(aapg Memoirs 88) P. M. Harris, L. J. Weber-giant Hydrocarbon Reservoirs Of The World (aapg Memoir No. 88)-american Association Of Petroleum Geologists (2006)

Copyrig ht © 2006 By the American Association of Perroleum Cc~>logists (AAPG) and SEPM (Society fo r Sedimentary Geology) All Rights Reserved

!SllN: 978-0-89181-704-8 0-89181-704-2

AAPC a nd SEPM grant permission for a single pho tocopy of an item fro m this publicatio n fo r personal use . Authorization for add itio nal copies o f items from this publication for personal or internal use is g ran ted by AAJ>G and SEPM provided that the base fee of $3.50 per copy and $.50 per page is paid d irectly to the Co pyrig ht Cleara nce Center, 222 Ros<>wood Drive, Danvers, Massachllsetts 01923 (phone: 978/750-8400). Fees are subject to change. Any form of electron.ic or d igital sca n.niJ>g or ot her d igital transformatio n of portions ot' t·his publication i.nto computer-readable M>d/or t·ra nsm ittable form for persoMI or corporate use re<.1 uires specia l permission f rom, and is subject to fee charges by, the 1\APG ru\d SEPM.

AAJ'>G Editor: Ernest A. Mancini Geoscience Director: )ames B. Blankenship

SEJ>M Special P ublicatio n Ed itors: Laura Crossey and Donald Me~'leill SEPM Publications Coordinator: Robert Clarke

Tf\is publicat io t\ is available fro m:

The AAPG Bookstore

SEPM

P.O. Box 9·79 T,~so. OK U.S.A. 74101-0979 Phone: l -918-584-2555 or '1-1:!00-364-AAPG (U.S.A. only) Fax: 1-918-260-2652 o r 1-800-898-2274 (U.S.A. only)

6128 E.1St 381h St«->et, Suite 308

E-mail: bookt>tore®aapg.org

Tulsa, OK U.S.A. 74135-5814

Phone: l-&J0-865-9765 (North America) or ·t-9'1 8-610-3361 Fax: 1-918-62H685 E ~ 1na iJ: eeUisW'sepm.org www.sepm.org

w-..\1"\v,aapg.org

Ceolosic.ll Society Publi•hiog 1-toU>e Unit 7, Brnssmill Enterpl'ise Centre BrassmiU Lane, Bath BA13JN U.K. Phone: +44-1225-445046 Fax: +44-'1 225-442836 E~roai J : [email protected] W\\11\'.geolsoc.org. tr.k

The American Associatio n of Petroleum Geologists (AAPG) does no t endorse or recommend products or services that may be cited_, used, o r discussed in AAPG publications o r in presentations a t e vents associated with AAJ'>C.

Table of Con tents Acknowledgments ······································ ~~-~~--..!<. About the Edjtors

vj

Introductj on

Giant Hydrocarbon Reservoirs of the World: From Rocks to Reservoir Characterization and Modeling P. M. Harris and L. [. Weber

Chapterl .................................

··----------------------------L----

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup, Precaspian Basin, Kazakhstan: Depositional Evolution and Reservoir Development [ . A . M. Kenter, P. M. Harris, [ . F. Collins, L. [. Webe1; G. Kuanyslwva, and D . [. Fischer

Chapter 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

Facies and Reservoir-quality Variations in the Late Visean to Bashkirian Outer Platform, Rim, and Flank of the Tengiz Buildup, Precaspian Basin, Kazakhstan [.F. Collins,[. A.M. Kenter, P.M. Harris, G. Kuanysl1eva, D . J. Fischer, and K. L. Stef(en

Chapter 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

97

Ghawar Arab-D Reservoir: Widespread Porosity In Shoaling-upward Carbonate Cycles, Saudi Arabia Robert P. Lindsay, Dave L. Cantrell, Geraint 111. Hugl1es, Thomas H. Keith, Harry W. Mueller Ill, and S. Duffy f<usse/1

Chapter 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

139

High-resolution Sequence Stratigraphy and Reservoir Characterization of Upper Tha mama (Lower Cretaceous) Reservoirs of a Giant Abu Dhabi Oil Field, Un ited Arab Emi rates Christian[. Stmhmeuge1; L.jim Weber, Ahmed Ghani, Klwlil AI-Mehsin, Omar Al-jeelani, Abdulla AI-Mansoori, l'aha AI-Dayyani, Lee Vaughan, Sameer A . Khan, and john C. Mitchell

Chapter 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Lower Cretaceous, Abu Dhabi (Un ited Arab Em irates) Lyndon A . Yose, Amy S. Ru(, Christian f. Strohmenger, jim S. Schuelke, Andy Combos, Ismail AI-Hosani, Slwmsa AI-Maskary, Gerald Bloch, Yousuf 111-Mehairi, and Imelda G. jolmson

173

Ch apter 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

213

Sequence Stratigraphy and Reservoir Architecture of the Burgan and Mauddud Formations (lower Cretaceous). Kuwait Christian f. Strolmrenger; Pemrv E. Patterson, Glraida AI-Salrlan, John C. Mitchell, Howard R. Feldman, Timothy M. Demko, E
H(

Broom/rail, and Neama ;1/.J\jmi

Ch apter 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

247

The Sequence Stratigraphy of the Maastrichtian (Upper Cretaceous) Reservoir at Wafra Field, Partitioned Neutral Zone, Saudi Arabia and Kuwait: Key to Reservoir Modeling and Assessment Dennis 111. Dull, /laynumd A. Garber, and 111. Scott Meddaugh

Ch apter 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

281

Stratigraphic Organization and Predictability of Mixed Coarse- and Fine-grained Lithofacies Successions in a Lower Miocene Deep-water Slope-channel System, Angola Block IS M. L. Porter, II. II. G. Sprague, M. D . Sul/ivmr, D. C. Jennette, II. T. Beaubouet T. R. Gar{ield, C. Rossen, D. K. Sicktt(oose, G. N./ensen, S. /. Friedmann, and D. C. Mohrig

Ch apter 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

307

Multiscale Geologic an d Petrophysical Modeling of the Giant Hugoton Gas field (Permian), Kansas and Oklahoma. U.S.A. Martin K. Dubois, 111/ttn P. BJ•rrres. Geoffrey C. Bo/1/irrg, and john H. Dovet011

Ch apter 10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

355

Key Role of Outcrops and Core.~ In Carbonate Reservoir Characterization and Modeling, Lower Permian Ful.lerton Field. Perm ian Basi.n. United States Step/ten C. Ruppel cmd Rebecca H. Jones

Chapter 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

395

Cart>onate Sequence Stratigrap hy and Petroleum Geology of the j urassic Deep Panuke Field, Offshore Nova Scotia. Canada John A. W Weissettberger, Richard 11. Wierzbicki, and Nancy f . Harland

Chapter 12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sedi mento logy, Sequence Stratigraphy, and Reservoir Ao:chltecture o f the Eocene 1-Hrador Formation, Cupiagua Field, Llanos Foothills, Colombia Juan Carlos Ramon cmd Anclres Fajardo

tv

433

Acknowledgments AAPG and SEPM are grateful and appreciative for the support of the AAPG Founda t io n .

The follow ing are gratefully acknowledged fo r their financial support:

Chevron EJf{onMobil

Contributions are applied against the production costs o f publicatio n, thus directly reducing the hoo k's

purchase price and ma king the volu me available to a greater a udience.

v

About the Editors Paul M. (Mitch) _f:l.a rris is n. Carbonate Reservoir Consultan t with Chevron Energy T('chnolog:y Compa ny in San Ramon, Califo rnia. He performs <·a rbona te research, 1echnical support projects, consulting, and train ing for the various operating u nils of Chevron. His \\10tk during t he last~ years has centered o n facies-re lated, stratigraphic. and diagenetic probl(;'ms th at p€'rtain lO carbom:ne reservoirs and expiO· ra tion play,o; in most carbo nate basins ,vorldwide. Mitch rt-ceived h is B.S. and M.S. degre,o.s from West Virginia University a nd Ph.D. from the University o f l\ttiami, Florida. He has published nu merous pa pers, edited several books, a nd is active In AAI'G a nd SEI'I\1. lie has been a Dlstlng11ished Lecturer and

Inte rnational Distinguished Lecturer fo r AAPG. was awarded 11onorary lvlembershlp from SEPM, and has received the Wallace E. Pratt Memorial Award and Robert H. Dott Sr. Memorial Award fro m AAPG. Mitch is also ad junct faculty at Rlce University, the University o f Miami, and the University o f Southern California .

L. J. Qim) W eber works for Ex.• onMobil ExploratJon Company as a Catbonate Stratigraphy Expert specializing in seque nce/seismic stratigraphy and reservoir chmac· tedzalion. liis curfent wo rk assignment invoJves techn ica l leadership an d pro ject toordination as a stratigraphic advisor. J im h
t he Cambro.Ordovicia n to th e Holocene.

vi

Introduction

Harris, C _,. M., and 1.. J. Wehc:r, 2006, Giant hydrocarbon re..'icrvoirs of the world: from rods to reservoir e.hara<.'1CrizatiOH and modeling,. iu P. M. Harris and 1.. J, Weber. eds., Giant hydrocarbon reservoirs of the world: From rocl-:s to reservoir <:haracterlz.atlon and mo
Giant Hydrocarbon Reservoirs of the World: From Rocks to Reservoir Characterization and Modeling P.M. Harris Chevron E11ergy Tectmology Company, Sm1 Ramon, California, U.S.A.

L.

J. Weber

fuonMo/Jil /Jxploration Company, Houston, Texas, U.S.A. INTRODUCTION

It is our goal that the technical content o f the chapters presented here result in discussion d irected toward

The SEPM/AAPG core workshop "Giant Hydroca r· bon Reservo irs o f the World: From Rocks to Reservoir Characterization and Modeling" and th is co mpan ion publication are an attempt to assemble information on giant ( rel="nofollow">500 MOEB recoverable reserves) reservoirs that is of value to a wide audience. Various examples and methods of reservoir characterization, develop· ment, and modeling practices are documented in this volume. Although far from exhaustive, this compila· lion includes a wide range of reservoirs when exam· ined from any perspective, i.e., location, geology, pro· duction history, and characterization. Figure 1 shows the geographic distribution of the reservoir examples that are Included In this volume. Agood understanding of geologic variability in time and space is a prereq uisite for any successful reservoir description. Geologic and engineering data obtlli ned from core are fundamental buildi.ng blocks in reservoir characterization. A com mo n goal is to describe the reservoir in sufficient detail to identify remain ing hy· drocarbons and then to produce these reserve_~ effi. ciently. To this end, a focus in this volume centers on aspects of geologic model ing as they relate to het· erogeneity In facies, which typic.a lly controls vari· ability In porosiry, permeability, and fluid saturations.

•

•

•

•

fundamental concepts and methods of reservo ir characterization, which include accurate represen· tation of internal and external reservoir geology, precise and quantitative description, and level of detail that satisfies simulation capabilities an understanding of the scales of rock hetero· geneity and its effect o n petrophysical and engi· neering properties and their relationships to fluid flow and hydrocarbon recovery improved methods o f reservoir description an.d development through application of high-resolution sequence and seismic stratigraphy and seismic visualization techniques identifying a lternative approache.~ to mo re e ffective reservoi r management practices

The reservoir examples described in this volume (!) explore historical and alternative approaches to reservoir description, characterization, and manage· ment and (2) examine appropriate levels and timing of data gathering, technology applications, evalua· tion techniques_, risk assessment, and nt.a nagement practices In various stages In the life of Individua l development projects.

COp)'rigllt 0 2006 by 1'hc American Associat:iou of PNrolcum CicoJogim.

001:10. 1306/1 21S872M8824?

1

2 I liarris and Weber FIGURE 1 . Reservoirs included in Giant Hydrocarbon Reser-

voirs of the World. Caspian Basin: (1) Tengiz platform and (2) Tenglz rim; Middle East: (3) Ghawar Arab-D, (4) Abu Dhabi Thamama, (5) Abu Dhabi Shualba, (6) Kuwait Burgan Mauddud, and (7} Wafra

Maastrichtian; West Africa: (8) Angola Block 15; North America: ( 9) Hugoton, (1 0) Fullerton, and (11) Deep Panuke; South America: (12) Cuplagua.

The giant reservoirs of th is volume account for approximately 0.5 trillion bbl recoverable hydrocarbons. Reservo ir examples a re both ca rbonate and siliciclastic, and collectively, they accoun t for a wide range of variability in reservoir parameters (e.g., gross rock volume, net-to-gross, porosity, and permeabi i· ity). Most of the reservoirs in this volu me occur o ut· side the United States and Canada, a nd until now, core from many of these reservoirs have not been widely observed . Enhanced recovery of hydrocarbons requires a criti· cal understanding of reservoir heterogeneity by both geoscientists and engineers. Spatial heterogeneity that affects fluid now occurs over a ll scales of investigation from the intr
SUMMARY OF RESERVOIR EXAMPLES Caspian Basin Kenter, Harris, Collins, Weber, Kuanysheva, and fischer discuss the t:entral pla tfo rm part oftheTengiz. field in "L
the B
Introduct ion I 3 facies are poorly bedded to massive boundstone breccia with subtypes based on clast co mposition, size, and packing; and upper-slope facies con sist of in-situ microbial bo undstone. Facies of the outer platform are shallow-platform skeletal, coated-grain, and ooid packsto ne to grai n.stone. Periodic large-sca le failure of the rim during both the Serpukhovian and Bash· kiria n resulted in a high degree of lateral facies discontinuity. Solution-en larged fractures, large vugs, and lost circlllation zones produced mainly during late diagenesis form a high-permeability, well-connected reservo ir in the rim and flank. Middle East

Lindsay, Cantrell, Hughes, Keith, Mueller, and Russell describe Ghawar field, which is the world's largest and most prolific o il field, in "Ghawar Arab-0 Reservoir: Widespread Porosity in Shoaling-upwa rd Carbonate Cycles, Saudi Arabia." Production occurs fro m the j urassic Arab-D carbonates. The upper half of the reservoir is dominated by exceptiona lly high reservoir quality, and t he lower half conta ins interbeds of less porosity. Th e reservoir is com posed o f two composite sequences. The upper composite sequence boundary is the top o f Arab-D carbonate, which is locally characterized by colla pse breccia. The lower composite sequence boundary between tbe Arab-D and the underlying j ubaila is m arked by deeper water cycles occu rring over gra in-dom inated cycles. Several h igh-frequency sequences are each composed o f cycle sets that each contai n approximately five individual carbonate cycles. These carbonates formed upon a broad, arid storm -domina ted ram p with a variety of rock types deposited . Diagenesis tha t is common within the Arab-D reservoir includes severa l dissolution even ts, recrystallization, and physical compaction. The resultant limesto ne porosity is a m ixture of interparticle (dom inant), moldic (c.ornmon), in traparticle (common), and microporosity (common) pore types. Less common dolostone porosity is a mixture o f moldic (less common), intercrystalline (less common). and intracrystalline (least common) pore types. The vertical seal for the reservoi r is the overlying Arab C-D an hydrite. In "High -resolution Seq uence Stratigraph y an d Reservoir Characterization o f Upper Thama ma (Lower C retaceous) Reservoirs o f a Giant Abu Dhabi Oil Field, United Arab Emirates" authors Strohmenger, Weber, Ghan i, AJ-Mehsln, Al-Jeelani, AJ-Mansoori, Al-Dan•ani, Vaughan, Khan, and Mitchell describe the Lower Cretaceous Kh araib (Barrem ian and early Aptian) and Shuaiba (Aptian) formations (upperTha-

mama Group) o f Abu Dhabi in wb icb important hydroca.rbon accumu lations occur in p latform carbonates. The Kharaib and Lower Shuaiba form ations contain th ree reservoir units separated by low-porosity and permeability dense zones. Core and well-Jog data fro m a gia nt oil field in Abu Dhabi and outcrop data fro m Wadi Rah abah in Ras Al-Khaimah were used to establish a seq uence-s tratigraphic framework and a litl1ofacies scheme. The Lower a nd Upper Kharaib Reservoir Units, as well as the upper dense zone, are part of a late transgressive sequence set of a scc.ondorder supersequence, built by two third-ord er composite sequences. The overlying Lower Shuaiba Reservoir Unit belongs to the la te transgressive sequence set and the early highstand sequence set o f this secondorder supersequence a nd comprises one third-order composite sequence. The th ree third-o rder composite sequences a re composed o f fourth-ord er parasequence sets that show p redo minantly aggradational and progradational stacking patterns, typical of greenhouse cycle.~. Reservoir lithofacies range from lowerramp to shoal crest to near backshoal open-platform deposits, whereas nonreservoir (dense) lithofacies represent an inner-ramp, restricted shallow lagoonal setting. Integration of subsurface and o utcrop data leads to more insightful and realistic geological models of the subsurface stratigraphy, and the geological model realizations based on core, outcrop, well-log, and seismic data co nstrain now-simulation models. Yose, Ruf, Strohmenger, Sch uelke, Gombos, AIHosani, AI Maskary, Bloch, Al-Mehairi, a nd Johnson integrate high-resolution three-dimensional (3-D) seismic with geologic and production data to describe the Lower Cretaceous (Aptian) reservoir in Abu Dhabi in "Volume-based Characterization of a Heterogeneous Carbonate Reservoir, Lower Cretaceous, Abu Dhabi (United Arab Emirates)." The. reservoir is positioned over a platform-to-basin transition and rec.ords a diverse range of depositional facies and stratal geometries. A second-order sequence set is divided into five depositiona l sequen ces. Sequences 1 and 2 are a transgressional phase showing the initial formation of buildup ma rgins and do minated by a lgaJ.prone f•cies. The subsequent h i.ghstand phase of Sequence 3 is mainly aggradational and records the proliferation of rudists across t he platfo rm !OJ>. A late highsta nd phase of sequences 4 and 5 is progradational showing th e progressive downstepping of the platform margin onto a low-angle slope. Three-dimen sional seismic data in the southern field area show a com plex mosaic of tidal channels, high-energy rudist shoals, and intershoal ponds. The geometry and reservoir-quality

4 1 Harris and \Veber

variations of these geologic features have a strong impact on reservo ir S\oveep and co nformance in the platform interior. ln the northern field area, seismic images of prograding slope clinoforms reveal system atic variations in architecture and reservoir qualitythat reflect multiple scales of stratigraphic cyclicity. A pat· tern gas llood has been Implemented in the clinoforms to add pressure support and improve recovery. Busi· ness applications of the reservoi r framework include (1) 3-0 seismic visua lization as a tool for optimizing well placement, identifying bypassed reservoirs and evaluating reservoir connectivity; (2) integra tion of quantitative, volume-based seismic informatio n into reservoir models; (3) m<Jximjzing recovery through full integration of a ll subsurface data; and (4) enh<~nced communicatio n a mong geoscientists and e ngmeers, leading to improved reservoir management practices. In "Sequence Strati.graphy and Reservo ir Arch itecture of the Burgan and Mauddud Formations (Lower Cretaceous), Kuwait" authors Strohmenger, Patter· son, AI-Sahlan, Mitchell, Feldman. Demko, Wellner, Lehmann, McCrimmon, Broomhall, and AI·Ajml pro· pose a new seq uence-stratlgraphic framework for the Burgan and Mauddud formations (Albian) of Kuwait, which resu lts in a predictable distribution of reset· voir and seal fades. The Burgan and Mauddud for· mations form two second-order composite sequences, the o ldest of which constitutes the lowstaod, trans· gressive, and h ighstand sequence sets of the Burgan Formation. This composite sequence is subdivided into high-frequen cy sequences that a re characterized by tidal-inlluenced, marginal-marine deposits in norUl · east Kuwait grading into more fluvial-dominated, con· tinental deposits to the southwest. The younger com· posite sequen ce consists of the lowstand sequence set of the uppermost Burgan formation and the trans· gressive. and highstand sequence sets of the overly· jng Maud dud l'ormation. This composite sequence is sand and mud prone in southern and southwestern Kuwait and is carbonate prone in northern and north· eastern Kuwait. The lowstand se<Juence set of the Bur· gan is subdivided into five high-frequency sequences. and the Mauddud transgress ive and highstand seq ue nce sets a re subdivided into eight h igh-frequency sequences. n 1e traditional lithostratigraphic Burgan Mauddud contact is time transgressive. The upper Mauddud h ighstand sequence set is carbonate prone and thins southward because of depositional thinning. AuU10rs Dull, Garber, and Meddaugh describe the Maastrichtian (Upper Cretaceous) reservoir in the giant Wafra o il field in "The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra field, Partitioned

Neutral Zone, Saudi Arabia and KuwaH: Key to Res· ervoir Modeli ng and Assessme nt." Oi l production is largely from subtidal dolomite formed on a very gently dipping, shallow, arid, and restricted ramp setting that transitioned between no rma l ~marine conditions to restricted lagoonal environments. The key to mod· ellng the reservoir was the construction of an appro· priately d etailed sequence -stratigraphic fram ework for use in building the geostatistical reservoir model. Within the sequence-stratigraphic framework, 10 ltigh· frequency sequences are correlated, a lbeit with some dllficulty, across the entue field. Ultimately, diagen· esis is a major factor in the distribution of po rosity and permeability. Dolomitjzation is pervasive, but fa· cies exerts some control on the distribution of pO· rosity. Average porosity o f the reservoir interval is 15%, with values as much as 45%, a nd permeability averages30 md wiUl core-plug measurements as much as 1200 md. The geostatistical model of the Maastrichtian reservoir demonstrates the layered a nd com · partmentalized nature of the reservoir a nd clearly shows that the location of the reservoir facies is con· trolled by the o riginal depositional fabric and sub· sequent dolomitization, both of which have been influenced by the paleotopography. Th is study w<Js undertaken to determine reservoir volumetrics, un· derstand the distributio n of inte rvals likely to yield h igher volumes of better quality oil, and provide a reservo ir property model fo r use in fluid-flow simulation. Such an understanding is critica l to efficiently develop the 1.5 billion bbl oil Maastricthian resource at Wafra field . West Africa

In "Stratigraphic Organization and Predictability of Mixed Coarse· and fine-grained Lithofacies Succes· sions io a Lower Miocene Deep-water Slope-channel System, Angola Block 15," Porter, Sprague, Sullivan, Jennette, Beaubouef. Garfield, Rossen, Sickafoose, )en· sen, Friedmann, and Mohrig describe the lower lvlio· cene slo pe-channel systems from Angola Block 15, which regional seism ic mapping, exploration drilling, and appraisal drilling have established as a worldclass development opportunity. One o f the major de· velopment targets in Block 15 is Burdigalian-aged slo pe-channel reservoirs, which are part of a system that traverses across the block in a n east - west direc. tion and can be continuously mapped on adjacent seism ic datasets for more than 30- 40 km (18- 24 ml). The channel system was a sediment fairway for the delivery of coarse-grained turdidites and mixed mud· dy and sandy debrites in to the Lower Congo basin.

Introduction I 5 Map patterns show distinctive changes in sinuosity, channel confi nement, and degree of amalgamation

bro
Dubois, Byrnes. Bohllng, a nd Doveton docume nt the reservoir characterization and modeling from pore to field scale of the giant Hugoton field in "Mul· tlscale Geologic and Petrophysical Modeling of the Giant Hugoton Gas field (Permia n), Kansas and Oklahoma." Their work on this mature Permian gas system is aiding in defining original gas in place, the nature and distribution of gas saturation, and reservoir properties. The Kansas- Oklahoma part of the field has yielded 963 billion m 3 (34 teO gas throughout a 70-year period from more than 12,000 wells. Most remaini ng gas is in lower permeabiUty pay zones o f the 170-m (557-ft)·thick, differentially depleted, layered reservoir system. The main pay zones have remarkable lateral continuity. They represent 13 shoaling-upward, fourth-order marine-continental cycles, comprising thin-bed ded (2-10-m; 6.6-33-ft), marine cMbonatc mudstone to grainstone and siltstones to very fine sandstones. The pay zones are separated by low-reservoirquality eolian and sabkha redbeds. Petrophysical properties vaty among major lithofacies classes. Neu ral n etwork p rocedures, stochastic modeling, and automation fadlitated building a detailed full-field 3-D cellular reservo ir model using a fou r-step workflow: (1) define lithofacies in co re and correlate to electric log cu rves (training set); (2) train a neu ral network and predict li thofacies at noncored wells; (3) po pu late a 3-D cellu lar mode l wi th lithofacies using stochastic methods; and (4) populate m odel with lithofacies-specific petrophysical properties and fluid saturations. Both the knowledge gained and the techniques and workflow employed have implications

for understanding and modeling reservoir systems worldwide that have similar geologic age and reservoir architectu re.

In "Key Role o f Outcrops and Cores in Carbonate Reservoir Characterization and Modeling, Lower Perm ian Fullerton field, Permian Basin, United States" authors Ruppel and jones discuss the rock-based model con.struction for fulle rton Clear fork field, which is a shallow-water platform carbonate reservoir of middle Permian age in the Permian Basin of west Texas. Fundamenta l steps in their study included (1) creating and applying an analogous outcrop depositional model; (2) describing and interpreting subsurface core and log data in terms of this initial model; (3) defining the sequence-stratigraphic arch ite~ture of the reservoir section; (4) developing a ~)'de-based reservoir framework; and (5) defin ing controls, interrelationships, and distribution o f porosity and permeability. Data used in th is analysis included co res. th in sections, 3-D and two-dimensional (2-D) seism ic data, borehole image Jogs, and outcrop models. Stratal architecture, differential dolomitization, karst fill, mineralogical variations, and rock-fabric distribution were incorporated into the model. 'lh ese components were used to con strain interpretation and d efinition of flow units, permeability distribution, and saturation. The rock-based methods demonstrated in this study provide key insights with broader application into the formatio n, characterization, and interpre tation of cMbonate platform reservoirs. Authors Weissenlle rger, Wierzbicki, and Harland describe the dee p Pan uke field in "Carbonate Sequence Stratigraphy and Pe troleum Geology of the jurassic Deep Pa nuke f ield, Off~hore Nova Scotia, Canada ." Deep Panuke, which is located 250 km (!55 mi) offshore of Halifax, Nova Scotia, Canada, contains gas-fiUed cavernous porosiry in jurassic carbonates of the Abenaki !'ormation. The Abena ki carbonates range from Bathonian to Neocomian in age and were deposited on broad carbonate platform attached to a siliciclastic hinterland. Seven third-order depositio nal sequences recogn ized in the Abenaki are correlated with geology and a 2-D seism iC grid; 3-D seismic data are used (Ot deli neation drilling and reservoir characterization. Lithologies range fro m fo reslope to reefal and shoal deposits near the platform margin. Open and restricted lagoon or tidal-flat deposits occur in th e platform interior. Siliciclastics are concentrated near sequence boundaries or are distributed along stri ke, close to point sources such as rive rs cutting through the platform . Th e reservoir occurs n car the platform margin in coral and

6 1 Harris and \Veber

stromatoporoid reef and associated skeleta l-peloidal and occasionally oolitic shoal deposits. It comprises a range of porosity types, from vuggy limestone and dolo mite to microporosity in limestone. Geochemical, isotOJ>e, and petrographic data suggest that the dolomitization and dissolution can be attributed to deep burial and hydrothermal fluids. Sout h America

In "Sedimentology, Sequence Stratigraphy, and Reservoir Architecture of the Eocene Mirador !'ormation, Cupiagua Field, Llanos Footh.ills, Colombia," Ramon and Fajardo document the stratigraph.ic architecture and facies distribution in a h igh-resolution ti me-space framework to define the 3-D reservoir zonati on of the Mirador Formation (Eocene) in the Cupiagua field. The Cupiagua structure is a large, eastverging. asymmetric anticlinal fold that trends no rthnortheast in the hanging wall o f the frontal fault. The Mirador Formation accounts for approximately 55% of the recoverable oil in Ute field. Three scale.~ of strati· graphic cycles are recognized based on stacking pat· tern and general lrend of facies successions: shortterm cycles or progradational and aggradational units stack systematically into intermediate-term cycles, which, in turn, are grouped into long-term cycles. The lower half of the Mirador consists of flood-plain facies with channel, crevasse splay/ and swamp and flood-plain facies successions. Bay-head delta and bayfill facies occur in the upper half of the Mirador Formation. The Lower 1\·firador consists or two intermediatesca le cycles s howing a seaward-stepping stacking pattern overla in by a third cycle with a landwardstepping pattern, and the upper Mirador conti nues the landward-stepping pattern . This upper unit con-

sists of tluee onlapping cycles composed of a succession of aggradatio nal channel deposits, p rogradational bay-head delta and bay-fill deposits with a landward -stepping stacking pattern. The Mirador is capped by restricted marine shales of the Carbonera Formation.

ACKNOWLEDGMENTS As organizers of the core workshop and editors of the companion voltune, we acknowledge several people and our respec.tive companies for their assistance, without which the workshop and volume would not have happened. Chevron and ExxonMobil provided generous financial assistance that enabled numerous students to attend the core workshop and subsidized printing costs o f the worksl•op volume. SEPM staff helped us o ve rco me n umerous o rganizational and logistica l issues that confronted us as we organized the workshop. AAPG staff and, in particular, Beverly Molyneux provided us invaluable help in the editing and publication of the workshop volume. We thank all of the authors wh o worked bard to prepare poster and core displays, oral presentations, and the detailed manuscripts t11at describe their respective fields. Finally, we thank the many who helped us in the technical review of the manuscripts for the workshop publication: Bob Alway, Steve Bachtel, Sherry Becker, Kelley Bergman, j ohn Bova,joel Colli ns. Bob Dalrymple, Laurence Droz. Ch.ip Feazel, Sean Guidry, Jurgen Grotseh, jean Hsieh, jon Kaufman, Mike Kozar, Dale leckie, j ose Matos, jim 1\kGovney, Gary Parker, Carlos Plrmez, George Pemberton, Linda l>rJce, Gene Rankey, Rick Sarg, Toni Simo, Krishnan Srinivasan, and Niall Toomey.

1

Kenter, J. A. M., P. M. Harris, J. F. Collins, L. J. Weber, G. Kuanysheva, and D. J. Fischer, 2006, Late Visean to Bashkirian platform cyclicity in the central Tengiz buildup, Precaspian Basin, Kazakhstan: Depositional evolution and reservoir development, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 7 – 54.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup, Precaspian Basin, Kazakhstan: Depositional Evolution and Reservoir Development J. A. M. Kenter1

L. J. Weber

Vrije Universiteit, Amsterdam, Netherlands

ExxonMobil Development Company, Houston, Texas, U.S.A.

P. M. Harris Chevron Energy Technology Company, San Ramon, California, U.S.A.

G. Kuanysheva

J. F. Collins

D. J. Fischer

ExxonMobil Development Company, Houston, Texas, U.S.A.

TengizChevroil, Atyrau, Kazakhstan

TengizChevroil, Atyrau, Kazakhstan

ABSTRACT

T

he Tengiz buildup, an intensely cored and studied isolated carbonate platform in the Precaspian Basin, contains a succession of shallow-water deposits ranging from Famennian to Bashkirian in age. From a reservoir perspective, Tengiz can be subdivided into platform (central and outer) and rimslope (flank) regions. The upper Visean, Serpukhovian, and Bashkirian form the main hydrocarbon-bearing interval in the platform. Depositional cycles (highfrequency sequences) in this interval are several to tens of meters thick for the Visean and Serpukhovian, and decimeter to meter scale for the Bashkirian. Cycles are made up of a succession of lithofacies overlying a sharp base that locally shows erosion, calcretes, meteoric diagenesis, and other evidence for subaerial exposure. At the base of the succession, tight peloidal mudstone and ash beds are associated with sequence boundaries and are thought to reflect lowenergy conditions developed in deeper platform areas at lowstand and during 1

Present address: Chevron Energy Technology Company, San Ramon, California, U.S.A.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215873M88374

7

8 / Kenter et al.

initial flooding. Above this, beds with in-situ articulated brachiopods signal initial open-marine but still low-energy conditions. Succeeding crinoid-dominated intervals represent maximum marine flooding and overlying skeletal-peloidal grainstones highstand shoaling phases. Visean and Serpukhovian cycles are generally easy to correlate from well to well over several kilometers distance. Volcanic ash beds are identified by gammaray spikes, and flooding intervals show as low-porosity zones. In contrast, Bashkirian cycles are thinner and incomplete, dominated by thin, peloidal mudstone intervals alternating with high-energy coated-grain and ooid grainstone, and are more difficult to correlate. High-frequency icehouse sea level fluctuations exposed the platform during each fall of sea level, and rapid flooding resulted in incomplete cycles and complex lateral facies changes that may explain relatively poor lateral continuity of log character. The distribution of reservoir rock types in the central platform is determined by burial diagenetic modification of an earlier reservoir system that includes meteoric alteration and porosity enhancement below major sequence boundaries and reduced dissolution along higher order sequence boundaries associated with the presence of volcanic ash. The lateral continuity of tight layers at sequence boundaries probably greatly affected later fluid flow as well as the ultimate distribution of cements, dissolution, and bitumen in the central platform reservoir. The burial diagenetic overprint included two major phases of reservoir modification. First, a corrosion and cementation phase significantly enhanced existing matrix porosity in the interior central platform while reducing porosity in the exterior central and outer platform by pore-filling equant calcite cement. This was followed by bitumen emplacement and associated corrosion. These processes not only exerted an overall porosity-reducing effect prior to and associated with bitumen invasion toward the exterior central platform, but also dampened or flattened the initial cyclic porosity variations and obscured relationships between pore types and permeability. The bitumen overprint is nearly absent in the innermost platform wells; bitumen concentrations are highest near the bases of the cycles, which may imply that the first fill of hydrocarbons migrated through the flanks laterally into the platform cycles.

INTRODUCTION Tengiz Field History The Tengiz field, located in western Kazakhstan, near the northeastern shore of the Caspian Sea (Figure 1), produces oil from an isolated carbonate platform (aerial extent of >110 km2 [42 mi2]) of Devonian and Carboniferous age. Tengiz was discovered in 1979 by the Ministry of Oil Industry of the Soviet Union. The discovery well Tengiz 1 (T-1) reached a total depth of 4095 m (13,435 ft). Development drilling of the Tengiz field commenced in 1983, and onsite construction of plant-processing facilities began in 1987. Field production officially began in April 1991. Since 1993, TengizChevroil, an in-country joint venture company run by Chevron, has operated Tengiz and the adjacent Korolev field.

Tengiz field produces a light, intermediate-sulfur, stabilized tank oil of approximately 478 API. As of midyear 2005, more than 115 wells have been drilled on Tengiz. The highest rate wells are located in the platform margin and slope in fractured carbonates with low (<6%) matrix porosity. Platform wells exhibit higher porosity (as much as 18%), but matrix permeability is typically low (<10 md).

Regional Paleogeography The Precaspian Basin is a large Paleozoic basin that occupies much of the area known today as the northern part of the Caspian Sea and adjacent landmass. During the Early Carboniferous, the Precaspian Basin was a major equatorial sag basin in the western part of Kazakhstan (Ross and Ross, 1985). Carbonate shelves formed along the basin margins, and isolated, broad

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 9

FIGURE 1. Index map locating North Caspian region and outlining area of the Tengiz field in blue, which is shown in more detail in Figure 2. carbonate platforms grew on preexisting basement highs (Cook et al., 1997). Together with other nearby fields, Tengiz forms an archipelago of isolated carbonate platforms that grew on a regional pre-middle Devonian structural high near the southeastern margin of the basin (Figure 2). The initial carbonate platforms in the archipelago formed sometime after the Middle Devonian and continued more or less uninterrupted until the early Bashkirian. The total thickness of carbonate that accumulated at Tengiz during that time has not been drilled but is represented by approximately 1.2 s two-way traveltime on the most recent three-dimensional seismic survey (Figure 3). Following platform termination in the early Bashkirian, the platforms were encased in a series of lithologies

associated with the closure of the Precaspian Basin. Initially, a veneer of Moscovian through Artinskian deep-water carbonates and volcaniclastics accumulated around and on top of the buildups. These deposits were followed by thick Kungurian evaporites, including halite that completely encased the platforms.

SUPERSEQUENCE FRAMEWORK In 2001, a joint study by ExxonMobil, Chevron, and TengizChevroil established a stratigraphic framework for the Tengiz platform based on seismic, biostratigraphic, core, and well-log data. This study was published in late 2003 (Weber et al., 2003) and stands

10 / Kenter et al.

FIGURE 2. Paleogeographic map showing the archipelago of isolated carbonate platforms, including Tengiz, that grew on a regional premiddle Devonian structural high near the southeastern margin of the Precaspian Basin.

as the definitive stratigraphic reference for Tengiz. The framework consists of a hierarchy of sequences that have continued to be maintained as Tengiz drilling proceeds. Through the main hydrocarbon-bearing interval at Tengiz (Famennian – Bashkirian), seven bounding discontinuity surfaces (sequence boundaries and maximum flooding surfaces [MFSs]) are recognized on seismic data (Weber et al., 2003) (Figure 4; Table 1). Two additional MFSs, Lvis1_MFS and Lvis2_MFS, are identified primarily from well control (cores and well logs). Four supersequences (Tournaisian – lower Visean, lower Visean – upper Visean, upper Visean – Serpukhovian, and Bashkirian) extend from the Famennian supersequence boundary (Fame_SSB) to the Bash_SSB (Weber et al., 2003). Each supersequence contains a transgressive sequence set (TSS) and a highstand sequence set (HSS) separated by an MFS. The upper two supersequences are the focus of this chapter. Three reservoir zones (Visean A, Serpukhovian, and Bashkirian) are contained in these supersequences and

constitute the platform reservoir part of the Tengiz buildup. Thin subaerial exposure caps on subtidal lithofacies form the most common type of cycle top in the Lvis_SSB to Bash_SSB succession on the Tengiz platform. The implication of this type of cycle boundary is that rapid base-level fall exposed subtidal lithofacies without development of an intervening tidal flat or peritidal succession. Alternatively, anchor points for tidal flats may not have been available because the depositional profile near the margin gradually deepened toward the platform break without the presence of a shallow barrier. In other words, the Tengiz platform was not conducive to producing extensive tidal-flat facies. Each cycle is more appropriately termed a ‘‘high-frequency sequence,’’ capped by a highfrequency sequence boundary. Cycle boundaries also occur where subaerial exposure is not evident from core and thin-section description. Without evidence of subaerial exposure, upper bounding surfaces of these cycles define parasequences.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 11

FIGURE 3. (A) Map of Tengiz showing the location of seismic line shown in (B). The map shows top reservoir (=top Bashkirian) structure with a contour interval of 100 m (330 ft). (B) Seismic profile showing general depositional areas of Tengiz buildup (central platform, outer platform, rim slope, or flank) and key reservoir zones (RZ). Vertical scale is two-way traveltime in milliseconds. See text for discussion.

Upper Visean–Serpukhovian Supersequence (Visean A and Serpukhovian Reservoir Zones) The upper Visean – Serpukhovian supersequence is defined as the interval between the Lvis_SSB and the Serp_SSB (Figure 4). It includes the Mikhailovsky, Venevsky, Tarussky, Steshevsky, Protvinsky, and Zapaltyubinsky regional horizons (Weber et al., 2003). The Zapaltyubinsky zone is not generally recognized on the Tengiz platform but is observed in wells that penetrate the slope and basin. The Zapaltyubinsky may therefore represent a basinward shift in sedimentation and form a lowstand sequence set (Weber et al., 2003). However, rim failure processes (see Collins et al., 2006) have obscured stratal patterns that clearly define the formation of such a depositional body. On the platform, the upper Visean – Serpukhovian supersequence is approximately 250 m (820 ft) thick; however, equivalent slope deposits (see Collins et al.,

2006) may exceed 600 m (1968 ft), where in-situ microbial boundstone and allochthonous debris aprons of platform- and slope-derived grainy carbonate and breccia accumulated (Figure 3). The upper Visean – Serpukhovian supersequence spans 14 m.y. (Gradstein et al., 2004). The Lvis2_MFS separates the TSS of the supersequence from the overlying HSS (Figure 5). The TSS consists of five composite sequences that are dominantly aggradational and do not extend beyond the platform break of the underlying lower Visean–upper Visean supersequence. Small-scale backsteps are possible, but not proven, at or near the platform break through the TSS. Deeper water, low-energy open-marine, platform lithofacies in these sequences include crinoid grainstone-packstone and poorly sorted, thick-walled brachiopod grainstone-packstone. Shallower water, high-energy, open-marine to back shoal lithofacies

of 100 m (330 ft). (B) Legend for log curves shown in cross section of (C). Gamma-ray (SGR and CGR) is shown on the left and porosity (PHIE) on the right. (C) Cross section through Tengiz showing major stratigraphic surfaces and reservoir zonation scheme. Stratigraphic surfaces, in ascending order, are Tournaisian (Tour_MFS); early Visean (Evis_SSB); late Visean (Lvis1_MFS, Lvis_SSB, and Lvis13_csb); Serpukhovian (Serp_SSB); and Bashkirian (Bash_SSB). Reservoir zones bracketed by these surfaces are Visean D (VisD), Visean C (VisC), Visean B (VisB), Visean A (VisA), Serpukhovian (Serp), and Bashkirian (Bash). The Visean A, Serpukhovian, and Bashkirian are the subject of this study.

FIGURE 4. (A) Map of Tengiz showing the location of cross section shown in (C). The map shows top reservoir (=top Bashkirian) structure with a contour interval

12 / Kenter et al.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 13

Table 1. Summary of sequence-stratigraphic nomenclature for Tengiz.* Stage Bashkirian

Reservoir Zones Unit 1

Bashkirian

Serpukhovian

Serpukhovian

Late Visean

Visean A

Visean B

Unit 2

Visean C

Early Visean

Visean D

Tournasian

Famennian

Tournasian

Unit 3

Famennian A Famennian B

Sequence

Top Surface

Sequence Number

Bash6 Bash4 Bash2.5 Bash1 Serp5 Serp4 Serp3 Serp2 Serp1 Lvis13 Lvis12 Lvis11 Lvis10.5 Lvis10 Lvis9.5 Lvis9 Lvis8 Lvis7 Lvis6 HRZ Lvis4 Lvis3 Lvis2 Lvis1 Evis9 Evis8 Evis7 Evis6 Evis5 Evis4 Evis3 Evis2 Evis1 Tour6 Tour5 Tour4 Tour3 Tour2 Tour1 Fame2 Fame1

Bash_SSB Bash4_csb Bash2.5_csb Bash1_csb Serp_SSB Serp4_csb Serp3_csb Serp2_csb Serp1_csb Lvis13_csb Lvis12_csb Lvis2_MFS Lvis10.5_sb Lvis10_csb Lvis9.5_sb Lvis9_csb Lvis_SSB Lvis7_csb Lvis6_csb Lvis5_csb Lvis1_MFS Lvis3_sb Lvis2_sb Lvis1_sb Evis9_sb Evis8_sb Evis7_sb Evis6_sb Evis_SSB Evis4_sb Evis3_sb Evis2_sb Evis1_sb Tour_MFS Tour5_sb Tour4_sb Tour3_sb Tour2_sb Tour1_sb Fame_SSB Fame_MFS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

*This framework was established by Weber et al. (2003), and picks in wells have, since then, been refined as a result of new wells and progressed knowledge of the field.

include well-sorted grainstone and grainstone-packstone consisting of algae, foraminifera, peloids, and coated grains. The upper Visean–Serpukhovian HSS is subdivided into an aggradational and progradational phase. The aggradational interval from Lvis2_MFS to Serp1_csb

(composite sequence boundary) is composed of three composite sequences (Figure 5). Significant progradation is observed in four subsequent composite sequences from the Serp1_csb to the Serp_SSB as shallow platform lithofacies fill available accommodation space and prograde to the present-day location of the

14 / Kenter et al.

FIGURE 5. North – south cross section through wells T-5044, T-5246, T-5447, T-220, T-6246, and T-6846; well locations are shown in Figure 4A. Cross section show detailed lithofacies and cycle correlations in the central platform region in the Visean A (VisA), Serpukhovian (Serp), and Bashkirian (Bash) supersequences. The lithofacies legend was simplified from that of Weber et al. (2003) by reducing the number of facies types and relating them to the level of energy during deposition. Sequence stratigraphy is adopted from Weber et al. (2003); triangles identify higher order sequences (fourth and higher order) correlated from well to well. See text for further explanation and discussion.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 15

16 / Kenter et al.

platform break. As much as 400–500 m (1600–1900 ft) of depositional topography was filled on the slope as shallow-water platform lithofacies prograded over predominantly slope boundstone of microbial origin (Figure 3). During the progradational part (Serpukhovian) of the upper Visean – Serpukhovian HSS, the Tengiz buildup was an overall flat-topped platform with a substantial slope apron collecting debris primarily sourced from autochthonous boundstone growing on the outermost platform and upper slope and, to a lesser extent, from the platform top. The platform interior is dominated by grainy textures (packstone and grainstone), with packstone dominating in the deeper subtidal areas and well-sorted grainstone in the higher energy shallow subtidal parts of the platform.

Bashkirian Supersequence (Bashkirian Reservoir Zone) The Bashkirian supersequence occurs between the Serp_SSB and the Bash_SSB (Figures 4, 5) and spans approximately 10 m.y. (Gradstein et al., 2004). The oldest Bashkirian horizon (i.e., Bogdanovsky) and most of the overlying Syuransky horizon do not appear to be present on the Tengiz platform (Brenckle and Milkina, 2001). The Akavassky and Askynbashsky horizons are generally present on the Tengiz platform above the Syuransky horizon. Although more work is necessary, results suggest that the early Bashkirian at Tengiz is transgressive, i.e., TSS of the Bashkirian supersequence (Figure 5). The HSS of the Bashkirian supersequence is very thin and condensed. Biostratigraphic analysis of core and cuttings typically dates the Bash_SSB as the top of the early Bashkirian. The late Bashkirian is condensed and represents starved sedimentation on an isolated platform, probably deposited in deep water (Weber et al., 2003). Throughout most of the central part of the Tengiz platform, the Bashkirian supersequence is uniform in thickness (80 – 100 m; 262 –330 ft). The Bashkirian platform break is generally aggradational through three composite sequences (Bash 1, 2.5, and 4) and coincident with the underlying platform break at the Serp_SSB. In isolated areas along the eastern margin of the platform, the Bashkirian section thickens to about 150 m (492 ft) (e.g., T-5056 and T-7252), with some of this thickening resulting from a facies change. Well-log correlations in these areas indicate that the uppermost composite sequence is locally 30–60 m (98–196 ft) thick (Weber et al., 2003), seismic data shows locally mounded seismic facies, and cores show skeletal packstone and local microbial boundstone on top of Serpukhovian microbial cement boundstone

several hundreds of meters thick. The remaining thickening could be explained by differential compaction between the massive and mechanically strong boundstone and the compacted central platform. The latter is supported by the petrographic observation that most grainstones are overcompacted.

VISEAN A, SERPUKHOVIAN, AND BASHKIRIAN PLATFORM ARCHITECTURE The Visean A, Serpukhovian, and Bashkirian reservoir zones are significant because they account for the bulk of the oil production from the platform part of the buildup. The platform is subdivided into different paleogeographic or depositional areas: central platform, outer platform, and rim slope (flank) (Figure 3). This chapter focuses on details of the central platform part of the Visean A, Serpukhovian, and Bashkirian intervals. To refine the central platform stratigraphy, a careful comparison of core and thin sections in a series of north –south wells (T-5044, T-5246, T-5447, T-220, T-6246, and T-6846) produced a detailed lithofacies correlation across the central platform region in the Visean A, Serpukhovian, and Bashkirian reservoir zones (Figure 5). Several general observations can be made on the cycle stacking pattern, overall depositional framework, and facies patterns visible in Figure 5; they are summarized here and discussed in more detail in subsequent sections. The succession of Visean A, Serpukhovian, and Bashkirian reservoir zones are, respectively, the TSS, HSS, and TSS of the two supersequences previously mentioned, and each consists of several composite and higher order sequences (cycles). Cycle stacking patterns (shoaling cycles as indicated by triangles in Figure 5) in the Visean A and Serpukhovian reservoir zones (and, respectively, TSS and HSS intervals) show a repetition of a thick cycle (15– 20 m; 49 – 66 ft) with commonly a basal crinoid- and brachiopod-rich sand followed by a series of thinner cycles (2 – 8 m; 6.6 – 26 ft). Those crinoid- and brachiopod-rich facies are interpreted as generally deeper water and relatively low energy and commonly extend all across the platform, which would suggest significant flooding following a lowstand of relative sea level. Ten to twelve composite sequences, each containing two to six higher order sequences (cycles), are shown in Figure 5. The boundaries of these composite sequences do not always exactly match those by Weber et al. (2003), although the number of composite sequences is essentially the same.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 17

The composite sequences described herein bundle into four larger sequences that each have a thick and laterally continuous brachiopod or crinoid interval near the base. Minor differences are observed between wells, but in general, the stacking pattern described above is very consistent. In the Bashkirian interval (mostly interpreted as TSS), the stacking pattern is more complex and shows a lower interval with thick cycles (10 m; 33 ft), followed by an interval dominated by thin (2–4-m; 6.6–13-ft) cycles, and an interval near the top of intermediate cycles (4 – 8 m; 13 – 26 ft). No clear stacking pattern is observed, and although lateral thickness variations are significant, the succession of thick, thin, and intermediate cycle intervals seems to be laterally continuous. The moderate differences between the sequences established by Weber et al. (2003) and those described herein may lead to slight modifications of the platform sequence-stratigraphy framework. One potential modification is changing the MFS in the Visean A – Serpukhovian supersequence from its position at Lvis2_MFS as previously discussed to the base of the thick brachiopod interval observed between Lvis12_csb and Lvis13_csb, which marks a shift from a series of thinning cycles below to thick cycle above and an overall change of generally deeper facies below to higher energy shallow-water facies above (Figure 5). These sequences of the Visean A, Serpukhovian, and Bashkirian intervals span a transition from greenhouse to icehouse cyclicity as suggested by observed trends in cycle thickness and general depositional environment. Another important observation is that sequences and cycles (third order and higher order) display a high degree of lateral continuity and constant thickness across the central platform area into the outer platform zones over distances of more than 10 km (6 mi) north to south. Careful inspection of this railroad pattern suggests the presence of a very low-angle top lap and possible truncation of underlying cycles at the Serp_SSB surface (the cross section of Figure 5 is flattened on this level). The truncation is variable but generally increasing to the south, indicating northward paleodip. In general, it can be stated that the central platform area as documented in this chapter is several kilometers inboard from the physical break between platform and flank. The outer platform contains areas that remain poorly defined because of a lack of core penetrations and/or uncertainty of the physical and sedimentological character of the central platform-toflank transition. Core penetrations close to the western edge of the platform are uncommon, but those available, wells T-44, T-6846, and T-6743, suggest moder-

ate but substantial differences from the central platform area and, therefore, a potentially discontinuous transition from platform to flank. The situation is more complicated toward the eastern margin, where apparent top-lap geometry of the Serpukhovian and Bashkirian platform develops over a Serpukhovian boundstone slope. In the underlying Visean A platform, the transition may be comparable to that recognized from the western margin, wherein the Serpukhovian and Bashkirian central to outer platform transition seems to be gradual and changes from deeper outer platform with intercalated boundstone intervals into boundstone upper slope. As a consequence, the lateral facies associations below the pseudo-top-lap may be significantly different from those above. See Collins et al. (2006) for further discussion related to these transitions.

Lithofacies Types in the Central and Outer Platform Core penetrations available for the Tengiz sequencestratigraphic framework developed by Weber et al. (2003) were limited to TengizChevroil wells T-220, T-5246, T-6846, and incomplete cores from numerous older Russian wells (figure 3 of Weber et al., 2003). The present study builds on that of Weber et al. (2003) by adding the information from cores that nearly completely span the Visean A, Serpukhovian, and Bashkirian in several key wells in the central (T-5044, T-5447) and southwestern outer platform (T-6246) (Figure 4A). These complete core penetrations for the first time allowed a detailed comparison of each individual cycle (and facies types within) in the study window. Evaluation of core observations, petrography, and wire-line and core gamma logs (porosity was excluded during this phase) resulted in a coherent catalog of 10 key lithofacies types that are ranked primarily based on the relative hydrodynamic energy during deposition (Table 2). This information was extracted through environmental indicators like grain size, sorting, matrix type, sedimentary structures, and grain types (biota). The resulting 10 facies or rock types can be grouped into three or four composite rock types, reflecting major differences in energy conditions during deposition. The catalog was also designed to assist in the characterization of plugs and petrography of the key wells at 1-ft (0.3-m) intervals and serve as a direct link between geological parameters to reservoir properties.

Cyclic Lithofacies Variations In the central platform, lithofacies stack into shallowing-upward cycles. The two supersequences and three reservoir zones making up the Lvis_SSB to

18 / Kenter et al.

Table 2. Summary of lithofacies, pore types, porosity, and bitumen content for central platform wells T-220 and

T-6246.*

Facies

Fabric

EoD

Abbreviation

General Description

New Facies Code Number

Algal grainstone (AG)

Algal dominated silt to fine sand G, very well sorted, with minor coated grains (smaller than 500 mm) Mixed skeletal-(peloidal) G, moderately well sorted, fine to medium sand, varying shell hash, minor crinoids, abundant intraclasts, oncoids, coral (massive) and Chaetetes fragments, forams, algae Ooid-dominated G, well to very well sorted, minor crinoids or shell hash, intraclasts, forams; locally rundstone with intraclasts; cross-lamination Skeletal G, fine to medium sand, well to very well sorted, coated grains, minor crinoids or shell hash, forams, intraclasts, algal fragments, locally overpacked; locally rudstone with intraclasts; cross-lamination Mixed skeletal-(peloidal) G, moderately well sorted, fine to medium sand, minor to abundant shell hash, moderate crinoids, intraclasts, forams, algae Skeletal-peloidal GP, coarse sand, moderately to poorly sorted, mixed crinoids and brachiopod shell hash, minor in-situ (thick-walled) brachiopods, peloids, algae, forams, whole and fragments of stick corals Skeletal-peloidal GP, medium to coarse sand, moderately to poorly sorted, dominant crinoids (1 – 2 mm or larger), minor whole brachiopods and hash, peloids, forams, (green and tubular) algal fragments Skeletal-peloidal GP, coarse sand, poorly sorted, brachiopods in-situ (thin vs. thick-walled; 1-5/20-50), bedded, variable crinoids, peloids, divers algal fragment, abundant and divers forams, whole and fragments of stick corals

46

Grainstone

High-energy reef flat, backshoal-inner platform

44

Grainstone

High-energy reef flat, backshoal-inner platform

43

Grainstone

High energy above wave base-intertidal platform

42

Grainstone

High energy above wave base-intertidal platform

41

Grainstone

Moderate to High energy above wave base-open platform

32

Grainstonepackstone

Moderate to high energy below or around wave base-open platform

31

Grainstonepackstone

Moderate to low energy below or around wave base-open platform

2

Grainstonepackstone

Ranging from mud to peloidal PW, silt-fine sand size with small oncoids, calcispheres, thin-walled bivalve fragments and ostracods Volcanic ash

13

Packstonegrainstone

Moderate to low energy below or around wave base open platform (abundant crinoids an thick-walled brachiopods) versus shallow protected platform interior (minor crinoids and thin-walled brachiopods) Low-energy restricted; restricted lagoon to supratidal

12

Clay mudstone

Skeletal intraclast grainstone (SIG)

Coated grain grainstone-rudstone (CgGR) Ooid grainstone (OG)

Well-sorted skeletal grainstone (SPG) Poorly sorted skeletal-peloid grainstone-packstone (SPP) Crinoid skeletal-peloid grainstone-packstone (CSP) Brachiopod skeletal-peloid grainstonepackstone (BSP)

Peloid packstonewackestone (PPW) Volcanic ash (VA)

Simplified Fabric Code

Low-energy restricted inner platform to postdrowning

All Grain-supported mud-lean; high energy Grain-supported muddy; intermediate energy muddy; low energy

41 – 46 2, 31, 32 13

*Lithofacies, also shown in Figure 5, are based on Weber et al. (2003), but simplified by reducing the number of facies and relating them to the level of energy during deposition. The vertical arrangement of lithofacies in this table resembles the ideal and most complete succession of lithofacies in a shoaling cycle or sequence. Pore types are 1 = interparticle; 2 = intraparticle; 3 = fenestral; 4 = shelter; 5 = growth framework; 6 = intercrystalline; 7 = moldic; 8 = microporosity; 9 = fracture; 10 = channel; 11 = vug; and 12 = enhanced dissolution. Note that bitumen analyses for well T-6246 are not yet complete. See text for discussion.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 19

T-220 N

24

Porosity

T-6246 Bitumen (% Rock Volume)

Dominant Pore Types

Range (%)

Mean (%)

Range

Mean

11, 1, 6, 2

2.25 – 17.59

13.08

0.16 – 0.31

0.23

0.18 – 16.35

4.46

0.08 – 0.41

18

N

Porosity Dominant Pore Types

Range (%)

Mean (%)

54/48

1, 7, 6, 11, 2, 9

0.84 – 14.80

4.90

0.28

22/22

1, 2, 7

0.97 – 12.07

4.58

78

11, 1, 7

0.42 – 23.66

10.09

0.13 – 6.86

1.35

36/36

1, 7, 11, 6, 9, 2

0.37 – 14.91

4.43

100

11, 1, 7, 2

0.78 – 21.36

11.79

0.01 – 2.73

0.40

103/78

1, 7, 6, 11, 2, 9

0.41 – 20.11

4.74

188

11, 1, 7, 6

0.19 – 23.18

11.64

0.04 – 6.36

0.23

217/114

1, 11, 2, 6, 7, 8

1.21 – 16.18

6.81

217

11, 1, 6, 7, 2

0.50 – 17.33

12.56

0.06 – 4.04

0.35

262/90

1, 11, 2, 9, 6, 7

0.97 – 15.30

6.91

83

11, 6, 1, 7, 9

0.34 – 14.79

5.37

0.05 – 5.54

2.58

79/14

1, 2, 11, 6, 7, 8

1.30 – 18.48

8.28

199

11, 1, 6, 2

1.45 – 17.53

10.51

0.09 – 5.49

0.68

207/27

11, 1, 6, 9, 2, 7

0.63 – 16.50

8.42

53

11, 9, 1, 6, 7, 2

0.07 – 17.55

3.62

0.11 – 3.72

0.67

49/14

1, 2, 11, 6, 7, 9

1.02 – 15.50

8.16

971 408

11, 1, 6, 7, 2, 9 11, 1, 7, 2, 6

0.07 – 23.66 0.18 – 23.66

10.44 11.15

0.01 – 6.87 0.01 – 6.87

0.74 0.24

1031 432

1, 11, 2, 6, 7, 9 1, 11, 7, 6, 2, 9, 8

0.37 – 20.11 0.37 – 20.11

6.89 5.77

510

11, 6, 1, 2, 7

0.34 – 19.34

10.58

0.05 – 5.54

0.84

549

11, 1, 2, 6, 7, 9, 8

0.63 – 18.48

7.67

53

11, 9, 1, 6, 7, 2

0.07 – 17.55

3.62

0.11 – 3.72

0.67

48

1, 2, 11, 6, 7, 9

1.02 – 15.50

8.16

3

20 / Kenter et al.

Bash_SSB interval comprise more than 50 cycles ranging in thickness from 2 to 15 m (6.6 to 49 ft) (Table 1; Figure 5). Commonly, cycles are bounded by thin volcanic ash layers less than a few centimeters in thickness that have been altered by burial diagenesis to resemble shale partings. Closely spaced volcanic ash partings are sometimes observed near bounding surfaces. Thicker ash beds do occur in some parts of the section (e.g., just above the Lvis_SSB and the Serp_SSB) and may exceed 1 m (3.3 ft) in thickness. Laminated ash beds were likely deposited (and preserved) during periods of low energy, either during exposure of the platform top or during the initial flooding and development of low-energy lagoonal deposits. The presence of thick (decimeter to meter) intervals of volcanic ash indicates major periods of exposure and/or deep initial flooding. These are therefore key surfaces that guided the selection of the supersequence boundaries. When presented in the context of their depositional setting related to energy (i.e., bottom agitation) at the time of deposition, lithofacies roughly form a succession of low to higher energy from the base to top of a depositional cycle (see Table 1). One caveat in relating the lithofacies with energy level is that there is no one-to-one correspondence. For example, similar low-energy facies occur both in deeper subtidal as well as in more restricted shallow depositional environments. For discussion purposes, such a succession is shown in Table 1 and partly demonstrates lithofacies stacking from base to top. A brief description is provided in the sections immediately following, whereas the implications for the general depositional system (spatial distribution of lithofacies types and cycle thickness) are discussed in subsequent sections. In brief, the description of lithofacies types and associated depositional energy, from base to top, is as follows: cycles generally start with a thin peloid packstone-wackestone that exhibits subaerial exposure features and are commonly interbedded with thin volcanic ash layers. These lithofacies are overlain by brachiopod and/or crinoid skeletal-peloid grainstone-packstone, which are, in turn, overlain by poorly sorted skeletal-peloid grainstone-packstone and/or well-sorted skeletal grainstone. This succession is, in turn, overlain by coated-grain grainstonerudstone and/or ooid grainstone and locally a skeletal intraclast grainstone or algal grainstone. Obviously, deviations from this generalization exist, and cycle thickness, stacking patterns, and relative contribution of each facies type vary within and across the major sequences.

Peloid Packstone-wackestone (PPW) and Volcanic Ash (VA) Peloid packstone-wackestone caps many shallowingupward carbonate cycles and can overlie any of the other lithofacies. Usually less than 0.5 m (1.6 ft) thick, this lithofacies exhibits rhizoliths, laminated crusts, alveolar fabric, uncommon fenestrae, low-diversity microfauna (ostracods and calcispheres), and uncommon megascopic biota (Figure 6). Peloid packstonewackestone includes the development of a pedolithic zone that occurs at most cycle boundaries, which is interpreted to include both the top of a cycle prior to exposure and the exposure event itself, as well as the initial reflooding event with intermittent exposure. Peloid packstone-wackestone nearly always has volcanic ash interbeds or dispersed volcanic ash (Figure 6A) and, probably as a result, is tight and well cemented. Volcanic ash is locally also dispersed within highenergy facies immediately below the Serp_SSB and below the Serp1_csb in well T-5447 (Figure 5). Those intervals are similarly tight and well cemented. The frequent occurrence of volcanic ash interbeds argues for nearly continuous fallout during deposition of the sequences; ash layers accumulated only during subaerial or low-energy conditions, thereby marking cycle boundaries on the gamma-ray logs. The preservation of volcanic ash is related to terrestrial fallout deposition during exposure and during initial low-energy conditions while the platform was submerged, probably in a shallow, restricted lagoon.

Brachiopod and Crinoid Skeletal-peloid Grainstone-packstone (BSP and CSP) An inferred rise in relative sea level resulted in slightly deeper conditions and deposition of skeletal grainstone to packstone at the base of a cycle. The basal part of the cycles can be either brachiopod skeletalpeloid grainstone-packstone or crinoid skeletal-peloid grainstone-packstone, but brachiopod skeletal-peloid grainstone-packstone is dominant. Brachiopod skeletal-peloid grainstone-packstone is interpreted to represent the continued flooding of the platform in relative low-energy conditions. Thick accumulation of such deposits could indicate deeper water or ongoing restriction caused by emergent outer platform shoals or barriers. The facies is initiated during initial flooding of the platform and is most robust during times of maximum flooding. Brachiopod skeletal-peloid grainstone-packstone is poorly sorted, composed of large, commonly in growth position, thick-walled brachiopods, variable crinoids, diverse algal fragments and foraminifera, and/or corals (Figure 7). Brachiopod

FIGURE 6. (A) Core photos from well T-5246 show peloid packstone-wackestone (PPW) alternating with volcanic ash layers (VA) below the Lvis_SSB. The boundary that shows evidence for subaerial exposure is overlain by poorly sorted skeletal-peloid grainstone-packstone (SPP) with elevated bitumen cement in the lower 70 cm (27 in.) (see Table 2 and Figure 5 for additional lithofacies information). Photomicrographs show the range of PPW textures. Common skeletal grains are tubular algal fragments (B, D), calcispheres (B, E), and minor coated skeletal grains and aggregrate grains (C, D). Width of photomicrographs is 4.20 and 1.58 mm (0.16 and 0.062 in.) (B, C – E, respectively).

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 21

FIGURE 7. (A) Core photos from well T-5246 show shoaling succession of skeletal-peloid grainstone-packstone (SSP) and well-sorted skeletal grainstone (SPG) immediately below the Lvis_9.0_csb. The boundary is overlain by peloid packstone-wackestone (PPW) and a thick interval of brachiopod-skeletal-peloid grainstone-packstone (BSP). Concentrations of in-situ brachiopods and local development of intraparticle and shelter porosity are clearly visible in the BSP interval. Photomicrographs (B – E) show the generally poorly sorted texture and microfractured nature of BSP. Brachiopods are dominant constituents and range from small thin-walled (less than 2 cm [0.8 in.] long and 1 mm [0.04 in.] thick) to large and thick-walled species (respectively, 5– 15 cm [1.9– 5.9 in.] and 1– 10 mm [0.04 and 0.40 in.]). Note the sutured, stylotized contacts of the brachiopods. Other common grain types are crinoid ossicles, peloids, algal fragments, and benthic foraminifera. Width of photomicrographs is 4.20 mm (0.16 in.).

22 / Kenter et al.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 23

skeletal-peloid grainstone-packstone intervals are generally continuous across the platform, appear to thin toward the outer platform, and have the highest frequency and volume in the upper part of the Visean A (between Lvis_13 and Lvis_10.5), where they generally form the basal parts of most cycles (Figure 5). Intervals of brachiopod skeletal-peloid grainstone-packstone occur sporadically in the lower part of the Visean A platform. Two very distinct and thick-bedded (4–10-m; 13–33-ft) intervals, one right above the Lvis9_csb and one below the Lvis13_csb, blanket the entire platform. Crinoid skeletal-peloid grainstone-packstone is coarse and poorly to moderately sorted crinoid-dominated skeletal-peloidal sand. The crinoids commonly have destructionally micritized outer surfaces; other common components are mostly broken brachiopods, green and red algal fragments, and foraminifera (Figure 8). Crinoid skeletal-peloid grainstone-packstone intervals are generally not continuous across the platform and seem to preferentially occur near the platform margin (wells T-6846, T6246, T-8, and T-44). Crinoid skeletal-peloid grainstone-packstone is interpreted as being deposited in moderate to low energy below or around wave base and may represent the time of maximum flooding. These intervals represent the maximum open-marine influx during the cycle, and only one interval (right above Lvis_10.5) covers the entire platform (Figure 5).

Poorly Sorted Skeletal-peloid Grainstone-packstone (SPP) and Well-sorted Skeletal Grainstone (SPG) Poorly sorted skeletal-peloid grainstone-packstone and well-sorted skeletal grainstone generally occur above brachiopod skeletal-peloid grainstone-packstone and crinoid skeletal-peloid grainstone-packstone, exhibit progressively better sorting and less mud, and therefore have been interpreted as shallow-water wave-agitated sand shoal deposits (Figures 9, 10, respectively). These grainstone lithofacies dominate the cycles for most of the Visean A reservoir and commonly show lateral shifts from one facies into the other. In the T-5246 well, well-sorted skeletal grainstone dominates the cycles, and there seems to be a general subtle increase away from the central platform (Figure 5). Immediately above the Lvis2_MFS, a southward thinning interval of wellsorted skeletal grainstone covers the entire platform.

Coated-Grain Grainstone (CgGR), Ooid Grainstone, (OG), and Skeletal-intraclast Grainstone (SIG) In a complete cycle, the previously described lithofacies are overlain by cross-bedded or uniformly laminated, well-sorted coated-grain and/or ooid grainstone,

which are interpreted as high-energy shallow subtidal open-platform to intertidal deposits (Figures 11, 12, respectively). Ooids are generally small, ranging from 0.5 to 1.0 mm (0.02 to 0.04 in.), and were most probably calcitic in origin from their radial structure. Coated grains are commonly skeletal fragments with coatings that are micritic and mainly constructional. Coated-grain grainstone has uncommon occurrences in the Visean A but is common in the Serpukhovian reservoir (Figure 5), where it is concentrated in the northern part of the platform amplifying a general trend from higher energy in the north to lower energy (and/or open marine) toward the south. Coatedgrain and/or ooid grainstones are locally overlain by a relatively thin interval of coarse-grained lithoclastic rudstone-grainstone, skeletal-intraclast grainstone, with centimeter-size subangular to well-rounded intraclasts and Chaetetes (demosponge) fragments, which are commonly coated by micrite (Figure 13). Some intervals display clast-supported flat pebbles, indicative of beach environments. Dominant grain types range from coated grains to ooids and algal grains.

Algal Grainstone (AG) Finally, very well-sorted and cross-bedded, silt to fine sand-size, algal grainstone occurs near the top of the Bashkirian sequences and locally in the Visean A sequence (Figures 5, 14). Algal grainstone intervals were also observed in the outermost platform (Serpukhovian) of the T-4556 well and may indicate the occurrence of this facies in a high-energy outer platform setting. Elsewhere on the platform, algal grainstone alternates with more open-marine well-sorted skeletal grainstone or coated-grain grainstone-rudstone; in these instances, algal grainstone is interpreted as being deposited in the intertidal zone near the top of cycles, perhaps sourced from the outer platform and filling the remaining accommodation space.

Lithofacies Mosaics Figure 15 shows hypothetical depositional models of higher order depositional cycles for both the Bashkirian sequence (Figure 15A) as well as the Visean A and Serpukhovian sequences (Figure 15B). The models show grain compositional and textural trends from the central platform through the outer platform and into the upper flank and indicate some associated bathymetric relief during the Visean A and Serpukhovian. In the central platform, Visean A and Serpukhovian platform cycles are vertically and laterally relatively predictable and show a succession, from base to top, that reflects depositional energy. Near

FIGURE 8. (A) Core photos from well T-5246 show succession of well-sorted coated-grain grainstone (CgGR) overlain by peloid packstone-wackestone (PPW) and capped by the Lvis_10.5_csb. The boundary is, in turn, overlain by crinoid skeletal-peloid grainstone-packstone (CSP) and poorly sorted skeletal-peloid grainstone-packstone (SPP) (see Table 2 and Figure 5 for additional lithofacies information). The CSP interval is only moderately stained by bitumen here in T-5246, whereas toward the outer platform, it is commonly completely stained black. Crinoid skeletal-peloid grainstone-packstone is coarse and poorly to moderately sorted crinoid-dominated skeletal-peloidal sands (B) with destructionally micritized outer surfaces (C) and common components as (mostly broken) brachiopods, (green and red) algal fragments and foraminifera (B – E). Common grain types are crinoid ossicles, brachiopod hash and whole valves, peloids, algal fragments (notably Koninckopora), benthic foraminifera. Width of photomicrographs is 4.20 mm (0.16 in.).

24 / Kenter et al.

contains a shoaling succession of well-sorted skeletal grainstone (SPG) capped by Lvis_9.5_csb. The boundary is overlain by peloid packstone-wackestone (PPW) with a thin volcanic ash ( VA) interbed near the top (see Table 2 and Figure 5 for additional lithofacies information). The skeletal-peloid grainstone-packstone (SSP) and well-sorted skeletal grainstone (SPG) are generally located above the B- and CSP; because they exhibit progressively better sorting and less mud, they have been interpreted as shallow-water wave-agitated sand shoal deposits. Common grain types found in SPP shown in (B – E) are crinoid ossicles, brachiopod hash, peloids, algal fragments, and benthic foraminifera. Grains are commonly coated by thin micrite rims and are poorly sorted. Width of photomicrographs is 4.20 mm (0.16 in.).

FIGURE 9. (A) Core photos from well T-5246 show a thick interval of poorly sorted skeletal-peloid grainstone-packstone (SPP) at the top. The lower half of the core

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 25

by thin peloid packstone-wackestone (PPW), well-sorted skeletal grainstone (SPG), and peloid packstone-wackestone (PPW), with large thin-walled brachiopods (see Table 2 and Figure 5 for legend). Well-sorted skeletal grainstones (SPG) are generally located above the BSP and CSP lithofacies, exhibit better sorting and less mud, and have been interpreted as shallow-water wave-agitated sand shoal deposits. Photomicrographs (B – E) show common grain types of SPG: crinoid ossicles, brachiopod hash, peloids, algal fragments, and benthic foraminifera. Width of photomicrographs is 4.20 mm (0.16 in.) (B, C, E) and 1.58 mm (0.06 in.) (D).

FIGURE 10. (A) Core photos from well T-5246 show shoaling succession of well-sorted skeletal grainstone (SPG) capped by Lvis_10.0_csb. The boundary is overlain

26 / Kenter et al.

FIGURE 11. (A) Core photos from T-5246 show shoaling succession of a thick interval of cross-beds, but laminations are obscured by bioturbation, well-sorted coated-grain grainstone (CgGR), which is overlain by a thin peloid packstone-wackestone (PPW). The Lvis_10.5_csb is overlain by coarse and poorly to moderately sorted crinoid-dominated skeletal-peloidal sands (CSP) (see Table 2 and Figure 5 for legend). In a complete cycle, mixed skeletal-peloidal sands (mostly SPG) are overlain by cross-bedded or uniformly laminated, well-sorted coated-grain and/or ooid grainstone (CgGR and OG), interpreted as high-energy shallow subtidal open platform to intertidal deposits. Common grain types in CgGR shown in (B – E) are well-sorted peloids, coated skeletal grains, algal fragments (tubular), and benthic foraminifera; crinoid ossicles and brachiopod hash are minor grains. Width of photomicrographs is 4.20 mm (0.16 in.) (D) and 1.58 mm (0.06 in.) (B, C, E).

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 27

FIGURE 12. (A) Core photos from T-5246 show part of an approximately 5 – 6-m (16 – 19-ft)-thick interval of cross-bedded, well-sorted ooid grainstone (OG). An exposure surface, with small intraclasts, separates a dark, bitumen-rich interval above from the underlying better cemented, yellow-cream-colored interval in the rightmost core slab (see also Table 2 and Figure 5 for legend). In the Bashkirian, a complete cycle of mixed skeletal-peloidal sands (mostly SPG) is generally overlain by cross-bedded ooid grainstone lithofacies (OG) interpreted as high-energy shallow subtidal open platform to intertidal deposits. Photomicrographs (B – E) show the dominant grain types of ooids and coated skeletal grains; algal fragments (tubular) and benthic foraminifera also occur. Width of photomicrographs is 4.20 mm (0.16 in.) (B, D) and 1.58 mm (0.06 in.) (C, E).

28 / Kenter et al.

grained skeletal-intraclast grainstone (SIG), and capped by the Serp1_csb. The boundary is overlain by thin intervals of PPW and SPG, brachiopod-skeletal-peloid grainstone-packstone (BSP), and another interval of SPG (see Table 2 and Figure 5 for legend). (B– D) Skeletal-intraclast grainstone (SIG) typically has centimetersize subangular to well-rounded intraclasts and Chaetetes fragments, which are commonly coated by micrite. Other grains are coated grains, ooids, and algal fragments. Width of photomicrographs is 4.20 mm (0.16 in.).

FIGURE 13. (A) Core photos from T-5246 show a shoaling succession of well-sorted skeletal grainstone (SPG), overlain by a relatively thin interval of coarse-

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 29

FIGURE 14. (A) Core photos from T-5246 show a thick interval of very well-sorted and cross-bedded, silt- to fine sand-size algal grainstone (AG) underlain and overlain by well-sorted coated-grain grainstone (CgGR); an interbed of volcanic ash (VA) occurs at the bottom (see Table 2 and Figure 5 for legend). Photomicrographs (B-E) show dominant grain types of AG: algal fragments, peloids, and uncommon benthic foraminifera. Typical algal species are Donezella (B, C), paleoberesellids such as Kamaena (D), and beresellids such as Beresella (E). Paleoberesellids are typically Visean and beresellids and Donezella occur in the Serpukhovian to Bashkirian. Width of photomicrographs is 4.20 and 1.58 mm (0.16 and 0.06 in.) (B, C-E, respectively).

30 / Kenter et al.

FIGURE 15. Models of hypothetical higher order depositional cycles for the Bashkirian (A) and the Visean A – Serphuhovian (B) showing grain composition and textural trends from the central Tengiz platform through the outer platform and into the flank. TST = transgressive systems tract; HST = highstand systems tract. See text for discussion.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 31

32 / Kenter et al.

the base, a thin, peloid packstone-wackestone that exhibits subaerial exposure features is commonly interbedded with thin volcanic ash layers. These lithofacies are overlain by brachiopod bioherms and/or crinoid skeletal-peloid grainstone-packstone, which are, in turn, overlain by poorly sorted skeletal-peloid grainstone-packstone and/or well-sorted skeletal grainstone. These lithofacies reflect lowstand and early transgressive restriction providing the environment for trapping volcanic ash fallout (peloid packstonewackestone) and maximum flooding during the crinoid skeletal-peloid grainstone-packstone. This succession is succeeded by highstand deposits that indicate shoaling to (nearly) sea level: coated-grain grainstonerudstone and/or ooid grainstone and locally a skeletal intraclast grainstone or algal grainstone. In the outer platform toward the east, cycles lack the lagoonal base and have a microbial boundstone interval (type 3 microbial boundstone; see Collins et al., 2006) instead, which is superceded by thicker poorly sorted mixed skeletalpeloidal and crinoid sand bodies. Some cycles include thin deposits of high-energy coated-grain grainstonerudstone and/or ooid grainstone and skeletal intraclast grainstone or algal grainstone. Brachiopod bioherms, crinoid skeletal-peloid grainstone-packstone, and poorly sorted skeletal-peloid grainstone-packstone and/or well-sorted skeletal grainstone are absent. The overlying Bashkirian cycles (Figure 15A) are thinner and, in the central platform, show alternating lagoonal peloid packstone-wackestone with subaerial exposure features and interbedded with thin volcanic ash layers, coated-grain grainstone-rudstone and/ or ooid grainstone and skeletal intraclast grainstone. Above the Bash2.5_MFS, coated-grain grainstonerudstone and/or ooid grainstone alternate with algal grainstone intervals, and volcanic ash beds are nearly absent. However, toward the outer platform areas, lowenergy transgressive deposits (brachiopod skeletalpeloid grainstone-packstone, crinoid skeletal-peloid grainstone-packstone, poorly sorted skeletal-peloid grainstone-packstone) thicken and may even include microbial boundstone intervals (T-4556), while highenergy cycle tops (coated-grain grainstone-rudstone, ooid grainstone) are thinning or absent (T-6743). Obviously, deviations from the generalizations of Figure 15 exist, as will be discussed in the following sections. Cycle thickness, stacking patterns, and relative contribution of each facies type vary within and across the major sequences. As an example, the bathymetric profile changes toward the Visean A–Serphukhovian outer platform to a deeper setting that brings microbial boundstone on the platform during flooding. In con-

trast, the Bashkirian sequence shows a flat-topped platform with parallel facies zones dominated by higher energy grainy lithofacies types. High-frequency cycles are much thinner in the Bashkirian, at meter scale, and show mostly alternating coated grain grainstone, ooid grainstone, and algal grainstone with sedimentary structures. The Bashkirian rim was probably flat during deposition and shows thinning of platform facies that border a nearly starved slope system.

Visean A and Serpukhovian: Trends and Depositional Environments Visean A cycles consist almost entirely of grainy facies (packstone and grainstone) that feature depositional environments ranging from restricted lowenergy brachiopod skeletal-peloid grainstone-packstone to moderate- to low-energy deeper subtidal environments characterized by more poorly sorted and open-marine poorly sorted skeletal-peloid grainstonepackstone as well as crinoid skeletal-peloid grainstonepackstone and high-energy well-sorted grainstone shoals (well-sorted skeletal grainstone and uncommon coatedgrain grainstone-rudstone). General trends in lithofacies distribution, cycle thickness, and stacking suggests the presence of three intervals in the Visean A and Serpukhovian reservoir zones, each displaying certain depositional trends (Figure 5). These are the intervals between the Lvis_SSB and Lvis10.5_csb, Lvis10.5_csb and Lvis13_csb (which straddles the second-order Lvis2_MFS), and Lvis13_csb to the Serp_SSB. In the Lvis_SSB to Lvis10.5_csb interval, the platform cycles are dominated by the poorly sorted, openmarine poorly sorted skeletal-peloid grainstonepackstone, with minor intercalations of brachiopod skeletal-peloid grainstone-packstone and/or crinoid skeletal-peloid grainstone-packstone. Crinoid skeletalpeloid grainstone-packstone intercalations generally occur near the base of cycles, whereas well-sorted skeletal grainstone intercalations appear near the top. Cycle caps are thin, decimeter-thick, micropeloidal packstone-mudstone intervals that have, in nearly all cases, one or two thin volcanic ash interbeds, but generally display moderate subaerial exposure. Cycle thickness ranges from 1–2 m (3.3–6.6 ft) near the base to more than 15 m (49 ft) in the middle of the interval with an average of 5–8 m (16– 26 ft). Poorly sorted skeletal-peloid grainstone-packstone intervals are dominant in the central platform around wells T-220 and T-5848 but gradually are partially replaced by higher energy sand shoals (well-sorted skeletal

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 33

grainstone) toward the outer platform areas (well T-5246 to the north and wells T-6246 and T-6846 to the southwest). Crinoid-rich intervals represent maximum, open-marine flooding and the deepest lithofacies. These occur preferentially in the outer platform toward the southwest and seem to thin and pinch out toward the central platform. One interval, however, runs contrary to this trend and appears to blanket the entire platform at the top of the Lvis_SSB to Lvis10.5_csb interval. Similarly, one nearly 10-m (33-ft)thick and very continuous brachiopod skeletal-peloid grainstone-packstone interval covers the platform right above the Lvis9_csb boundary. This general depositional picture suggests an overall low-energy platform, subtidal or below wave base, with slightly shallower outer platform sand shoals and a dominant southern open-marine influence. The cycles in this succession never filled the total accommodation space, despite the relatively low-amplitude sea level oscillations. The Lvis10.5_csb to Lvis13_csb interval, which straddles the supersequence flooding surface (Lvis2_MFS), shows a similar central versus outer platform facies pattern as the previous interval but has much thinner, 2–5-m (6.6–16-ft)-thick cycles and frequent brachiopod skeletal-peloid grainstone-packstone intervals. These brachiopod interbeds seem to thin away from the platform center and are replaced by higher energy mixed skeletal-peloidal sands to the south and highenergy, well-sorted sands to the north. In addition, the section has a crinoid-rich base, a thick interval of high-energy sand shoals in the middle, and a thick brachiopod interval (more than 10 m) near the top. Each of these intervals is continuous across the entire platform. The crinoid-dominated base indicates maximum deepening of the platform coinciding with the Lvis_MFS surface. Although relative depositional conditions did not change and accommodation space was still underfilled, it suggests that sea level fluctuations increased in frequency. The Lvis13_csb to the Serp_SSB interval in the HSS of the supersequence has similar thin cycles, 2– 5 m (6.6– 16 ft) in thickness, but those show more pronounced subaerial exposure near the cycle tops. Furthermore, there is an overall trend of higher energy coated-grain and ooid sand shoal deposits with elevated amounts of dispersed volcanic ash toward the north (wells T-5447, T-5044, and T-5444) and slightly deeper, lower energy, environments toward the south, near the outer platform (wells T-6246 and T-6846). Cycles were nearly completely filled in the northern platform area, multiple ash layers occur near sequence boundaries, and there is an increased expression of

exposure toward the Serp_SSB. These observations suggest the development of a slightly deeper and lower energy outer platform toward the southeastern margin (possibly caused by structural tilting). The thinning of cycles, increasing exposure of cycle tops, and lateral facies differentiation during the Serpukhovian coincide with the development of an unusual deeper water outer platform toward the northeastern to eastern margins of the Tengiz platform, where upper slope facies consisting of algal-microbial, skeletalmicrobial, and microbial boundstone prograded several kilometers out into the basin on a preexisting Visean A and B shelf edge. In these regions of the outer platform, which generally have lower energy and probably deeper water facies, exposure expression is reduced or absent (see Figure 15).

Bashkirian: Trends and Depositional Environments Bashkirian cycles are thinner and less systematic than those recognized in the Visean A and Serpukhovian, most likely because of rapid sea level fluctuations produced during global icehouse conditions. The Bashkirian consists of three intervals, of which the lower (Serp_SSB to Bash1_csb) is essentially a 15– 20-m (49–66-ft)-thick ooid sand body covering the entire platform (Figure 5). This unit is nearly tabular in shape, has locally thin, mixed skeletal-peloidal openmarine intercalations, and is cyclic with thin mudstone cycle caps and associated exposure surfaces. In the central platform, the lower half of the interval commonly contains high bitumen content and has low porosity; toward the southwestern outer platform (T-6246) and north (T-5246), nearly the entire section has bitumen and strongly reduced porosity. The middle part of the Bashkirian is bounded above by a stratigraphic surface (Bash2.5_MFS) that was not part of the original framework published by Weber et al. (2003). The Bash1_csb to Bash2.5_MFS is dominated by thin (a few meters) shoaling cycles of sorted skeletal subtidal open-marine sands, overlain by highenergy intertidal coated grain to ooid sand shoals, and capped by thin peloidal packstone-mudstone intervals with general extensive subaerial exposure and intercalated centimeter- to decimeter-thick volcanic ash beds. In the uppermost part of the Bashkirian (Bash2.5_MFS to Bash_SSB), ooid to coated grain sands occur in the basal part of shoaling cycles and are overlain by silt-size (high-angle), cross-bedded, wellsorted algal sands. Lagoonal packstone-mudstone intervals and associated volcanic ash beds are uncommon or

34 / Kenter et al.

lacking completely. The algal fragments are silt-size, dominant, and probably tubular red algae in origin. The waning of ooid domination and introduction of algal silt above the Bash2.5_MFS may indicate a general deepening stratigraphic trend toward the Bash_SSB, which is consistent with an interpretation of the subsequent drowning of the platform (Weber et al., 2003). Cross-bedding was possibly generated by oceanic currents sweeping the deeper water platform top during the progressively longer highstands; the platform was still exposed during shorter lowstands associated with high-frequency and high-amplitude (as much as 40–60-m [131–196-ft]) sea level changes of the icehouse world. Expressions of subaerial exposure, such as dissolution, recrystallization, and preservation of ash layers are also nonuniform from cycle to cycle. Despite this, Bashkirian cycles retain some similarities to Visean A and Serpukhovian cycles, such as the presence of high-energy shoals and deeper platform skeletal grainstone and the association of semirestricted to restricted facies near sequence boundaries. In the Bashkirian, high-energy shoals are represented by oolites, pisolites, and coated grains, whereas the deeper and restricted facies are similar to those present in the Serpukhovian and Visean A.

Interplay between Minor Tilting and Deposition A remarkable observation from the cross section in Figure 5 is that the railroad-track pattern of sequence and cycle boundaries below the Serp_SSB is nearly perfectly parallel, with the exception of the section in T-6846. Another observation is the presence of a minor angular unconformity of several degrees between the Serp_SSB and the underlying cycle breaks, wherein both the Serp_SSB and cycle breaks dip to the south a few degrees. In addition, the Serp_SSB appears to have variable erosional relief of as much as 5 m (16 ft) across the platform as indicated by the apparent increased erosion in well T-5447. These observations suggest several things. First, fairly gradual and constant vertical aggradation across the central platform resulted in a flat platform top with minimal local relief as a result of immediate compensation of possible variations in depositional topography. In addition, the platform was tilted to the south prior to the Serp_SSB, subsequently eroded (based on the missing Bogdanovsky and Suryansky biozones), generating as much as 5-m (16-ft) relief, and finally, tilted a few more degrees along the same axis during the post-Bashkirian. Well T-6846 shows about a 5-m (16-ft) thicker section relative to other wells of Figure 5 between the Lvis10.5_csb

and Lvis_SSB, a nearly 5-m (16-ft) thicker section between the Lvis13_csb and Lvis10.5_csb, probable minor erosion at the Lvis13_csb, and several additional meters missing section below the Serp_SSB. Although not proven, these differences may indicate that the southwestern margin tilted downward some 10 m (33 ft) or more prior to the Serp_SSB and was eroded at the time of the Serp_SSB following minor tilting to the north.

RESERVOIR QUALITY AND DIAGENESIS Pore Types Air permeability data from approximately 11,000 plugs clearly suggest the presence of three general reservoir environments based on plug-scale matrix properties: central platform, outer platform, and rim slope (flank) (Collins et al., 2006). The Visean A, Serpukhovian, and Bashkirian central platform reservoir can be regarded as having distinctly different gross reservoir properties that are generally well behaved as a whole. Porosity-permeability crossplots and porosity histograms suggest different behavior between the northern wells (T-220 and T-5246) and southern wells (T-6246 and T-6846) in the central platform (Figure 16). Trends of generally lower mean porosities and better behaved porosity-permeability relationships occur in the southern wells, whereas bimodal porosity distributions occur in the northern wells. Pore-type characterization of nearly 2000 plugs (and associated thin sections) from the nearly bitumenfree T-220 well and the moderately bitumen-cemented T-6246 well show the presence of varied pore types like moldic, vugs, microporosity, interparticle porosity, intercrystalline, and intraparticle porosity and the general absence of fractures (Figure 17). The porosity classification scheme of Choquette and Pray (1970) defined numerous carbonate pore types, either fabric selective (including interparticle, intraparticle, intercrystalline, and moldic) or not fabric selective (including fracture and vug). There were implications for porosity-permeability relationships, of course, but this was not emphasized. Lucia’s (1995, 1999) landmark study showing the direct link between rock type, porosity type, and reservoir quality (porositypermeability relationships or transforms) changed the context of some of these classic pore types. Lucia (1995, 1999) recognized interparticle pore types and vuggy pore types. Interparticle pores, which have a more straightforward porosity-permeability transform that can be tied to rock type, grain size, and sorting, are instances where the pore space occurs between particles and is not significantly larger that the particles.

FIGURE 16. Subtle porosity-permeability differences are observed between four wells in the central platform; see Figure 4 for well locations and Figure 5 for their facies and stratigraphy. K-Phi plots (top) and porosity histograms (bottom) suggest different behavior between the northern wells (T-220 and T-5246) and southern wells (T-6246 and T-6846) with generally lower mean porosities and better behaved permeability-porosity relationships in the southern wells and bimodal porosity distributions in the northern wells.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 35

36 / Kenter et al.

FIGURE 17. Pore and cement types recognized in the T-220 and T-6246 wells are generally similar. However, relative contributions by vuggy, intercrystalline, and moldic pore types are significantly higher in T-220 for all facies types than in T-6246, which may be related to the larger average grain size and higher contribution by the mud-lean rock types in T-6246. Similarly, cement types in T-6246 show a higher abundance (frequency) of equant calcite and syntaxial cements than in T-220. This may suggest a larger volume of deep burial cements filling pores at T-6246, but clearly, more quantitative data are required for such assessment. Pore types across the bottom of the figures are 1 = interparticle; 2 = intraparticle; 3 = fenestral; 4 = shelter; 5 = growth framework; 6 = intercrystalline; 7 = moldic; 8 = microporosity; 9 = fracture; 10 = channel; 11 = vug; and 12 = enhanced dissolution. Cement types are 1 = fibrous; 2 = bladed; 3 = equant to rhombic; 4 = coarse crystalline; 5 = botryoidal; 6 = syntaxial; 7 = micritic; 8 = meniscus; 9 = microstalactitic; 10 = poikilotopic; 11 = neomorphic spar; and 12 = grain rims. Plotted are relative dominance of pore or cement types in terms of visual presence with the number indicating their ranking. See text for discussion. Vuggy pore types are typically larger than the grains or occur within the grains, and these can be separate or touching. Separate vugs are connected through the interparticle pore system, whereas touching vugs themselves form an interconnected system. At Tengiz, separate vugs that add to porosity but not permeability include moldic and intraparticle porosity, so these must be subtracted from total porosity to use interparticle transforms. Touching pore types include some vugs (solution-enlarged interparticle pores) and fractures, which add porosity and permeability independently from the interparticle component, such that interparticle transforms must

be modified. Clearly, real-world reservoirs in limestone can be even more complex with the addition of microfractures, intercrystalline porosity, or microporosity connecting separate vugs, and/or the problem of assessing in two dimensions with petrography the three-dimensional complexity of porosity. The following sections describe the pore types commonly observed in Tengiz samples and their relative abundance in the central platform. Pore types were identified in the Tengiz cores and thin sections generally following the terminology of Choquette and Pray (1970) to first assess their relation to lithofacies and then explore the relation

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 37

FIGURE 18. Photomicrographs illustrating the dominant pore types in the central platform area. (A) Vugs are nonfabric-selective pores commonly larger than the dominant grain size (although different interpretations exist in the literature). Examples of vugs shown here are the result of enhanced dissolution in a grainstone following early equant rim cementation. (B) Vugs developed in a packstone-grainstone, dissolving both grains and matrix but leaving crinoid ossicles and brachiopod fragments intact. (C) Overpacked ooid grainstone with vugs that postdate early cementation but predate blocky calcite and bitumen emplacement. Vugs are a dominant pore type in the central platform and seem to be mostly related to deep-burial dissolution instead of initial meteoric processes. (D) Advanced stage of enhanced dissolution of moldic and interparticle porosity (and minor intraparticle) leading to the destruction of cement bridges and development of vugs. (E) Skeletal and coated grain grainstone dominated by interparticle porosity with minor cement fill. (F) Overpacked ooid grainstone with minor blocky calcite and bitumen occluding solution-enhanced interparticle porosity. (G) Skeletal and peloid grainstone with rims of equant spar lining interparticle pores and minor moldic porosity. (H) An example where the interparticle porosity is completely occluded by equant to rhombic spar and bitumen cement. Width of photomicrographs is 4.20 and 1.58 mm (0.16 and 0.06 mm) (A – G, H, respectively). See text for discussion. between lithofacies and reservoir quality (Table 2). Common pore types in the Tengiz central platform are vugs, interparticle, microporosity, intercrystalline, moldic, and intraparticle with generally the highest permeability occurring in vuggy-dominated matrix. The most common pore type is vuggy, which is, in part, fabric selective, being found in matrix, grainy areas (solution-enhanced interparticle pores), or associated with stylolites (Figure 18A–C). The second most common pore type is interparticle porosity (Figure 18D–H), followed by microporosity (Figure 19A, B). Vuggy porosity includes moldic porosity when considered in terms of porosity-permeability trans-

forms (Lucia, 1995, 1999). Figure 19C – E shows successive stages of the development of moldic porosity at Tengiz: in its early phase where micrite grains are gradually being dissolved following partial fill of interparticle porosity by equant to rhombic spar cement (Figure 19C); a phase of progressive dissolution shown by corrosion of the mold boundaries, nearly breaking through at contacts and corroding earlier cement (Figure 19D); and an advanced stage where enhanced dissolution left only micrite envelopes. Intraparticle porosity, which is also included as vugs in terms of porosity-permeability transforms, is shown in Figure 19D–F.

38 / Kenter et al.

FIGURE 19. Photomicrographs illustrating additional dominant pore types in the central platform area. (A, B) Examples of microporosity, pore size smaller than 25 Mm, developed in grains (A) and matrix (B). This type of pore is generally difficult to observe in thin sections because of their thickness, commonly about 25 – 45 Mm. (C) Moldic porosity in its early phase: micrite grains are gradually being dissolved following partial fill of interparticle porosity by equant to rhombic spar cement. (D) Moldic porosity showing evidence for enhanced dissolution shown by corrosion of the mold boundaries, nearly breaking through at contacts and corroding earlier cement. (E) Moldic porosity showing further stage of enhanced dissolution leaving only micrite envelopes. Bitumenstained micrite appears to slow down dissolution. Interparticle porosity is partially filled with calcite cement. (F) Intraparticle porosity developed in benthic foraminifera. (G) Intraparticle porosity developed in a green alga, Koninckopora. (H) Intraparticle porosity developed in a Chaetetes fragment (not all pores are filled by blue epoxy). Width of photomicrographs is 4.20 mm (0.16 in.) (A, B, F, G) and 1.58 mm (0.06 in.) (C, D, E, H). See text for discussion.

Examples of an advanced stage of dissolution of moldic and interparticle porosity leading to the destruction of cement bridges and development of vugs with minor blocky calcite and bitumen occluding porosity are shown in Figure 18D and F. Commonly, moldic porosity co-occurs with interparticle porosity as shown in Figure 18G; interparticle porosity is completely occluded by equant to rhombic spar and bitumen in some samples (Figure 18H). Microporosity in grains or matrix is difficult to observe in Tengiz samples because pore sizes are much smaller than the thickness of a typical thin section. Microporosity includes fine intercrystalline porosity because both are

generally beyond normal petrographic resolution and, therefore, difficult to distinguish. Fractures and fissures are generally uncommon in the central platform, and where present, they are mostly occluded by cement. In some few cases where the fractures remain open, they range from microhairline fissures to sharp microfractures (Figure 20A–F). Stylolites are generally stained by bitumen and commonly associated with small vugs (Figure 21A – C). Examples demonstrate that such dissolution postdates stylolite formation.

Parameters Controlling Porosity and Permeability The relative contributions by vuggy, moldic, and intercrystalline porosity are significantly higher for all facies types in well T-220 than in well T-6246. In contrast, the relative contribution by interparticle porosity is higher at T-6246, a change that is most likely related to the higher percentage of grainy, mud-lean

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 39

FIGURE 20. Photomicrographs illustrating dominant pore types in the central platform area. (A, B) Generating thin sections from core plugs with open fractures is difficult if not impossible. Shown here are examples of thin fractures that are filled by blocky calcite cement. (C, D) Examples of hairline fissures or fractures filled by bitumen. Note that the fissure in (C) postdates blocky calcite interparticle cement fill but predates bitumen emplacement and confirms the very late timing of bitumen invasion. (E, F) An exception to the rule: vague outline of open fractures that have been enhanced by late-stage dissolution in a skeletal-peloidal grainstone with high interparticle porosity. Fractures may connect vugs as shown in these samples. Width of photomicrographs is 4.20 mm (0.16 in.). See text for discussion.

rock types in that well (Figure 22). Nevertheless, at a gross scale, none of the individual pore types or combination of pore types seem to have a significant firstorder relationship with primary depositional facies nor have any significant control on porosity or permeability. This suggests that the central platform area could be regarded as one single reservoir region wherein changing pore types and lithofacies do not significantly impact true reservoir connectivity. This suggestion is rather surprising and seemingly contradicts Lucia’s (1995, 1999) reservoir rock-type classification that suggests pore types control permeability behavior. Although the observations above suggest minimal first-order control by pore type or facies, several secondorder relationships should be acknowledged. Porosity distributions and pore-type contributions for the muddy lagoonal facies (peloid packstone-wackestone), moderately sorted grainy facies with mud matrix (brachiopod skeletal-peloid grainstone-packstone, crinoid skeletal-peloid grainstone-packstone, and poorly sorted skeletal-peloid grainstone-packstone) and mud-lean well-sorted grainy facies (well-sorted skeletal grainstone, coated-grain grainstone-rudstone,

ooid grainstone, skeletal intraclast grainstone, and algal grainstone) that make up the cyclic pattern shown in Figure 5 do show subtle but significant differences (Figure 22). Pore type contributions in the grainy facies groups are very similar, but muddy lagoonal facies have a higher contribution of intraparticle and fracture porosity. The grainy facies in well T-6246 have a higher contribution of interparticle porosity but also differ in the overall distribution of porosity. These distributions are remarkably different between the two wells with bimodal and generally higher porosities in grainy facies and low porosities in lagoonal facies in well T-220, whereas the porosity distributions in well T-6246 show highly reduced and flattened distributions for the mud-lean facies compared to T-220 (Figure 23). Similarly, the mud-rich grainy facies have reduced porosity, but the lagoonal mudstone has higher porosity, displaying a bimodal distribution that is possibly related to a higher contribution of fracture porosity. The above suggests that T-220 porosity has a relatively higher contribution by pore types that are the result of corrosive processes, like vug and moldic porosity, an observation that can be made for T-5246 as well. In addition to pore type, observations were made on the type of pore-filling agents, which resulted in two significant trends. First, the distribution of porefilling cement types for both wells T-220 and T-6246

40 / Kenter et al.

FIGURE 21. Photomicrographs illustrating the dominant pore types in the central platform area. (A) Pressure solution stylolite with dark-brown micrite and possibly bitumen. (B) As in (A), but dissolution associated with stylolite. (C) As in (A, B), but here, dissolution postdating stylolite formation is clearly visible; note that the vug is not breaking through the bitumen seam. (D) Poikilotopic syntaxial overgrowth cements on crinoid ossicles are locally dominant and reduce interparticle porosity. However, invasion by bitumen and associated dissolution was able to corrode cement and crinoid ossicles. (E) Coated grainstone where equant to rhombic spar is occluding interparticle pore space, probably replacing earlier marine cements, and only sparse molds and remaining interparticle pores remain. (F) Micritized coated grainstone with evidence for minor corrosion of grain boundaries followed by burial blocky calcite that occludes part of the interparticle porosity. Micrite is locally invaded by bitumen. (G) Detail of partially open fracture that was occluded by blocky calcite cement and followed by bitumen. (H) Porosity in poorly sorted grainstone was occluded by cement and completely recrystallized to fine calcite spar cement. Width of photomicrographs is 4.20 mm (0.16 in.) (A – C, E – H) and 1.58 mm (0.06 in.) (D). See text for discussion.

appears to be comparable, with the exception of a relatively higher contribution by equant and syntaxial calcite at T-6246 (Figure 17). Another obvious difference between the T-220 and T-6246 wells is the amount of bitumen; Figure 5 shows qualitatively higher amounts of bitumen in T-6246 (and T-6846) than in T-220. For well T-220, bitumen volume is mostly associated with muddy grain-rich to mud-rich facies (Figure 24A) and dominant pore types like interparticle, intercrystalline, and vug with minor contributions by intraparticle, moldic, and fracture (Figure 24B). Bitumen content shows a negative relationship with porosity (Figure 24C) and permeability (Figure 24D).

Figure 25 better shows the influence of bitumen volume on the porosity versus permeability relationship in well T-220 (no such data are yet available for T-6246). Increasing bitumen content above a threshold of approximately 0.5% rock volume is negatively correlated with K-Phi. Although bitumen commonly occurs in grainy facies with matrix pore types, it also occurs with late diagenetic pore types like vugs as well as with nonmatrix pore types like fractures. Bitumen content significantly impacts the cyclic and stratigraphic porosity variations as shown by Figure 26; it also causes the observed decrease in porosity toward the outer platform. Porosity logs corrected for bitumen volume are easier to correlate and confirm that the distribution of bitumen increases toward the outer platform (Figure 26). Clearly, bitumen reduced the porosity toward the outer part of the central platform and even more so into the outer platform itself (Figures 4, 5).

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 41

FIGURE 22. Pore-type contributions for muddy lagoonal facies (13 = PPW) and moderately sorted grainy facies with mud matrix (2 = BSP, 31 = CSP, 32 = SPP) and mud-lean, well-sorted grainy facies (41 = SPG, 42 = CgGR, 43 = OG, 44 = SIG, 46 = AG) in both T-220 and T-6246 wells. Pore types are 1 = interparticle; 2 = intraparticle; 3 = fenestral; 4 = shelter; 5 = growth framework; 6 = intercrystalline; 7 = moldic; 8 = microporosity; 9 = fracture; 10 = channel; 11 = vug; and 12 = enhanced dissolution. Plotted are relative dominance of pore types in terms of visual presence with the number indicating whether pore types were first, second, or third in importance. See text for discussion. Volcanic ash also produces low-porosity (tight) zones. These tight zones tend to occur around cycle boundaries and within intervals that have dispersed volcanic ash. One mechanism explaining these observations associates volcanic ash with inhibited downward fluid percolation during early diagenesis associated with sequence and cycle boundaries. This condition generally reduced the effect of meteoric dissolution. In addition, silica leached from the volcanic ash was, in some cases, redeposited as chert and silica cement below the sequence boundaries, thereby adding to the porosity reduction of the associated lagoonal intervals. Finally, the dissolution associated with the bitumen emplacement did not affect those tight zones.

Processes and Timing of Diagenesis The present-day distribution of reservoir quality in the central and (southwestern) outer platform in the Lvis_SSB to Bash_SSB platform succession (Visean A, Serpukhovian, and Bashkirian reservoir zones) was

determined by the combined effect of late diagenetic modification, specifically bitumen invasion and associated corrosion, of an earlier reservoir system whose porosity and permeability distribution was established by highly cyclic depositional and early diagenetic processes (Figure 27). Early diagenesis includes meteoric alteration in cyclic platform facies associated with major sequence boundaries and reduced dissolution along sequence and high-order cycle boundaries associated with the presence of volcanic ash. Diagenetic processes associated with subaerial exposure produced both an increase and reduction of porosity (Figure 27). Brown fibrous, mycrostalactitic, meniscus, and pendant cements are commonly associated with meteoric vadose and marine vadose diagenesis, and these are observed in Tengiz cores to several meters below a suspected exposure surface (Figure 28A – D). Possible examples of alveolar fabrics indicating soil formation below an exposure surface are shown in Figure 28E. Early dissolution related to

matrix (2 = BSP, 31 = CSP, 32 = SPP), and mud-lean well-sorted grainy facies (41 = SPG, 42 = CgGR, 43 = OG, 44 = SIG, 46 = AG). See text for discussion.

FIGURE 23. Porosity distributions for the major facies types in T-220 and T-6246: muddy lagoonal facies (13 = PPW), moderately sorted grainy facies with mud

42 / Kenter et al.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 43

FIGURE 24. Relationship between bitumen (Bit rock), rock fabric (PP fabric), pore type (pore 1), porosity (Phi), and permeability (K ) in well T-220 (no such data are yet available for T-6246). Bitumen content is mostly associated with grainy mud-lean to mud-rich facies (A) and dominantly in interparticle, intercrystalline, and vug pore types with minor contributions by intraparticle, moldic, and fracture pores (B). Bitumen contents above 0.5% rock volume shows a negative relationship with porosity (C) and permeability (D). Pore types are 1 = interparticle; 2 = intraparticle; 3 = fenestral; 4 = shelter; 5 = growth framework; 6 = intercrystalline; 7 = moldic; 8 = microporosity; 9 = fracture; 10 = channel; 11 = vug; and 12 = enhanced dissolution. Textures are 1 = rudstone; 2 = grainstone-rudstone; 3 = grainstone; 4 = grainstone-packstone; 5 = packstone-grainstone; 6 = packstone; 7 = packstone-wackestone; 7 = wackestone-packstone; 8 = wackestone; 9 = wackestone-mudstone; 10 = mudstone-wackestone; 11 = mudstone; and 12 = boundstone. See text for discussion.

exposure may have led to early compaction as shown by the frequent occurrence of fitted fabrics (Figure 28H), where the thin cement rim is included in the compaction and postdated by blocky calcite cement and bitumen. Later stage diagenesis includes pressure solution culminating in stylolites with concentrations of darkbrown micrite and bitumen (Figure 21A, B). Dissolution is sometimes observed postdating the stylolite formation (Figure 21C). Poikilotopic syntaxial overgrowth cements on crinoid ossicles are locally a dominant cement, reducing interparticle porosity (Figure 21D). Later burial processes also reduced porosity by cementation and recrystallization (Figure 21E – H). Examples show progressive occlusion of pore space during

this process, which includes cementation of pore space created by fracturing (Figure 21G). Burial dissolution events that added porosity are of great interest and paramount importance to the Tengiz reservoir and its interpretation with respect to porosity modeling (Figure 29A–F). A very common observation in samples that have interparticle porosity is enhanced (late-burial?) dissolution that enlarges and connects interparticle pore space into larger vugs. These vugs are lined by bitumen, attesting to the timing of the dissolution relative to bitumen emplacement (Figure 29A). Grain boundaries are commonly corroded by this dissolution event, as is syntaxial overgrowth cement in some cases (Figure 29B, C). Extensive late deep burial dissolution prior to oil migration has been

44 / Kenter et al.

FIGURE 25. Influence of bitumen volume on the porosity (PHI) versus permeability (K ) relationship in well T-220 (no such data are yet available for T-6246). Increasing bitumen content is negatively correlated with K-Phi above a certain threshold of about 0.5% rock volume. Although bitumen is common in grainy facies with matrix pore types, it also occurs with late secondary pore types like vugs indicative of dissolution as well as with nonmatrix pore types like fracture porosity. See text for discussion.

reported as a major reservoir porosity generation mechanism in, at least, several other carbonate fields (Moore, 2001; Esteban and Taberner, 2003; Zampetti et al., 2003; Sattler et al., 2004). Most of these authors suggest corrosion by mixing of formation fluids with an external fluid at higher temperatures, and such process may be responsible for the burial corrosion in the Tengiz central platform as well. Similarly, the corrosion event postdating the bitumen invasion (and cementation) may represent yet another such leaching phase with different fluids. Fluid-inclusion and stable isotope studies will shed light on the origin of the substantial burial corrosion at Tengiz; see Collins et al. (2006) for discussion of preliminary geochemical data and a probable dissolution model. Recrystallization is also an important process in the central platform diagenesis. As an example, Figure 29D is a grainstone that was recrystallized to equant and blocky spar. This recrystallization followed early partial cementation of interparticle porosity and was, in turn, followed by late dissolution that produced moldic porosity. Note that early moldic porosity was

occluded by blocky calcite (see micrite envelope), whereas this later dissolution phase generated the open molds that are lined by bitumen. More common are finely recrystallized fabrics associated with intercrystalline and minor moldic porosity, as well as the presence of scattered bitumen (Figure 29E, F). Although the recrystallization phase seems to postdate (corrode and truncate) blocky calcite cementation, the exact timing remains unclear because of the destructive nature of the process. The important diagenetic overprint in the Tengiz central platform as outlined in Figure 27 can be summarized as follows. Syn- and postdepositional dissolution in meteoric environments generated moldic, enlarged interparticle and intraparticle, and microporosity pore types, as well as vuggy microkarst. Cementation of lagoonal intervals associated with the presence of discrete and dispersed volcanic ash significantly reduced porosity and created permeability baffles around cycle and sequence boundaries. Intermediate shallow burial produced commonly observed fitted fabrics and low- to medium-amplitude stylolites. The lateral continuity and persistence of tight layers around sequence and cycle boundaries as well as in local highenergy intervals probably greatly affected later fluid flow and, thus, the ultimate distribution of cements, dissolution, and bitumen in the central platform reservoir. The late diagenetic overprint was responsible for the enhanced dissolution, increasing any existing

for well locations and Figure 5 for additional correlations). K-Phi plots and porosity histograms suggest different behavior between northern central platform wells (T-220 and T-5246) and southern central platform wells (T-6246 and T-6846), with generally lower mean porosities and better behaved permeability-porosity relationships in the southern wells and bimodal porosity distributions in the northern wells. Bash = Bashkirian; Serp = Serpukhovian; VisA = Visean A. See text for further discussion.

FIGURE 26. Subtle but significant differences in petrophysical and geological properties are observed between these four wells in the central platform (see Figure 4

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 45

46 / Kenter et al.

FIGURE 27. Paragenetic sequence for the central platform. Significant syndepositional, intermediate (shallow)-burial, and deep (late)-burial events are listed in chronological order. In the lower part of the figure, those events that are particularly important to porosity creation and destruction are listed. MO = moldic; BP = interparticle; IP = intraparticle; MP = microporosity; MVUG = micro vug. See text for discussion.

porosity in the matrix and forming vugs and minor fractures, whereas equant calcite cement reduced porosity away from the interior central platform. The subsequent bitumen invasion exerted an overall porosityreducing effect increasing toward the outer platform. These late-burial events that increased and reduced porosity significantly dampened or flattened initial vertical, nearly cyclic, porosity variations linked to primary depositional properties.

Spatial Distribution of Central Platform Reservoir Quality In the absence of any significant first-order control by pore type or lithofacies, the following sections describe the reservoir quality in the context of the transition from central to outer platform environments. This lateral change is likely a result of the early cyclic depositional and diagenetic processes, including the preservation of volcanic ash, coupled with late-burial dissolution and bitumen emplacement.

Figure 30 presents a model that summarizes porosity generation and reduction events in the central Tengiz platform from the standpoint of gammaray and porosity logs. Porosity changes are linked to the major diagenetic phases following deposition of alternating thin, low-energy lagoonal intervals with volcanic ash beds and thick intervals of high-energy grainstone. Detailed log and core correlation shows that lagoonal interval and ash beds are not always associated and/or continuous between well locations, and cycles may vary slightly in thickness over kilometers distance. Three important diagenetic stages that affect the reservoir quality and distribution are observed. First, porosity curves with a blocky shape are the result of early-burial processes, including low porosity around cycle boundaries related to the presence of thin ash beds and corrosion and cementation related to meteoric diagenesis (Figure 30A). Locally thick grainstone intervals have low porosity and show the presence of elevated levels of dispersed ash concentrated in frequent stylolites. Second, a

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 47

FIGURE 28. Photomicrographs illustrating some of the dominant cement types in the central platform area. (A, top to the right) Brown fibrous pendant cement attached to lower part of brachiopod fragments is interpreted as the result of meteoric diagenesis and, therefore, suggests subaerial exposure above. (B) Mycrostalactitic or pendant cements, which are commonly associated with vadose diagenesis, are observed to several meters below a suspect exposure surface. (C, D) Meniscus and pendant cements, indicating vadose to marine vadose diagenesis occlude much of the primary porosity in these samples. Burial dissolution and/or corrosion of preserved interparticle porosity are followed by blocky calcite cement and bitumen. (E) Possible example of alveolar fabrics indicating soil formation below exposure surface. (F) Possible vadose cementation followed by enhanced corrosion or dissolution of grain boundaries and cement, blocky calcite cement, and bitumen. (G, H) A combination of minor cementation and compaction leads to fitted fabrics or overpacked fabrics. Here, a thin cement rim is included in the compaction process and followed by burial blocky calcite spar cement and bitumen. Width of photomicrographs is 4.20 mm (0.16 in.) (A, C, D, E, F) and 1.58 mm (0.06 in.) (B, G, H). See text for discussion.

burial event led to corrosion and enhanced blocky porosity curves in the interior central platform and porosity reduction in the exterior central platform; alternatively, corrosion in the exterior central platform was lower compared to the interior central platform (Figure 30B). Finally, bitumen that was emplaced during a late-burial event acts as cement and reduces porosity (Figure 30C). Bitumen volume increases from the interior to the exterior central platform and margin and obscures correlation using the porosity curves. Some evidence suggests that another corrosion phase is associated with the bitumen emplacement; in the

rim and flank area, observations indicate a corrosion event following the bitumen cement (Collins et. al., 2006). Packstone to mudstone intervals around the cycle boundaries in the central platform wells typically display significantly lower porosity and permeability relative to the overlying grainstone-packstone lithofacies (Figure 31, sequences between Lvis10.5_csb-Lvis9.5_csb in wells T-5246 and T-220). The latter have porosity ranges between 0 and 24% with a mean of 10.5 and 11.0% (Table 2). The packstone to mudstone intervals are generally associated with gamma-ray spikes caused by the presence of volcanic ash and are generally tight with a mean porosity of 3.5%. Within the central platform, the cyclic signature of alternating blocky high-porosity and low-porosity zones associated with gamma-ray spikes is dominant. Lateral and

48 / Kenter et al.

FIGURE 29. Photomicrographs illustrating additional pore types of the central platform area. (A) Enhanced late-burial(?) dissolution joining interparticle pores into larger vugs prior to bitumen lining the pore walls. The primary grain boundaries appear to have been corroded by the dissolution. In addition, following bitumen emplacement, grains were partially dissolved by an additional dissolution event. (B) Bitumen corroding both the crinoid ossicle boundaries as well as those of the syntaxial overgrowth cement. (C) As in (B), but advanced stage of bitumen emplacement corroding grain boundaries and leading to pseudopackstone to wackestone fabric. (D) Recrystallization is a dominant process in the central platform diagenesis. Here, an example where the grainstone fabric was recrystallized to equant and blocky spar following early partial cementation of interparticle porosity and followed by late dissolution generating the moldic porosity. Note that early moldic porosity was occluded by blocky calcite (see micrite envelop), whereas a later dissolution phase generated the open molds that are lined by bitumen. (E, F) Finely recrystallized fabrics associated with intercrystalline and minor moldic porosity, as well as the presence of scattered bitumen, are very common in the central platform. Although the recrystallization phase seems to postdate (corrode and truncate) blocky calcite cementation, the exact timing remains unclear because of the destructive nature of the process. Width of photomicrographs is 4.20 mm (0.16 in.) (C, F) and 1.58 mm (0.06 in.) (A, B, D, E). See text for discussion.

vertical changes in distribution and type of the intermediate grainstone to packstone lithofacies have only a minor effect on the present-day porosity. The interval on Figure 31 in well T-5246 represents highenergy grainstone, and although the interval in well T-220 is lower energy packstone-grainstone facies, no visible difference is observed on the porosity log.

A remarkable deviation in the central platform occurs in the lower part of the Bash1_csb, where mostly well-sorted ooid grainstone is invaded and cemented by bitumen all across the platform. The resulting porosity range is 0–10%, with a mean of 4%, whereas in bitumen-poor ooid grainstone, the porosity ranges between 0 and 21%, with a mean of 12% (Table 2). The fact that bitumen (as much as 3.5%

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 49

FIGURE 30. Hypothetical log correlation scenarios in the central platform, with gamma ray shown on the left and porosity on the right for each well(s). Porosity generation and reduction events that are linked to the major diagenetic phases are indicated for a stratigraphy that includes thin, low-energy lagoonal intervals alternating with volcanic ash beds (yellow and green horizons) and thick intervals of high-energy grainstone (white). Detailed correlation shows that lagoonal intervals and ash beds are not always associated and/or continuous between well locations, and cycles (triangles) may vary slightly in thickness over kilometers distance. (A) Porosity curves with a blocky shape are the result of early-burial processes, including porosity loss around cycle boundaries related to the presence of thin ash beds and corrosion and cementation related to meteoric diagenesis (porosity is blue). Locally thick grainstone intervals have low porosity and show the presence of elevated levels of dispersed ash concentrated in frequent stylolites. (B) Late-burial events leading to corrosion and enhanced blocky porosity curves in interior central platform (red) and porosity reduction in the exterior central platform (yellow); alternatively, corrosion in the exterior central platform was lower compared to the interior central platform. (C) Bitumen, which is emplaced during late burial, acts as cement and reduces porosity. Bitumen volume increases from the interior to the exterior central platform and margin and obscures correlation using the porosity curves.

rock volume) invaded this far into the central platform may be related to the continuous underlying pressure baffle, at the Serp_SSB, which preexisted prior to the bitumen and associated dissolution phase (Figure 32A).

Bitumen occurs within fine matrix and as a porefilling phase, and two spatial trends are observed (see Figure 5, 25): (1) a general increase from exterior to interior central platform wells and (2) elevated bitumen levels near the base of grainy facies between

Figure 5. The lateral facies and thickness continuity of the sequences between Lvis10.5_csb and Lvis9.5_csb are shown between the T-5246, T-220, T-6246, and T-6846 wells. The cyclic and blocky high-low porosity character is visible in nearly bitumen-free wells (T-5246 and T-220) and represents, respectively, alternating grainy facies and lagoonal mud intervals with volcanic ash. Although the interval in T-5246 represents high-energy and essentially mud-free grainstone, and that in T-220 is lower energy and packstone-grainstone facies, no visible difference is observed on the porosity log. See text for discussion.

FIGURE 31. Cross section showing lithofacies, gamma-ray (left), and porosity (right) logs; well locations are shown in Figure 4, and lithofacies legend is shown in

50 / Kenter et al.

tight lagoonal intervals. In addition, some ooid grainstone (right above the Serp_SSB) and crinoid grainstone-packstone intervals (mostly between the Lvis_SSB and Serp_SSB) show increased bitumen content all across the central platform. The late timing of bitumen relative to other diagenetic overprint suggests the introduction through the flanks and into the platform part of the Tengiz buildup, with the largest lateral extent in the most permeable units and residual deposition controlled by gravity, that is, on top of the immediate underlying impermeable cycle boundary (see also Figure 26). Additional intervals with elevated bitumen extend to T-220 but disappear upsection and in the direction of T-5246 (Figures 5, 32B). In the north (well T-5044) and southwest (wells T-6246 and T-6846) parts of the central platform, overall bitumen content in the Lvis_SSB to Bash_SSB grainy facies reduces porosities to a mean of 5.8– 7.6%, which is significantly reduced from the 10.6 – 11.2% recorded in well T-220 (Figure 5). In general, increased bitumen content is commonly confined to the lower part of a cycle. The bitumen-cemented interval exhibits a sharp, lower contact with underlying, well-cemented peloid packstone-wackestone lithofacies, has a gradational upper contact, and appears dark gray to black in core. Although the blocky log character changes to a more irregular and spiky shape with increasing bitumen content, no obvious relationship exists between the amount of porosity reduction and primary lithofacies type. Instead, the presence of the tight peloid packstone-wackestone baffles seems to control the occurrences of bitumencemented interval (see Figures 31, and 33, well T-220 versus well T-6246, no bitumen but lateral facies change). The bitumen effect progressively increases away from the platform center and, at the scale of the plug sample, reduces the porosity and permeability as suggested by the relationship shown in Figure 24. In addition to porosity and permeability reduction by bitumen and volcanic ash layers, the presence of dispersed volcanic ash in high-energy grainstone facies locally also has an effect (Figure 34). Volcanic ash dispersed in the well-sorted skeletal grainstone and coated-grain grainstone-rudstone lithofacies of the Serp1_csb sequence at well T-5447 reduces porosity to nearly zero and generates a nearly flat porosity curve, whereas the same facies interval at well T-220 retains high porosity. Similarly, the tight and laterally continuous interval associated with the Serp_SSB is probably also associated with elevated levels of dispersed volcanic ash (see Figure 5).

FIGURE 32. Cross sections showing lithofacies, gamma-ray (left), and porosity (right) logs; well locations are shown in Figure 4, and lithofacies legend is shown in Figure 5. The lateral facies and thickness continuity of sequences between the Serp_SSB and Lvis13_csb are shown for the T-220 and T-6246 wells in (A) and between the Bash2.5_MFS to Bash4_csb for the T-5246, T-5447, T-220, and T-6246 wells in (B). The lower part of the Bash1_csb, mostly well-sorted ooid grainstone (OG), is invaded and cemented by bitumen all across the platform. Additional intervals with elevated bitumen extend to T-220 but disappear upsection and toward T-5246. See text for discussion.

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 51

Figure 5. The lateral facies and thickness continuity of sequence Lvis9.5_csb to Lvis9_csb are shown for wells T-5246, T-5447, T-220, and T-6246. Toward the southwestern (interpreted as outer platform) margin and away from the central platform to the north (T-5044), overall bitumen in the Lvis_SSB to Bash_SSB section increases (as much as 2 – 4% rock volume bitumen in some parts of T-6246), and generally, elevated levels of bitumen are commonly confined to the lower part of the cycle. See text for discussion.

FIGURE 33. Cross section showing lithofacies, gamma-ray (left), and porosity (right) logs; well locations are shown in Figure 4, and lithofacies legend is shown in

52 / Kenter et al.

IMPLICATIONS FOR RESERVOIR MODELING The stratigraphic framework introduced for the central platform part of the Tengiz buildup in the preceding sections is the basis for the layering scheme that is being used in reservoir models for the field. Fullfield models use the layering portrayed on Figure 4, whereas more detailed models focusing on parts of the platform incorporate more detailed layers as shown in Figures 5 and 25. The lithofacies summarized in Table 2 have been grouped and used to subdivide layers in the reservoir models into regions where porosity and permeability might be expected to vary. The complex diagenesis that occurs in the central platform and the outer platform has overprinted the original (depositional) porosity to an extent, as summarized in Figure 30, that it cannot be ignored when trying to interpret reservoir quality from wire-line logs and correlate it within a reservoir model. Ongoing studies of the detailed analysis of the types and amounts of porosity, as well as pore-filling agents are being used to better understand the log signatures of porosity and improve porosity-permeability transforms. Much work remains in the detailed characterization of the Tengiz reservoir and in correctly portraying the significant geologic characteristics of the Tengiz platform in reservoir models.

SUMMARY AND CONCLUSIONS An integrated (and ongoing) study of core, well logs, and discrete petrophysical measurements provided a first comprehensive assessment of depositional evolution and reservoir quality in the late Visean to Bashkirian interval in the Tengiz central platform. The study resulted in the following observations and tentative conclusions. Depositional cycles (=higher frequency sequences) are up several to tens of meters in thickness and made up of a succession of lithofacies overlying a sharp base with variable evidence for subaerial exposure. At the base of the cycles, tight peloidal mudstone and volcanic ash beds are associated with sequence boundaries and represent flooding events. They are overlain by brachiopod- and crinoid-dominated intervals that represent maximum marine flooding and near the top by skeletal-peloidal grainstone that are interpreted as the highstand-shoaling phase. Nonskeletal grainstones are most common in the Bashkirian. Visean and Serpukhovian cycles are generally easy to correlate laterally, whereas Bashkirian cycles are best explained by the transition to high-frequency icehouse sea level

in Figure 5. The lateral facies and thickness continuity of sequences between Serp_SSB and Lvis13_csb are shown for wells T-5447 and T-220. In addition to bitumen-reducing porosity, dispersed volcanic ash in high-energy grainstone facies locally has an even greater effect on porosity in the central platform. Volcanic ash dispersed in the SPG and CgPR lithofacies in the Serp1_csb sequence at T-5447 reduces porosity to nearly 0%, generating a flat porosity curve, whereas the same facies interval at T-220 retained high porosity. See text for discussion.

FIGURE 34. Cross section showing lithofacies, gamma-ray (left), and porosity (right) logs; well locations are shown in Figure 4, and lithofacies legend is shown

Late Visean to Bashkirian Platform Cyclicity in the Central Tengiz Buildup / 53

fluctuations, which generate incomplete cycles and complex lateral facies changes. The distribution of reservoir quality in the central platform was initially determined by early diagenesis associated with cyclic deposition, which resulted in a stacked system of high-porosity intervals bordered by tight higher order sequence boundaries with reduced dissolution associated with the presence of volcanic ash. This diagenetic phase includes lowporosity, thick (discontinuous and not platformwide) grainstone intervals with high amounts of dispersed volcanic ash. The lateral continuity of tight cycle boundaries probably greatly affected later fluid flow as well as the ultimate distribution of cements, dissolution, and bitumen in the central platform reservoir. Burial diagenetic overprint, an initial corrosion phase followed by a pore-filling phase, and finally, bitumen invasion and associated corrosion, produced an overall porosity reduction toward the exterior central platform. Bitumen content decreases from the exterior to the central platform and is highest near the bases of the cycles. This suggests that the first fill of hydrocarbons migrated through the flanks laterally into the platform cycles. Both corrosion and bitumen emplacement flattened the initial cyclic porosity variations and further complicated porosity and permeability relationships.

ACKNOWLEDGMENTS The authors thank the following individuals, who provided insightful comments or suggestions and, in some cases, data to greatly improve this manuscript. Perhaps most importantly, Michael Clark (Chevron) is thanked for initiating an intense coring and core analysis program that formed the basis for most of our geologic studies and for his strong support of the sequence-stratigraphic framework. Ben Robertson (Chevron) provided important discussion during the continued coring program and supporting geological analyses of outcrop analogs and literature reviews. Numerous coworkers in Chevron, ExxonMobil, and TengizChevroil assisted with our core and log studies and participated in lively discussions on many topics. Giovanna Della Porta (Cardiff University) and Frans van Hoeflaken (independent consultant) are acknowledged for critically reviewing the petrographic and lithofacies descriptions. Steve Jenkins (TengizChevroil) offered numerous comments and observations, which tested the ideas expressed in this study in their practical application for three-dimensional

54 / Kenter et al.

geologic modeling of the Tengiz platform. Comments by AAPG reviewers helped to greatly improve the manuscript. Finally, we thank TengizChevroil and its shareholder companies (Chevron, ExxonMobil, Kazmunaigaz, and BPLukArco) for permission to publish this study.

REFERENCES CITED Brenckle, P. L., and N. V. Milkina, 2001, Foraminiferal timing of carbonate deposition on the Mississippian – early Pennsylvanian Tengiz platform, Kazakhstan: Paleoforams 2001 — International Conference on Paleozoic Benthic Foraminifera, Abstracts, Ankara, August 20 – 24, p. 2001. Choquette, P. W., and L. C. Pray, 1970, Geologic nomenclature and classification of porosity in sedimentary carbonates: AAPG Bulletin, v. 54, p. 207 – 250. Collins, J. F., J. A. M. Kenter, P. M. Harris, G. Kuanysheva, D. J. Fischer, and K. L. Steffen, 2006, Facies and reservoir-quality variations in the late Visean to Bashkirian outer platform, rim, and flank of the Tengiz buildup, Precaspian Basin, Kazakhstan, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir chacracterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 55 – 95. Cook, H. E., W. G. Zempolich, V. G. Zhemchuzhnikov, and J. J. Corboy, 1997, Inside Kazakstan: Cooperative oil and gas research: Geotimes, v. 42, no. 11, p. 16 – 20. Esteban, M., and C. Taberner, 2003, Secondary porosity

development during late burial in carbonate reservoirs as a result of mixing and/or cooling of brines: Journal of Geochemical Exploration, v. 78 – 79, p. 355 – 359. Gradstein, F. M., J. G. Ogg, and A. G. Smith, 2004, A geologic time scale: International Commission on Stratigraphy (ICS) under: www.stratigraphy.org. Lucia, F. J., 1995, Rock-fabric/petrophysical classification of carbonate pore space for reservoir characterization: AAPG Bulletin, v. 79, p. 1275 – 1300. Lucia, F. J., 1999, Carbonate reservoir characterization: New York, Springer-Verlag, 222 p. Moore, C. H., 2001, Carbonate reservoir-porosity evolution and diagenesis in a sequence stratigraphic framework: Developments in Sedimentology, v. 55, p. 444. Ross, C. A., and J. R. P. Ross, 1985, Carboniferous and Early Permian biogeography: Geology, v. 13, p. 27 – 30. Sattler, U., V. Zampetti, W. Schlager, and A. Immenhauser, 2004, Late leaching under deep burial conditions: A case study from Miocene Zhujiang carbonate reservoir: South China Sea: Marine and Petroleum Geology, v. 21, p. 977– 992. Weber, L. J., B. P. Francis, P. M. Harris, and M. Clark, 2003, Stratigraphy, lithofacies, and reservoir distribution, Tengiz field, Kazakhstan in W. M. Ahr, P. M. Harris, W. A. Morgan, and I. D. Somerville, eds., Permo-Carboniferous carbonate platforms and reefs: SEPM Special Publication 78 and AAPG Memoir 83, p. 351 – 394. Zampetti, V., W. Schlager, J. H. Van Konijnenburg, and A. J. Everts, 2003, Architecture and growth history of a Miocene carbonate platform from 3D reflection data: Luconia Province, offshore Sarawak, Malaysia: Marine and Petroleum Geology, v. 21, p. 517 – 534.

2

Collins, J. F., J. A. M. Kenter, P. M. Harris, G. Kuanysheva, D. J. Fischer, and K. L. Steffen, 2006, Facies and reservoir-quality variations in the late Visean to Bashkirian outer platform, rim, and flank of the Tengiz buildup, Precaspian Basin, Kazakhstan, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 55 – 95.

Facies and Reservoir-quality Variations in the Late Visean to Bashkirian Outer Platform, Rim, and Flank of the Tengiz Buildup, Precaspian Basin, Kazakhstan J. F. Collins

G. Kuanysheva

ExxonMobil Development Company, Houston, Texas, U.S.A.

TengizChevroil, Atyrau, Kazakhstan

D. J. Fischer

J. A. M. Kenter1

TengizChevroil, Atyrau, Kazakhstan

Vrije University, Amsterdam, Netherlands

K. L. Steffen P. M. Harris

ExxonMobil Development Company, Houston, Texas, U.S.A.

Chevron Energy Technology Company, San Ramon, California, U.S.A.

ABSTRACT

T

engiz field is an isolated carbonate buildup in the southeastern Precaspian Basin, containing a succession of shallow-water platforms ranging in age from late Famennian to early Bashkirian. Platform backstepping from Tournaisian through late Visean resulted in approximately 800 m (2625 ft) of bathymetric relief above the Famennian platform. This was followed by as much as 2 km (1.2 mi) of Serpukhovian progradation, which formed a depositional wedge around the older platforms referred to as the Serpukhovian rim and flank. Rim and flank facies include lower slope mudstone, volcanic ash, and platformderived skeletal packstone to grainstone interbedded with boundstone breccia; middle-slope poorly bedded to massive boundstone breccia with subtypes based on clast composition, size, and packing; upper-slope in-situ microbial boundstone; and outer-platform to shallow-platform skeletal, coated-grain, and ooid packstone to grainstone. The upper-slope microbial boundstone represents the 1

Present address: Chevron Energy Technology Company, San Ramon, California, U.S.A.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215874M881469

55

56 / Collins et al.

dominant source of clasts in the middle- and lower-slope breccias. Periodic largescale failure of the rim during both Serpukhovian and Bashkirian time resulted in a high degree of lateral facies discontinuity. Solution-enlarged fractures, large vugs, and lost circulation zones produced mainly during late diagenesis form a high-permeability, well-connected reservoir in the rim and flank. This diagenetic overprint is associated with the presence of bitumen and extends upward into overlying Serpukhovian and Bashkirian platform facies and inward into adjacent late Visean platforms, where it has substantially altered reservoir properties that remained after early diagenesis related to cyclic depositional processes.

INTRODUCTION

STRATIGRAPHY

The Tengiz field is located in onshore western Kazakhstan at the northern end of the Caspian Sea’s eastern shoreline, among an archipelago of isolated carbonate buildups that grew on a regional pre-Middle Devonian structural high (Figure 1). Tengiz was discovered in 1979 by the Oil Ministry of the former Soviet Union and, since 1993, has been operated by TengizChevroil (TCO), an in-country joint-venture company. The field is covered by a three-dimensional (3-D) seismic survey encompassing about 1000 km2 (385 mi2), acquired in 1998. As of year-end 2005, more than 115 wells have penetrated the reservoir at Tengiz (Figure 2), which produces an intermediate sulfur oil (478 API). The highest initial production rates are from wells located in the buildup rim and flank areas, in fractured carbonates with low (<6%) matrix porosity. Wells located toward the center encounter higher porosity (as much as 18%) but lower matrix permeability (<10 md) and lower initial rates. The Tengiz buildup formed sometime after the Middle Devonian and grew more or less uninterrupted until the early Bashkirian by aggradation of a series of grain-dominated platforms. The total thickness of carbonate is unknown, but is represented by approximately 1.2 s TWT on 3-D seismic data. The deepest borehole to date bottomed in the middle Famennian at 6032 m (19,790 ft) subsea, more than 2100 m (6890 ft) below the top of the reservoir and more than 700 m (2297 ft) below the top of the Famennian platform. Following platform demise in the Bashkirian, the Tengiz buildup was buried initially by a thin veneer of Moscovian through Artinskian carbonates and volcaniclastics and then by thick Kungurian evaporites (mainly halite) that completely encased the platform. Post-Permian salt diapirism produced a series of salt pillars and clastic-filled withdrawal basins above the platforms, which locally adversely affect 3-D seismic image quality (Figure 3A).

In 2001, a joint study by ExxonMobil, Chevron, and TCO established a stratigraphic framework for the Tengiz platform based on seismic, biostratrigraphic, core, and well-log data that existed up to that time. This study was published (Weber et al., 2003) and stands as the definitive stratigraphic reference for Tengiz. The published framework includes a hierarchy of second-order, third-order, and fourth-order sequences that have been maintained as Tengiz drilling proceeds. This architecture is a locally applicable, fit-forpurpose framework, although it more or less conforms to the second-order eustatic cycles of Ross and Ross (1987). This chapter describes facies and reservoir-quality variations in the Visean A through Bashkirian platform margins at Tengiz (Figure 3B). This interval forms the late highstand of a second-order supersequence set that began with deposition of a series of backstepping platforms in the Tournaisian. By the end of the Visean, the platform area had decreased from 210 km2 (81 mi2) in the late Famennian to about 90 km2 (35 mi2). During the Serpukhovian, the Tengiz platform prograded as much as 2 km (1.2 mi) basinward, filling much of the accommodation space created during backstepping, and forming a microbial boundstone-cored depositional wedge known as the rim and flank around the late Visean and older platforms. This wedge attained a maximum thickness of more than 800 m (2625 ft). Following a brief depositional hiatus on the platform (Brenckle and Milkina, 2003), early Bashkirian platform carbonates accumulated more or less aggradationally above the Serpukhovian. Because this chapter also discusses reservoir connectivity between the Serpukhovian rim and flank, the overlying Bashkirian outer platform region, and the adjacent Visean A platforms, a brief description of how platform facies correlate to rim and flank facies is included here. For a detailed account of the nature

FIGURE 1. North Caspian region and location of the Tengiz field.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 57

58 / Collins et al.

FIGURE 2. Structure contours on top of the Tengiz buildup. Current well locations and the path of the seismic profile in Figure 3 are indicated. Contour interval = 100 m (330 ft).

than 10 m (33 ft) to several tens of meters. They consist of dipping, bedded to massive skeletal packstone to rudstone containing abundant large crinoids and brachiopod fragments, minor colonial coral detritus, and occasional platform-derived breccias. In the central platform, fourth-order sequence boundaries are associated with subaerial exposure and deposition of low-energy or restricted facies containing thin volcanic ash layers (Kenter et al., 2006). In the Serpukhovian outer-platform cycles, these features are reduced or absent at the sequence boundaries.

of the Visean A, Serpukhovian, and Bashkirian central platform regions, the reader is referred to Kenter et al. (2006).

Visean A, Serpukhovian, and Bashkirian Platform Cycles Visean A, Serpukhovian, and Bashkirian fourthorder central platform sequences generally adhere to the model shown in Figure 4. These sequences feature depositional environments ranging from highenergy grainstone shoals to slightly deeper environments characterized by more poorly sorted, openmarine grainstone to packstone, or in some cases, mudstone and faunally restricted, fine-grained carbonate sand. Outer platform cycles vary over the interval but are generally characterized by an increase in algal grains and micritic intraclasts. Because of the unique nature of microbial facies in the rim and flank, greater attention is given to Serpukhovian outer-platform cyclicity. These cycles are commonly poorly expressed and range in thickness from less

DEPOSITIONAL ENVIRONMENTS General Features of the Tengiz Rim and Flank The terms ‘‘platform,’’ ‘‘rim,’’ and ‘‘flank’’ were originally derived from seismic features and are now used for drilling purposes to denote geographic regions in the field. The platform region comprises boreholes that penetrate the stack of backstepping platforms, whereas boreholes that encounter any part of the adjacent progradational wedge are described as rim and flank wells. As mentioned earlier, the wedge arises as a result of Serpukhovian progradation, but includes rocks of Serpukhovian and Bashkirian age (see Figure 3). The rim area refers to the inner, structurally elevated rim crest or raised rim, whereas the flank refers to the outer sloping surface. The raised rim is interpreted to have formed mainly by differential compaction caused by a mechanical contrast between grainy platform facies and more rigid upper-slope microbial boundstone facies in the rim and has resulted in as much as 50 ms of seismic relief and as much as 200 –250 m (656 – 820 ft) of

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 59

FIGURE 3. Tengiz stratigraphic framework and the Serpukhovian rim and flank region. The seismic line (A) shows the east-west asymmetry in the shape of the wedge and in the width of the structurally elevated portion (raised rim). The seismic profile also shows the absence of Moscovian – Artinskian volcanic sediments on the upper flank of the buildup, placing an anhydrite layer at the base of the salt in direct contact with the reservoir, as well as the source rock interval associated with subsalt facies in the same interval in surrounding lows. These are two potentially important factors in rim and flank diagenesis. The schematic framework (B) is derived from Weber et al. (2003), with details from this study added in the rim and flank area. In particular, the diagram shows the rim and flank to consist of an early debris apron that accumulated around the base of the buildup, followed by progradation of a boundstone-cored upper wedge (black dashed line).

60 / Collins et al.

FIGURE 4. Idealized Tengiz fourth-order platform sequence, showing textural and compositional variations associated with platform bathymetry for the Bashkirian, Serpukhovian, and late Visean. Outer-platform, rim, and flank facies illustrated apply particularly to the Serpukhovian. For other intervals, outer-platform facies, in general, are characterized by an increase in algal grains, micritic intraclasts, and coarse crinoid-brachiopod debris. Sea level lowstands are associated with exposure of shallow-platform shoals, contemporaneous formation of an outer margin complex of algal-intraclast grainstones and possible coral-algal patch reefs, and local development of restricted conditions on the platform. Partly restricted-marine conditions that persisted during the early transgression were dominated by brachiopods, including the formation of outer-platform brachiopod mud mounds. Widths of facies belts are not shown to scale.

structural relief1 (see Figure 2). Evidence for platform compaction includes: 1) a lack of volumetrically significant deep lagoonal facies in the central areas of the Visean A, Serpukhovian, and Bashkirian platforms 2) upper Serpukhovian and Bashkirian platform cycles in the raised rim that contain shallow facies similar to age-equivalent cycles in the central platform 3) significant grain-to-grain compaction observed in thin sections from the platform facies 4) observation of high early cement volumes in the rim and flank facies, indicating early development of mechanical rigidity 5) a lateral facies progression in Serpukhovian cycles that deepens basinward from the platform

1 Weber et al. (2003) estimated that approximately two-thirds of the raised rim relief was caused by differential compaction, and one-third was caused by a bioconstructed margin associated with upper Bashkirian cycles.

into the rim, based on comparisons between Tengiz and other microbial boundstone-cored platforms (Kenter et al., 2005). The break in slope separating the raised rim and flank areas does not necessarily indicate the limit of Serpukhovian progradation. In some regions, this break appears to be controlled by faulting or mass wasting. In addition, differential compaction of a basinward-thickening wedge of middle- to lowerslope facies may be responsible for the slight basinward dip of the raised rim observed in some areas.

Rim and Flank Depositional Environments The rim and flank contain carbonate facies deposited in water depths ranging from a few meters to several hundred meters. Facies descriptions come mainly from cores2 and from Formation MicroImager (FMI) 2 Approximately 1200 m [3937 ft] of core from 15 wells available as of year-end 2005 were examined for this study.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 61

logs calibrated to cores. Facies interpretation from FMI logs was hampered by use of oil-based mud during drilling, which tended to reduce conductivity contrasts between porous and nonporous rock, and by frequent hole washouts and zones of lost circulation that adversely affected image quality. Rim and flank facies are associated with a depositional profile consisting of the following environments, in the order of increasing paleowater depth: 1) cyclic shallow-platform deposits 2) outer-platform skeletal-intraclast rudstone, grainstone, packstone, and minor boundstone 3) upper-slope microbial boundstone 4) middle slope welded to mosaic boundstone breccia 5) lower-slope detrital boundstone breccia and grainy platform-derived deposits 6) lower-slope and debris apron boundstone breccias and bedded, fine-grained periplatform deposits. These are partly based on environments and facies recognized in outcrops of the Asturias platform margin in the Cantabrian region of northern Spain and from the Capitan margin in west Texas and New Mexico (Figure 5). The Tengiz outer platform, slope, and debris apron facies are described in the ensuing paragraphs and are also summarized in Appendix 1. The cyclic, shallow-platform facies are described in Kenter et al. (2006).

Outer-platform Facies The Asturias and Capitan margins were both formed by deeper water microbially influenced boundstone margins (Kirkland et al., 1998; Della Porta et al., 2003) and support the interpretation of considerable original outer-platform relief above the Tengiz upper slope during the Serpukhovian. The Asturias margin had outer-platform relief of between 20 and 50 m (66 and 164 ft) during times of progradation (Bahamonde et al., 2000; Kenter et al., 2005). Estimates of the relief across the Capitan outer platform margin show variations through time from 10 m to more than 60 m (33 to more than 197 ft) (Kerans and Harris, 1993; Tinker, 1998). Total paleobathymetric relief from the shallow Tengiz platform to the start of the upper-slope microbial boundstone at the edge of the outer platform was at least tens of meters (Weber et al., 2003) and probably higher. Apparent thickening in the outer platform of the uppermost Serpukhovian fourth-order cycles, plus the limited presence of subaerial exposure at the tops of outer-

platform cycles, indicates that relief exceeded at least some of the ambient low-amplitude eustatic variations, estimated to be between 25 and 50 m (82 and 164 ft) (Ross and Ross, 1987, 1988). Whereas the succession of facies comprising cycles in the central platform area is well documented in cores (see Kenter et al., 2006), the descriptions of Serpukhovian outer-platform cycles contained here are generalizations based on a limited number of complete cycles present in the current suite of cores. The basal parts of some outer-platform cycles were observed to contain breccias composed of colonial coral detritus or, in other cases, rounded, platformderived grainstone clasts (Figure 6A, B). The breccias contain either microbial matrix textures and cements or a matrix of coarse skeletal debris. They may represent lowstand deposits, indicating updip platform exposure, or development of outer-margin coral patch reefs. In other sequences, brachiopod mud mounds a few meters thick (Figure 6C) are encountered near the base. The middle and upper parts of cycles are characterized by wackestones, packstones, and rudstones containing large crinoids, brachiopod shell fragments, and solitary rugose corals (Figure 6D), indicating an increase in water depth. This facies is associated with dipping beds (Figure 6E), indicating deposition on a paleoslope, and may also exhibit microbial matrix textures. Cycle tops tend to be characterized by a reduction in crinoid-brachiopodcoral grains and an increase in coated-grain or algalintraclast grainstone, resembling some shallowplatform grainstones. Subaerial exposure effects like those described from the central platform (Kenter et al., 2006) are commonly absent, but large dissolution voids filled with multilayered geopetal sediment and multiple generations of carbonate cement are found, as are brecciated grainstone intervals with evidence of clast rotation and mixing. Isolated, sediment-filled dissolution voids are also scattered throughout the section, but appear to increase in frequency at or near the tops of cycles. These may be associated with longer term exposure at supersequence boundaries (Serp_SSB and Bash_SSB), or they may be unrelated dissolution features whose distribution was influenced by stratigraphic permeability variations.

Upper- to Middle-slope Facies The upper slope consists mainly of in-situ microbial boundstone and boundstone breccia. Similar facies are recognized from the Asturias margin (Bahamonde et al., 2000; Della Porta et al., 2003) as mentioned

62 / Collins et al.

FIGURE 5. Outcrop analogs with facies and depositional relationships similar to the Tengiz outer platform, rim, and flank. Like Tengiz, the platform margins illustrated are characterized by dipping, basinward-deepening platform facies that pass downward into upper-slope, microbially influenced boundstones and foreslope facies with reef-derived breccias. Based on these profiles, the structurally raised rim visible on Tengiz seismic data (see Figure 3) is largely an artifact of postdepositional differential compaction (see text for discussion). Microbial boundstone types A, B, and C noted in the Asturias example closely match boundstone types recognized in Tengiz cores.

previously, from Triassic platforms in Italy (Blendinger, 2001) and from Famennian platforms in the Canning basin (Playford, 1984; Kerans et al., 1986; Stephens and Sumner, 2003). Tengiz cores and FMI logs suggest that the in-situ boundstone interval is perhaps 150– 200 m (492–656 ft) thick. In-situ microbial boundstones (Figure 7) are light colored in core, with textures that include relatively featureless

micritic to peloidal fabrics (type A) or irregular laminar fabrics and amalgamated semiconcentric laminar masses (type B). A third boundstone type (type C), characterized by massive to peloidal fabric with common skeletal grains (bryozoans, algae, forams, ostracods, pelecypods, crinoids, corals, and sponges), represents the transition between upper-slope and outer-platform facies. Small- to medium-size fenestrae filled with banded

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 63

FIGURE 6. Serpukhovian outer-platform facies. Breccias composed of platform-derived clasts (A) or colonial coral fragments (B) occur toward cycle bases. The breccia in (A) contains cement and clotted fabric of possible microbial origin, whereas the matrix in (B) consists of coarse crinoidal rudstone. Brachiopod mud-mounds (C) as much as a few meters thick also sometimes occur near cycle bases. The middle and upper parts of typical outer-platform cycles contain abundant packstone and wackestone with large crinoids, brachiopods, and solitary corals (D), as well as intervals with dipping beds (E), indicating deposition on a paleoslope.

cements are commonly abundant, and larger cementlined cavities filled with internal sediment or skeletal accumulations are observed. Boundaries between microbial masses are sometimes formed by networks of elongate cavities that are also frequently cement filled. The boundstone interval at Tengiz could be extended down to more than 300 m (984 ft) if certain types of breccias are included. The upper slope interval in the Asturias margin is comparable in thickness and includes both in-situ microbial boundstone, and nearly in-situ boundstone breccia generated by rapid marine cementation of incipient boundstone failure (Bahamonde et al., 2000; Della Porta et al., 2003). To date, similar breccias held together by thick marine cements have not been observed in abundance at Tengiz, although core coverage is still limited. However, breccias consisting of irregularly shaped to

subangular boundstone clasts with stylolitic (welded breccia) or fracture (mosaic breccia) contacts are found in the Tengiz middle to upper slope (Figure 8) and may be analogous to the Asturias intact breccias. The matrix consists mainly of several generations of calcite cement, secondary microbial encrustations, microbial cement, and minor amounts of platform-derived skeletal sediment, thin-bedded marly volcanics, or argillaceous carbonate. Very little open primary pore space was observed in these breccias. Formation MicroImager logs over upper-slope microbial boundstone and breccia intervals show little evidence of cyclicity, suggesting that sea level variations that affected the platform had little or no effect on production of microbial boundstone. As a result of deposition below the range of sea level fluctuations and productivity across a wide range of water depths,

64 / Collins et al.

FIGURE 7. Serpukhovian microbial boundstone facies, shown in order of increasing paleodepth. (A) Type C boundstone (skeletal-micritic) with corals and crinoids, and numerous cement-filled, irregular-shaped fenestrae. The matrix is skeletal rich with micritic to peloidal microbial texture. (B) Type C boundstone containing numerous bryozoans (arrows) and thick-banded cements of probable marine origin. (C) Type B boundstone (laminar cemented), characterized by irregular to rounded, concentrically laminated microbial masses. These boundstones contain as much as 75% cement. (D) Type A boundstone (peloidal-micritic), featuring a dominantly peloidal to micritic matrix, scarce fauna, and numerous fenestrae or tubular voids.

including subphotic depths, the boundstones were apparently able to prograde continuously seaward during all stages of sea level rise and fall.

Middle- and Lower-slope Facies The middle and lower slope consists of mainly detrital facies (Figure 9), including sedimentary boundstone breccias; coarse skeletal rudstones; fine-grained breccias; massive or poorly bedded allochthonous grainstones; and thin-bedded to laminated grainstones, packstones, wackestones, mudstones, argillaceous carbonates, volcanic ashes, and cherts. Clast size and packing varies in the breccias, with textures ranging from matrix-supported floatbreccia to welded breccia with stylolite-bounded clasts (Figure 9D – F). Clast sizes range from less than 1 cm (0.4 in.) to several meters.

The matrix contains variable amounts of bioclastic debris, fine-grained carbonate, argillaceous mudstone, and volcanic ash. The most visible skeletal components are large crinoids and brachiopod shell fragments similar to outer-platform facies. Relationships between clast size, clast packing, matrix grain size, and matrix composition have not been fully evaluated, but they appear to be poorly correlated. Clasts are dominantly upper-slope microbial boundstone fragments, although platform-derived clasts and reworked slope clasts are increasingly observed in breccias near the base of the section. Some detrital breccias are massive and unbedded over very thick intervals, whereas others are layered and contain intervals of bedded, grainy periplatform facies. Formation MicroImager logs again do not indicate an ordered cyclicity that can be related to the platform sequences.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 65

FIGURE 8. Upper- to middle-slope diagenetic and sedimentary breccias. (A) Fractured microbial boundstone healed by banded, probably marine cements. (B) Partly cemented mosaic breccia containing earlier cemented voids dissected by later fractures and voids containing bitumen and equant calcite spar (arrows). Larger open voids are mostly a product of leached cement fills in the secondary breccia. Note also that the early, cemented fracture in (A) is also intersected by a later, bitumen-filled fracture. (C) Mosaic to welded breccia containing microbial boundstone clasts with stylolitic contacts. (D) Sedimentary floatbreccia with irregularly shaped boundstone clasts, and crinoid-rich, skeletal-lithoclast rudstone matrix.

DEPOSITIONAL HISTORY OF THE TENGIZ RIM AND FLANK The additional facies control provided by FMI logs (Figure 10) proved helpful in interpreting the rim and flank regions. Attempts to extrapolate fourthorder platform correlations into the rim and flank on ordinary wire-line logs proved difficult because of the presence of apparently nonstratiform porosity variations and diagenetic effects manifested as anomalous radiogenic responses on gamma-ray logs. Additionally, rim and flank stratigraphy is complicated by large-scale failure and mass wasting because of instability, both during and after deposition. Seismic and biostratigraphic control indicates that the interval contains breccias of Serpukhovian age as well as Bashkirianaged breccias with Serpukhovian clasts. Boundstone

breccia is the dominant facies present in rim and flank cores recovered to date; therefore, its distribution is a critical element of reservoir prediction. This requires a deeper understanding of depositional and failure processes at the Tengiz margin. The analogous Asturias and Capitan platforms are also associated with middle- and lower-slope breccia deposits. The Asturias breccias occupy a basinwardthickening middle- to lower-slope wedge (see Figure 5), whereas the Capitan margin is associated with both middle- and lower-slope breccias and basinal debris flows out in front of the shelf (Melim and Scholle, 1995). These dominantly progradational margins have maximum upper-slope angles of 30 – 358 (Tinker, 1998; Della Porta et al., 2003; Kenter et al., 2005). Slope angles for the Tengiz rim and flank are known from structural and seismic profiles of the Bashkirian

66 / Collins et al.

FIGURE 9. Middle to lower slope and debris apron facies. Middle- to lower-slope facies include coarse skeletal (crinoidbrachiopod) rudstone (A), skeletal wackestone to packstone with dipping beds (B), and thin-bedded volcanics, mudstone, packstone, and fine grainstone (C). Debris apron facies include coarse breccia with meter-size boundstone clasts (D), packbreccia with platform-derived, crinoid-rich matrix (E), and float- to packbreccia interbedded with allochthonous grainstone and packstone (F). Note the relatively low dip angles in (F).

surface and from dipmeter data. In the steeper sloped upper flank, the Bashkirian is thin, consists of mostly fine-grained deeper water facies, and, thus, forms a prominent seismic reflector that closely approximates the Serpukhovian profile. Measured angles vary between 20 and 258 for apparent accretionary slopes, whereas angles approaching 308 are measured in areas of obvious detachment or failure. These measurements represent angles determined from a compacted body of rock and do not necessarily reflect original depositional dips. Because of the chaotic nature of boundstone and breccia facies at Tengiz, FMI dip measurements commonly produce inconsistent results, but some intervals indicate that preserved bedding dips of 25 –308 in the upper slope may be typical, and angles of 35 – 458 occur locally. Directionality is variable, however, and dips do not always point directly away from the platform. Multiple azimuth changes have been recorded vertically in a single well or in comparing two nearby wells. Formation MircoImager logs also suggest that facies distribution is more complex than expected for simple progradation. Whereas the Asturias and Capitan margin profiles have regular shapes, the Tengiz flank has an irregular shape on

seismic profiles, even considering geometries created by depositional or compactional drape over underlying backstepping platforms. This shape varies systematically around the perimeter of the buildup and, combined with facies obtained from core and FMI logs, allows the rim to be divided into subregions governed by different depositional processes.

Allochthonous and Accretionary Rim Sectors The rim and flank area consists of two distinct regions, the allochthonous and accretionary sectors (Figure 11). The allochthonous sector refers to the southwestern and northwestern parts of the rim. It is characterized by a lower, distally thickened apron that onlaps the Tournaisian and Visean platforms, which itself appears to be downlapped by a narrow progradational wedge (Figure 12A). The raised rim is reduced or, in some places, completely absent along this trend (Figure 12B). The accretionary sector refers to the northern, northeastern, and southeastern regions of the rim, where it is characterized by a conspicuous raised rim crest and a considerably wider progradational wedge (Figure 13A).

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 67

FIGURE 10. Formation MicroImager log to core facies calibrations. FMI facies, indicated by different colors in the depth columns (depths are in meters), include boundstone-breccia (BBr), boundstone-horizontal fabric (Bhz), breccia, rudstone, grainstone, and laminated fine-grained (lam-FG). BBr facies, characterized by a chaotic image lacking horizontal layering (C – E), includes both boundstones and breccias from core, which, except occasionally, were not distinguishable on FMI logs. Bhz facies corresponds to poorly bedded outer platform skeletal rudstone and type C microbial boundstone (F). FMI rudstone is characterized by nodular texture and some horizontal layering (A, D) and includes lithoclast-skeletal rudstones, fine-grained breccias, and poorly bedded skeletal packstone, wackestone, and floatstone from core. Formation MicroImager grainstones, identified by a low-contrast image with faint layering (B), correspond to generally tightly cemented allochthonous grainstones. Lam-FG facies, marked by frequent high-contrast layers (A), are equivalent to thin-bedded, deep-water volcanics, mudstones, and grainstones in core. In some intervals, image quality is adversely affected by bitumen, which produces a snowy texture overprinted on the facies (C).

In the allochthonous sector, the lower apron consists of bedded to massive boundstone breccia interlayered with bedded platform-derived sediments, suggesting relatively continuous failure of a boundstone margin. In the northern half of the accretionary sector, the flank slope is typically irregular. This is partly caused by a series of local slump scars and ap-

parently bottomless faults, but also to a prominent midslope bulge. The lateral continuity of this latter feature (see Figure 11) could be accounted for by slight downslope movement of upper-slope material or alternately by downlap of a late-stage progradational wedge onto a protruding debris apron (Figure 13B), analogous to the allochthonous sector. Seismic data

68 / Collins et al.

FIGURE 11. Dip magnitude on the top of the reservoir (Bashkirian surface) measured from 3-D seismic data, showing the extent of the accretionary and allochthonous sectors. The map shows a thin band of elevated dip (dashed white line) that is directed toward the platform interior caused by differential compaction between platform facies and upper-slope microbial boundstones and indicates the position of the platform margin prior to progradation. The zone of low dip between this band and the slope break is the raised rim and indicates the limit of progradation (possibly modified by failure processes). The boundary between the accretionary and allochthonous sectors occurs where the band of compactional dip intersects the slope break (white arrows). In the accretionary sector, the slope break is marked by several local scarps indicative of localized failure (yellow lines). In the allochthonous sector, the raised rim is thin or absent, indicating either nearly complete rim failure or a poorly developed progradational wedge. The locations of seismic profiles in Figures 12–15 are also shown. suggest the existence of an early debris apron in the accretionary sector in another way as well. The northern and southern margins of the Tengiz platform represent transitional regions between the allochthonous and accretionary sectors. Both areas are characterized by an upper wedge that gradually widens eastward while maintaining a downlapping relationship to laterally continuous debris apron deposits (Figure 14). The accretionary and allochthonous sectors demonstrate internal and comparative differences with respect to the total volume of rock in the rim and flank, as well as the relative proportions in the upper wedge and lower apron. This asymmetry could result from both extrinsic and intrinsic controls. For example, one intrinsic control might be the pre-Serpukhovian profile created by the stacking pattern of Tournaisian and Visean platforms. In areas of significant backstepping, a bench or gentle slope might have provided stability for progradation (see Figure 13A), whereas in areas where platform stacking is more vertical, instability and rim failure might dominate. Alternatively,

primary microbial production may have varied around the platform. In the accretionary sector, elevated productivity could have resulted in faster accumulation of a debris apron and, thus, earlier initiation of upper wedge progradation. In a lower productivity allochthonous sector, the debris apron required a longer time to accumulate and was perhaps still in progress at the end of the Serpukhovian, accounting for the more limited upper wedge development. Tectonic activity is a possible external mechanism that could also account for rim asymmetry. Rim failure could have been restricted to, or enhanced in the allochthonous sector because of alignment with regional stress regimes or as a result of directly facing pressure waves from the tectonically active southern margin of the Precaspian Basin (see Figure 1). Whether the apron is mostly an early or a late feature is therefore still somewhat uncertain because of these possibilities. The significant role of rim failure and the timing uncertainty precluded erecting a stratigraphic framework in the rim and flank that can be specifically

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 69

FIGURE 12. Seismic profiles from the allochthonous rim sector. In (A), a slightly raised rim indicates a narrow progradational wedge of upperslope microbial boundstone that formed after deposition of an extensive debris apron. The figure in (B) shows a region where the raised rim is absent, indicating failed progradation. Formation MicroImager logs from T-7040 indicate the upper wedge deposits at this location most likely consist of coarse breccia. See Figure 11 for profile locations.

tied to Serpukhovian and Bashkirian fourth-order platform cyclicity; however, core control, FMI facies, biostratigraphic data, and, to a lesser extent, wire-line logs were still used to separately constrain the distribution of facies in the lower apron and the upper wedge as a result of their relative spatial separation.

Upper Wedge Facies Distribution From previous discussions, the raised rim effect is interpreted to indicate the presence of mechanically rigid upper-slope microbial boundstone facies in the upper wedge. In places along the allochthonous sector, the upper wedge is partly or completely detached from the platform, and the raised rim effect is absent. In some cases, the wedge appears to have detached more or less intact, whereas in others, it

forms a mound-shaped pile that is almost continuous with the distal apron (Figure 15). To date, the upper wedge is penetrated by 14 wells with core slabs and/or FMI images (T-463, T4346, T-4635, T-70, T-7040, T-5056, T-5857, T-7252, T-6457,T-4556,T-5059, T-5454, T-7450, and T-7052). These wells indicate facies consistent with a prograding boundstonecored wedge in areas of accretion in both sectors and facies dominated by breccia in areas of detachment in the allochthonous sector (Figure 16). Specific cored examples include cyclic shallow platform overlying deeper water outer-platform facies (T-4556 and T-7052); upper-slope in-situ boundstone facies (T-7450 and T-5056); and upper-slope boundstone and welded-to-mosaic breccia from a detached but relatively intact upper wedge (T-4635, Figure 15A). Other cored examples include middle-slope breccias (T-5056, T5857, and T-7252), and middle- to lowerslope clast- and matrix-supported boundstone breccia interbedded with platform-derived crinoid-brachiopod rudstone to packstone (T-463). The Bashkirian in the upper wedge consists of shallow-platform facies in the raised rim area and, on the basis of mainly short cores and core chips from older wells, deeper water facies composed of laminated to thin-bedded mudstones, packstones, fine grainstones, and volcanics in the flank.

70 / Collins et al.

FIGURE 13. Seismic profiles from the accretionary rim sector. The flank profile in (A) is smooth and lacks highamplitude internal reflections. Here, a bench formed by a backstepping of the Visean A and B platforms may have provided stability for early upper wedge progradation. The flank profile in (B) shows a distinct midslope bulge. Formation MicroImager data from T-5059 and T-5660 demonstrate that the upper part of the Serpukhovian interval in the bulge contains a prograding boundstone sequence, whereas the lower part contains bedded boundstone breccias (see Figure 17). The change occurs near a zone of high-amplitude, subparallel reflectors on the seismic line, which have been interpreted to indicate areas in the accretionary sector where deposition of early debris aprons filled in some of the accommodation space as a prerequisite to upper wedge progradation. See Figure 11 for profile locations.

Lower Apron Facies Distribution The lower apron is penetrated by four wells in the allochthonous sector with cores and/or FMI data (T-16, T-6337, T-3938, and T-3948). Assuming that an early apron is also present in the accretionary sector, five other wells potentially have representative core and/or FMI data (T-47, T-5963, T-5059, T-5660, and T-6261). In particular, the FMI facies from T-5660 and T-5059, combined with three separate cored intervals from nearby T-5056 in the raised rim, indicate a possible distinction between apron breccias and upper wedge facies in the accretionary sector (Figure 17). Apron facies are best known from cores in the allochthonous sector, which reveal that, in general, the apron consists of mixtures of upper-slope-derived

boundstone breccia and platform-derived grainy facies, but which also indicate significant lateral variations in bed thickness, clast size, and amount of platformderived grainy material versus breccia. In two separate cored intervals from T-6337, for example, both clast- and matrix-supported breccias are frequently interbedded with fine-grained periplatform mudstone, packstone, and grainstone. Overall clast size is larger compared to the T-463 core from the upper wedge, but the platform-derived fraction is finer grained. By contrast, a long, cored interval at the T-3948 core consists almost entirely of poorly bedded boundstone breccia made up of very large clasts, with very few interbeds of fine-grained platform-derived material. The T-3938 FMI log suggests that the apron interval there also consists of massive boundstone breccia with few interbeds of platform-derived facies. The lateral extent of this massive, coarser apron breccia facies appears to coincide

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 71

FIGURE 14. Seismic profiles from the northern (A) and southern (B) rim and flank regions. These areas are transitional between the allochthonous and accretionary sectors and show both a lower debris apron and significant upper wedge progradation, suggesting that variations in wedge development did not necessarily occur at the expense of the apron. See Figure 11 for profile locations.

with chaotic seismic character and, therefore, may be regionally mappable within limits imposed by image quality (see Figure 16). Apron breccias in the allochthonous sector contain increasing numbers of transported slope and platform clasts toward the base of the interval. Cores from T-3938 and T-3948 demonstrate that somewhat finer grained breccias with abundant platform and slope clasts are associated with a resistivity decrease on petrophysical logs. This low-resistivity zone is not observed on logs from the accretionary sector, but because this could be caused by a change in porosity or other diagenetic effect, its absence does not necessarily indicate that similar breccias are not pre-

sent there. Cores across the base of rim and flank breccias have been recovered at T-6337, T-3938, T-3948, and T-7252. Formation MicroImager logs from these and other wells demonstrate that the breccias were deposited abruptly on top of a continuous interval containing mainly thin, dipping beds of deeper water facies. Biostratigraphic analyses from the cores indicate that in both sectors, the onset of breccia deposition occurred close to the Visean–Serpukhovian boundary. Biostratigraphic data also demonstrate that Bashkirianaged sediments containing Serpukhovian boundstone clasts occupy as much as 300 m (984 ft) at the top of the apron breccias in the allochthonous sector, indicating that significant postdepositional rim failure occurred in that region. The Bashkirian interval appears to be associated with a distinctive pattern on gamma-ray and resistivity logs (Figure 18), and well correlations suggest that Bashkirian breccias are variable in thickness but laterally extensive (see Figure 16). The apparent absence of equally thick Bashkirian deposits or Bashkirian-aged breccias in the accretionary sector is an important distinction that perhaps indicates that relative depositional stability was achieved in this region prior to the end of the Serpukhovian. Cross sections illustrating the depositional concepts, core and FMI facies control, and typical porosity and

72 / Collins et al.

FIGURE 15. Seismic profiles illustrating different modes of upper wedge failure. The figure in (A) shows a completely detached but largely intact upper wedge. Core from T-4635 contains mainly type C boundstones and breccias, indicating original deposition in shallower water. The figure in (B) shows a narrow upper wedge with a downslope bulge, indicating possible partial failure. The debris apron in this area consists of poorly bedded boundstone breccia, as indicated on FMI logs from T-3938. Biostratigraphy from drill cuttings revealed a thick Bashkirian-aged section at the top of the apron breccias. The figure in (C) shows local fault detachment in the upper wedge. In addition, chaotic seismic character in the debris apron is separated from the upper wedge slope by a zone of relatively continuous dipping reflections from a distinct lower slope facies. T-3948 cores reveal that the apron at this location consists of massive boundstone breccia with very large clasts. See Figure 11 for profile locations.

DIAGENESIS

gamma-ray expression of the lower apron and upper wedge in the Tengiz allochthonous and accretionary rim sectors are provided in Figures 18 and 19.

As a long-lived shallow platform, Tengiz was expected to have a complex distribution of pore types because of its multifaceted stratigraphic hierarchy and associated facies variations, repeated exposures to marine, meteoric, and shallow burial diagenesis, and, because of the age range involved, the influences of both greenhouse and icehouse global themes. Core and thin-section observations, combined with preliminary geochemical analyses, provide insights into processes of porosity creation, destruction, and preservation in outer platform, rim, and flank reservoirs and also illuminate relationships that determine reservoirquality distribution in the rim and flank. These relationships suggest that present-day reservoir-quality variations are strongly influenced by a late diagenetic overprint involving fracturing, dissolution, and bitumen cementation.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 73

FIGURE 16. Facies distribution in the rim and flank. The Visean A outline indicates the platform configuration just prior to Serpukhovian progradation. The raised rim area (upper wedge) represents the preserved extent of Serpukhovian platform progradation. The outer limit of the upper-slope boundstone was determined by an assumed vertical thickness of about 200 m (656 ft) below the slope break. The lower apron contains Serpukhovian and Bashkirian coarse breccias, mapped from seismic character and well-log correlations. Serpukhovian breccias occur on both sides of the platform, whereas Bashkirian apron deposits are generally restricted to the allochthonous sector, indicating depositional instability in that area throughout the Serpukhovian and Bashkirian. Facies control is indicated for rim and flank wells only.

Early Dissolution, Cementation, and Bitumen Formation Early pore types consist of primary and early diagenetic matrix pores and larger voids from at least one significant dissolution event. Existing cores suggest that this early dissolution was more prevalent in the outer-platform and upper-slope facies compared to the middle and lower slope. Early dissolution may have occurred during platform margin karst events associated with major sequence boundaries or as a result of Kohout-type convection driven by large fluiddensity contrasts between platform-top groundwaters and seawater (Kohout et al., 1977; Saller, 1984; Sanford et al., 1998). Such contrasts, controlled by temperature and salinity, could be especially impor-

tant for platforms like Tengiz with considerable vertical height. Early voids commonly contain several cement generations. These cements, starting with inclusion-rich calcites of probable marine origin and culminating with mediumto coarse-grained clear equant calcite of burial origin, form a sequence that destroyed much of the early porosity (Figure 20). The residual pore space commonly contains bitumen, either as a final fill phase or as beads and partial linings on the equant calcite. The early cement sequence is therefore referred to collectively as the prebitumen sequence. Petrographic relationships between bitumen and equant calcite (Figure 20C) suggest that they were emplaced approximately contemporaneously. Equant calcite and bitumen occur in some dissolution features as the only phases, indicating a second fracturing and dissolution event prior to bitumen emplacement. Fluid-inclusion homogenization temperatures indicate that the equant calcite was precipitated at temperatures in the 80–958C

74 / Collins et al.

FIGURE 17. Selected core and FMI images from T-5660 (top), showing facies representing the upper wedge sequence overlying a possible debris apron sequence in the accretionary sector. The upper wedge sequence consists of distal bedded rudstone and fine breccia toward the base (E, F), which are overlain by successively coarser breccias (B – D). Above the cored interval, the section consists of massive boundstone or breccia (A). The apron (G – J) consists of interbedded lam-FG, rudstone, and breccia FMI facies (the latter presumably containing microbial boundstone clasts). For definitions of FMI facies, see Figure 10. The cross section (bottom, see Figure 16 for well locations) shows that the inferred apron section in T-5660 is associated with reduced radioactivity on the spectral gamma-ray (SGR) log compared to the wedge interval and may be correlated to nearby wells on this basis. Other log curves shown include computed gamma-ray (CGR, calculated with uranium subtracted) and deep resistivity (DRES).

FIGURE 18. Facies and stratigraphic relationships in the allochthonous rim sector. Wells are shown in approximate order of increasing distance from the edge of the platform (see Figure 16 for well locations). The thick, allochthonous Bashkirian section at T-6337 is supported by biostratigraphic analysis of the cored interval. Comparison of spectral gamma-ray (SGR) and computed gamma-ray (CGR, calculated with uranium subtracted) logs indicates uranium enrichment throughout the rim and flank interval. Note the low-resistivity zone on the deep resistivity (DRES) curves in T-3938 and T-3948 at the base of the apron. Also note the similarities in the FMI facies and gamma-ray logs in T-6337 and the interpreted apron sections in T-5660 and T-5059 from the accretionary sector (Figures 17, 19). For FMI facies definitions, see Figure 10.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 75

for well locations). Raised rim wells contain Serpukhovian middle-slope breccia and upper-slope boundstone overlain by upper Serpukhovian to Bashkirian outerand shallow-platform facies. Wells closer to the platform (T-4556 and T-7450) are associated with reduced uranium enrichment, as shown by comparing the spectral gamma ray (SGR) with the computed gamma ray (CGR, calculated with uranium subtracted). Uranium enrichment also tends to be higher in upper wedge boundstone breccia and associated facies in flank wells, above the apron facies interpreted in T-5059 and T-5963. FMI facies and deep-resistivity (DRES) logs are shown for some wells. For FMI facies definitions, see Figure 10.

FIGURE 19. Stratigraphic and facies relationships in the accretionary rim sector. Wells are shown in order of increasing distance from the platform (see Figure 16

76 / Collins et al.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 77

range (Appendix 2); thus, prebitumen cementation was completed shortly before hydrothermal temperatures were achieved in the reservoir. Equant calcite is a volumetrically important cement in the rim and flank and is referred to as 908 calcite on the basis of the fluid-inclusion data.

Fractures and Later Dissolution (Corrosion) Larger fractures observed in core can be separated into sets according to their fill phases. Earlier fractures, like early dissolution voids, are extensively filled with prebitumen cement phases. Later fractures contain only bitumen, or equant calcite accompanied by bitumen, and are filled to partly open. Systematic frac-

tures that are mainly open and lack cement are present, representing an even later generation following bitumen deposition, and at least one set of fractures appears to be aligned with the present-day stress regime (D. G. Carpenter, 2005, personal communication). Partly open and open fractures are commonly associated with local solution enlargement along fracture walls, and with bitumen cementation and corrosion in the surrounding rock matrix. Matrix corrosion occurs as microporosity in the micritic components of the rock, for example, skeletal grains and microbial fabrics (Figure 21A). In facies containing micritic matrix, microporosity accompanied by small vugs can sometimes be extensively developed. In addition to

FIGURE 20. Earlier stages of diagenesis in the Tengiz rim and flank. Voids are filled by a prebitumen cement sequence (A and B) that includes inclusion-rich, probably marine calcite (1), prismatic to equant zoned calcite and inclusion-rich dolomite (2), and clear blocky equant calcite (3). This sequence is commonly followed by bitumen (4). Prebitumen cements filled much of the primary pore space, as well as porosity created during early diagenetic dissolution and fracturing. Some voids are filled only with equant calcite and bitumen (C and D), indicating a separate, later fracturing and dissolution event. Even later dissolution is suggested by corrosion along crystal growth planes (5) in the equant calcite. Fluidinclusion homogenization temperatures of 80 – 958C for the equant calcite indicate the existence of prehydrothermal conditions in the reservoir prior to bitumen formation. Scale bars = 1 mm.

78 / Collins et al.

matrix corrosion adjacent to solution-enlarged fractures, bitumen and corrosion are noted between cements and walls of partly filled fractures. Bitumen and corrosion are also sometimes observed within inclusionrich prebitumen cements or along cleavage and growth planes of individual cement crystals. Bitumen and matrix corrosion are both observed to decrease away from vugs and enlarged fractures, suggesting that bitumen, corrosion, and large-scale dissolution may be genetically linked. Petrographic observations indicate that significant corrosion and solution enlargement postdates the early cement sequence. A key question is whether most of this occurred before, during, or after bitumen formation or a combination of the three. Examples where bitumen-associated

microporosity and bitumen-free microporosity occur in the same thin section are common (Figure 21B), and in some cases, dissolution is observed that appears to postdate bitumen formation (Figure 21C). This suggests that some matrix corrosion probably accompanied bitumen emplacement, and that some dissolution may also have occurred later along the same pathways. However, many other instances indicate mainly passive fill of open void space by bitumen, accompanied by only minor micrite corrosion. Other postbitumen diagenesis includes fractures, noncarbonate cement such as fluorite and quartz (based on fluid-inclusion temperatures; see Appendix 2), and possibly some calcite spar (Figure 21D). The existence of syn- and postbitumen cementation and dissolution

FIGURE 21. Bitumen-related and postbitumen diagenesis in the Tengiz rim and flank. Bitumen is dominant in some voids and fractures (A, 4) and is commonly associated with corrosion of microbial fabrics (A, 5). Some micrite corrosion appears to be free of bitumen (B, 6) and possibly postdates bitumen formation. Postbitumen diagenesis also includes dissolution (selective dissolution of dolomite rhombs in a partial dedolomite, C) and late fracturing (D). The example in (D) also shows postbitumen calcite cement. A key question in the Tengiz rim and flank is how much postbitumen dissolution and fracturing contributes to present-day reservoir quality and distribution. Scale bars = 1 mm, except where indicated.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 79

implies that an active hydrothermal system was probably a factor in the development of the rim and flank reservoir. Two scenarios are proposed: hydrothermal fluid mixing and geothermal convection. Mechanisms for hydrothermal mixing includes fluids with different H2S concentrations or thermochemical sulfate reduction (TSR) reactions between hydrocarbons and sulfate minerals. These possibilities are suggested by: 1) the elevated H2S content (13%) of hydrocarbons in the reservoir 2) the presence of known TSR byproducts in the reservoir 3) the thermal regime, estimated from fluid-inclusion studies, under which these products were precipitated 4) the presence near or within the reservoir of geological conditions and ingredients favorable for TSR reactions. Geothermal convection is supported by twodimensional numerical flow models of Tengiz using temperatures derived from basin modeling, known thermal rock and fluid properties, and a simplified layering scheme based on bulk reservoir permeabilities. It is also possible that both convective and H2Srelated processes contributed to rim diagenesis and reservoir quality at different times or at the same time.

Hydrothermal Fluid Mixing and H2S-related Diagenesis According to basin models, the Tengiz reservoir attained hydrothermal temperatures during the early stages of post-Triassic rapid subsidence (Figure 22). During and after this time, basin-generated hydrothermal fluids may have periodically charged the reservoir, mixing with formation fluids. Thin sections and quantitative x-ray diffraction data from core plugs reveal the presence of anhydrite, dolomite, chalcedony, quartz, barite, fluorite, fluorapatite, and pyrite as potential hydrothermal species in the rim and flank. Most of these minerals exist in quantities less than 4% by weight. Species that may have a hydrothermal origin and have been measured as much as 50% and higher by weight in individual samples include fluorapatite, anhydrite, and dolomite (anhydrite and dolomite may also have other origins). Chalcedony and quartz are also locally abundant. Homogenization temperatures of 110 – 1258C have been obtained from examples of quartz and fluorite (Appendix 2). Significant calcium carbonate dissolution can occur in reservoirs where fluids with different H2S

saturations are mixed (Hill, 1995), or when H2S and CO2 generated by TSR mix with oxygenated formation water in a carbonate reservoir (for example, see Hill, 1990). In the absence of oxygenated formation waters, the amount of carbonate dissolution that occurs in the presence of H2S and CO2 is controlled by pH of the formation waters and the presence of iron oxides or aqueous sulfides (Stoessell, 1992). Carbonate dissolution commonly occurs because of carbonic and/or sulfuric acid generated during these reactions (Hill, 1987; Machel, 1987, 1989). Once H2S saturation became established in the reservoir, either by migration of H2S and CO2 gases or after migration of hydrocarbons, additional dissolution may have occurred as a result of periodic mixing with external hydrothermal fluids during even later burial. The high burial temperatures, elevated H2S levels, and the presence of bitumen, dedolomite, and anhydrite suggest the possible involvement of TSR in diagenesis of the rim reservoir. In the Tengiz region, potential reactants include hydrocarbons generated from Carboniferous to Permian subsalt source rocks adjacent to the buildups (Lisovsky et al., 1992) and an anhydrite layer at the base of the thick Permian halite formation that encases the buildups. Figure 23 shows two possible regions where reactions might have occurred: 1) basinal areas between platforms where hydrocarbons were being generated; in this scenario, carbonate dissolution occurred after migration of H2S and CO2 from reaction sites into the reservoir 2) in the reservoir itself, with carbonate dissolution resulting from reactions that occurred when liquid hydrocarbons entered the reservoir The latter mechanism is possible because the absence of Moscovian–Artinskian volcanics on the upper slope of the Tengiz flank places the anhydrite layer in direct contact with the reservoir. It has the advantage of also accounting for, as a common by-product of TSR, at least some of the ubiquitous bitumen in the rim and flank. Reservoir temperatures of 80 –958C achieved prior to bitumen emplacement (based on prebitumen equant calcite) indicate that bitumen formation probably occurred as the reservoir entered the hydrothermal regime. Carbonate reservoirs exposed to TSR commonly contain anhydrite, dedolomite, sulfide minerals, elemental sulfur, and uranium and iron enrichment (Hill, 1990, 1995). In particular, anhydrite and dedolomite of apparent nonevaporative or nonsubaerial origin are commonly observed in cores and thin sections

80 / Collins et al.

FIGURE 22. Temperature and burialhistory plots for the Tengiz rim and flank reservoir from basin modeling. The vertical depth range indicates the suite of samples from different parts of the reservoir interval used in fluidinclusion analyses (Appendix 2). Bitumen formation in the reservoir began, probably under hydrothermal conditions, after cementation by equant calcite (top) during Mesozoic subsidence. Potential times of burial fracture generation include stresses generated during salt diapirism, uplift (approximately 220 – 175 Ma), rapid Mesozoic subsidence, and the present-day stress regime (bottom).

from the Tengiz rim, and GR logs from rim and flank penetrations typically show uranium enrichment (see Figures 18, 19). The presence of dedolomite suggests the involvement of fluids with excess Ca2+, possibly from TSR-related anhydrite dissolution.

Geothermal Convection Groundwater circulation can be important in the diagenetic history of carbonate reservoirs, particularly for isolated platforms (Jones et al., 2004). Platform margins are favorable sites for convective circulation during burial, as well as during deposition. Depositional or early postdepositional convective processes were previously briefly cited with respect to early dissolution at the Tengiz margin. Geothermal convection can occur during burial under certain conditions (Machel and Anderson, 1989; Morrow, 1998; Wendte et al., 1998; Heydari, 2000). Two-dimensional numerical fluid-flow simulations indicate that largescale geothermal convection is feasible in the Tengiz

rim and flank (G. D. Jones and Y. Xiao, 2005, personal communication), mostly because of the high vertical permeability possible in fractured reservoirs and because of a relatively large center-to-edge temperature gradient in the buildup caused by the thick, conductive envelope of halite around the exterior. Once convection cells are established in a closed system, areas of net dissolution and net cementation can occur while maintaining average saturation equilibrium in the cell with respect to the reservoir. Because of the inverse relationship at constant pH and pCO2 between calcium carbonate solubility and temperature, dissolution occurs where saturated or nearly saturated fluids are cooling (rising cell limbs), and cementation occurs where such fluids are warming (descending limbs). However, this process results in mass transfer of carbonate from one area to another, and closed-system cells may, over time, be forced to readjust in shape or shut down in response to changing permeability patterns. Numerical models further indicate that convection cells, if they remain active long enough, can transfer sufficient calcium carbonate to significantly affect porosity in the reservoir (G. D. Jones, 2005, personal communication). Several factors may have facilitated long-term active convection in the rim and flank of Tengiz. The absence of Moscovian to Artinskian sediments on the upper slope of Tengiz (Figure 23) may have allowed continuous seawater exchange for rim and flank convection throughout an extended period following deposition. Second, Tengiz appears to have

FIGURE 23. Evolution of the Tengiz rim and flank high-permeability reservoir, based on data and analyses contained in the text. Prior to burial (A), the reservoir consisted of fractures and large vugs caused by syndepositional instability and platform margin karsts. Initial large-scale dissolution and cementation patterns may have been controlled by Kohout convection and by meteoric and mixing processes. During early burial (B), they may have been controlled by geothermal convection, promoted by still-open fractures and by deposition of thermally conductive halite. Conductivity loss and differential compaction caused by salt diapirism may have reduced or eliminated geothermal convection (C), although compactional stresses may have rejuvenated the fracture system. Prebitumen cements had significantly reduced nonmatrix porosity and permeability by the end of this stage. New tectonic stresses again reactivated the fracture system, possibly facilitating movement and distribution of hydrothermal fluids in the reservoir (D). Diagenesis related to thermochemical sulfate reduction (TSR) is believed mainly responsible for solution enlargement of open fractures and lost circulation zones (LCZ), either from migration of CO2 and H2S gas, or as a direct result of the liquid hydrocarbon charge. These fluid charges may also be responsible for enhanced dissolution in adjacent platform facies, forming the transitional reservoir zone (TRZ). Note that overall fracture density, but not necessarily openness, increased in reservoir over time.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 81

82 / Collins et al.

porosity (fractures, large vugs, and lost circulation zones) in some facies, the rim and flank are considered a dualporosity reservoir. Different reservoir facies in the rim and flank can be recognized based on both matrix and nonmatrix pore types.

Matrix Reservoir Types Air permeability data combined with thin sections from core plugs were used to guide the recognition of matrix reservoir rock types in the Tengiz rim and FIGURE 24. Tengiz rim and flank paragenesis. Cementation events are indicated in flank. This data, while not black, dissolution events are indicated in white, and events that involve mainly replacement or recrystallization are shown in gray. Approximate absolute timing has quantitatively accurate for been established for zoned prismatic spar, dolomite cement, and subhedral equant characterization at reservoir calcite (150 ± 25 Ma; see Figure 22 ) from fluid-inclusion results. Based on petrographic conditions, nevertheless reprelationships, these cements were emplaced just prior to bitumen deposition (see resent a relatively consistent Figure 20) and the time when hydrothermal temperatures were attained in the reservoir. All other diagenetic products are shown in relative order only, as part of either laboratory test that is useful for distinguishing reservoir the prebitumen or postbitumen sequences. Other phases might be constrained by processes related to burial history, which has an implied time scale (see Figures 22, 23). rock types based solely on rock properties. The classification scheme used for analysis of Tengiz plugs was had a complex fracture history (Figure 24). Multiple modified from that of Lucia (1999) and recognizes fracture events and/or repeated reactivation of prefive matrix pore-type classes (interparticle porosity, vious fractures in the rim and flank may have mainmicroporosity, touching vugs, separate vugs, and fractained or increased vertical permeability over time, tures) based on independent permeability behavior allowing geothermal convection to remain active duras a function of porosity. Applied to Tengiz, these ing burial. data permit the distinction of three main reservoirs based on matrix properties: RESERVOIR QUALITY AND DISTRIBUTION The distribution of reservoir rock types in carbonate reservoirs is a product of interactions between depositional facies and diagenetic processes. Reservoir quality should tend to average out somewhat when comparing deeper well penetrations or wells from the same environment in the field because these wells encounter a wider range of stratigraphic ages, facies, and potential diagenetic themes in the Tengiz spectrum. Reservoir quality at Tengiz has so far shown a stronger relationship to specific well locations than to facies or stratigraphy. A partial explanation for these geography-based reservoir-quality variations lies in understanding the relative contributions of dissolution, cementation, and fracturing to matrix and nonmatrix porosity in the outer platform, rim, and flank. Because of the significant contribution of nonmatrix

1) The rim and flank reservoir (Figure 25) encompasses upper-, middle-, and lower-slope facies containing mainly microbial boundstone or breccia (but also including some interbedded platform-derived facies). 2) The transitional reservoir zone (Figure 26) consists of shallow-platform or outer-platform facies with significant bitumen cement and matrix dissolution. This term is used instead of outer platform because the reservoir type is not restricted to that facies. It should not be confused with the term ‘‘transition zone,’’ commonly used to describe certain conditions in hydrocarbon columns. 3) The central platform reservoir includes the central regions of the Tengiz platform that are

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 83

FIGURE 25. General characteristics of the rim and flank reservoir. (A) Local solution enlargement of a partly cemented fracture-breccia in microbial boundstone. Dissolution is found along the fracture walls and between cement crystals. Bitumen-filled microporosity (darker color) extends a few centimeters from the fractures, but matrix porosity is otherwise low. A lost circulation zone is associated with this feature. (B) Microbial boundstone with low matrix porosity and systematic bitumen-filled hairline fractures. (C) Probable fractured zone in skeletal-lithoclast rudstone with extensive matrix corrosion, resulting in core rubble. The dark color is caused by pervasive bitumen cementation. (D) Microbial boundstone (clast in breccia) with bitumen and miroporosity developed along microbial filaments. (E) Microbial boundstone (clast in breccia) with vugs formed by localized solution enlargement between microbial masses. relatively unaffected by bitumen plugging and associated effects (see Kenter et al., 2006). It can be demonstrated that much of the Visean A, Serpukhovian, and Bashkirian platform reservoir is in pressure communication with the rim and flank. As shown in Figure 23, the transitional reservoir zone refers to parts of the outer regions of these platforms that may have been altered by the incursion of various diagenetic fluids active in the rim. For example, the extent of this region may represent the equilibration limit of corrosive hydrothermal fluids that charged the reservoir or the innermost edges of active convection cells. The greatest distinction in matrix properties is between the transitional reservoir zone, affected by bitumen plugging and matrix dissolution, and the central platform reservoir (Figure 27). Because similar depositional facies are present in both the outer and central regions for shallow-platform cycles, bitumen and dissolution are mostly responsible for the observed differences. In some cycles, early porosity and permeability in the transitional reservoir zone may have been adversely affected by the presence

of deeper, outer-platform facies or by the presence of microbial or marine cements (for example, see Figure 6). Figure 27 shows a more subtle contrast between the transitional reservoir zone and the rim and flank reservoir. Because bitumen and dissolution affected both of these reservoirs, the difference is more likely caused by the grainy nature of transitional reservoir zone facies, which probably had higher matrix porosity prior to bitumen and dissolution, in contrast with rim and flank facies that are dominated by early-cemented microbial boundstone and boundstone breccia.

Transitional Reservoir Zone The porosity range for the transitional reservoir zone determined from plug data is intermediate between the central platform range and that of the rim and flank reservoirs (Figure 27A). The data also show a large amount of permeability scatter around the central platform trend (Figure 27B). The transitional reservoir zone matrix is characterized by larger pore sizes because of solution enlargement of early interparticle and moldic porosity and because smaller

84 / Collins et al.

FIGURE 26. Characteristics of the transitional reservoir zone, which affects the outer regions of the Visean A, Serpukhovian, and Bashkirian platforms. Facies include skeletal packstone-rudstone with brachiopods and large crinoids (D, F, J, and K), sponge or coral boundstone (E and H), and occasional high-energy ooid grainstone (G). These examples have in common dissolution enlargement and bitumen cementation in the matrix. Dissolution enhancement includes matrix vugs (D, arrows) and enlarged fractures or stylolites (D, J, and K). Bitumen (darker colors in core examples) is somewhat fabric selective in (E – H). The example in (F), interpreted as a karst feature, shows differences in relative bitumen abundance relative to the fracture and to the clasts in the cavity. In thin section, bitumen is commonly associated with minor corrosion (cement corrosion in B and C, arrows). Bitumen and solution enlargement have offsetting effects on porosity and permeability, as shown in (A), in which both bitumen-cemented pores (arrows) and solution-enlarged pores are visible.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 85

pores tend to be bitumen filled (Figure 26A, B). Larger pores exhibit only partial fills or incomplete linings. The reduced porosity range in the transitional reservoir zone, thus, appears to be caused by bitumen cementation of smaller pores, whereas the permeability scatter appears to be caused by a combination of bitumen plugging and solution enhancement that resulted in a fundamental change in matrix pore size, shape, and connectivity. For the same porosity value or porosity range, the transitional reservoir zone has slightly better average matrix permeability compared to the central platform reservoir. Thus, whereas bitumen plugging appears to have reduced the porosity range in the transitional reservoir zone, an expected reduction in permeability over that range has been offset by dissolution. Because significant matrix porosity likely remained in the transitional reservoir zone after depositional and early diagenetic processes, cyclic controls on platform porosity probably influenced patterns of later dissolution and cementation. The net effect of porosity changes caused by bitumen plugging and matrix dissolution overprinted across facies results in a noticeable change in the expression of cyclicity

on porosity logs. In particular, dissolution somewhat disguises the loss of porosity observed at sequence boundaries in the central platform (Kenter et al., 2006) and adds high-frequency vertical porosity variations in the cycles. Log character thus provides a means of constraining the width of the transitional reservoir zone along the outer edge of the platform. Based on current well and core control, the width appears to vary vertically and may range from less than 500 m (1640 ft) to more than 2 km (1.2 mi).

Rim and Flank Reservoir The data from the rim and flank reservoir in Figure 27 indicate a further reduction in matrix porosity range but permeability scatter similar to the transitional reservoir zone. Although it was originally anticipated that matrix reservoir quality might depend on the amount of grainy matrix in different facies, comparisons of matrix-rich breccias, more tightly packed breccias containing little or no matrix, and in-situ microbial boundstones do not reveal greatly different porosity ranges (Figure 28). It is also surprising, given that matrix dissolution features are less

FIGURE 27. Plug porosity and permeability data comparing the matrix reservoir from the central platform, transitional reservoir zone (TRZ), and rim and flank areas. The central platform trend includes almost 5000 plugs (only a partial set is plotted) and demonstrates a single trend despite a variety of facies, textures, and reservoir rock types represented. The rest of the data consists of approximately 1700 plugs from the rim and flank and 700 from Bashkirian and Serpukhovian outer-platform facies representing the TRZ. Increased permeability scatter and reduced porosity ranges characterize both the TRZ and the rim and flank trends compared to the central platform (A). In the TRZ, reduced porosity is caused mainly by bitumen, whereas the permeability scatter is caused by a change in pore connectivity from the combination of bitumen plugging and solution enlargement. In the rim and flank, the porosity range is reduced further (B), because of the presence of tightly cemented microbial boundstone facies. The permeability scatter is partly a function of solutionenlarged pores, but also because of the increased importance of vuggy-moldic pore types and microfracture porosity.

86 / Collins et al.

FIGURE 28. Histograms from plug data comparing matrix porosity ranges in the rim and flank by facies (A) and by well (B). Boundstones and breccias have similar distributions, rudstones have slightly better porosity (based on higher frequency in the 3 – 5% bins), whereas fine-grained platform-derived facies have a lower porosity range. Overall, however, porosity variability between facies is instead weak compared to the variations by well, suggesting a strong diagenetic overprint. abundant overall and appear to be more poorly connected, that rim and flank matrix permeability is comparable to that of the transitional reservoir zone. This may be partly an artifact of the increased abundance of small fractures, producing a greater amount of plug damage. However, dissolution is observed along some small fractures in plugs, so whereas the number of plugs with high permeability may be inflated, the range is probably real. It should also be remembered that these are comparative statistical relationships that only apply where matrix porosity exists and cannot be extrapolated to generalizations about the overall distribution of porosity. Porosity in upper- and middle-slope microbial boundstones and breccias was substantially reduced by sediment fill and prebitumen cementation of early pore types. These include fenestrae, microbial growth and framework cavities, and other primary voids, as well as early fractures and dissolution voids. In middle- to lower-slope and apron facies, it was expected that the presence of skeletal rudstones or floatbreccias with grainy matrix might improve reservoir quality in some areas because of possible preservation of primary porosity, compared to areas containing tightly packed breccias or breccias with finer grained or micritic matrix. However, like the upper slope, matrix reservoir quality in most middleto lower-slope facies is affected by bitumen, corrosion, and prebitumen cement, in particular syntaxial calcite (of probable early to moderate burial origin) and clear equant calcite (including 908 calcite of burial origin). Early marine cements prevalent in the

upper-slope facies are less noticeable. Burial cementation and dissolution processes may be part of the reason why there is no good evidence that breccias, in general, have better reservoir quality compared to other facies, or that better reservoir quality is associated with a particular type of breccia.

Matrix and Nonmatrix Reservoir Distribution Nonmatrix features consisting of fractures, large vugs, and lost circulation zones are an important component of the rim and flank reservoirs. These elements are absent or greatly reduced by comparison in the central platform area. The distribution of these features and their relationships to matrix properties are the basis of rim and flank reservoir characterization. As described previously, open fractures, vugs, and lost circulation zones appear to be linked to areas of bitumen cementation and solution enlargement (see Figure 25). Whereas bitumen is associated with dissolution along fractures and in the surrounding matrix, it is also responsible for plugging up some fractures and vugs. Furthermore, bitumen and dissolution affect matrix reservoir quality to varying degrees in fractured intervals, and matrix dissolution and bitumen can sometimes be more extensively developed in areas of lower fracture density. Understanding these variations in spatial and genetic relationships is a key to improved reservoir-quality prediction. Bitumen-associated matrix dissolution includes microporosity, micrite corrosion, vugs, and enlarged primary voids. In detrital middle- to lower-slope and

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 87

apron breccias, residual matrix porosity controlled by primary texture and prebitumen cementation appears to have regulated the extent to which matrix corrosion is developed locally. Incompletely cemented grainy facies lend themselves to more extensive corrosion, and small-scale variations in these parameters sometimes result in apparent bedding or facies control, obscuring a relationship to larger scale fracture density. In upper-slope boundstones and breccias, microporosity occurs mainly in primary microbial fabrics and entrained micritic skeletal grains. Because little or no matrix is present in these rocks, a general relationship to open and partly open fractures is more clearly evident. The amount of matrix corrosion and dissolution that develops depends on the type of boundstone (which, in turn, determines the density of entrained grains and microbial fabrics) and on the amount of prebitumen cement (commonly abundant because of high volumes of marine cement). Whereas this combination can produce a complex-looking distribution of corrosion and bitumen relative to primary textures in boundstones and breccias, a clear spatial connection exists between the matrix and nonmatrix components of bitumen and associated dissolution, and variations in matrix components are determined more by prebitumen cementation patterns than by fracture density or facies.

Early Karst Features Early karst features are commonly cited as important influences on porosity and permeability distribution in carbonate reservoirs. The overall effect may be positive or negative (Esteban and Wilson, 1993; Wagner et al., 1995). Early karst features can vary considerably in spatial distribution and relationship to stratigraphy (for example, Choquette and James, 1987; Lucia, 1995; Mylroie and Carew, 1995), although for isolated platforms, dissolution is commonly concentrated near the platform margin (for example, Cander, 1995; Tinker et al., 1995). Early karsts may also influence later diagenesis, serving as targets for selective re-excavation or forming significant impediments to later fluid flow (Ford, 1995). Large-scale features interpreted as early karst are present at Tengiz; however, observed relationships between these features, the distribution of later diagenetic features (bitumen cement and corrosion), and overall porosity and permeability suggest that early karsts are largely incidental to reservoir quality. Potential early karst features include local subaerial exposure effects associated with sequence boundaries, brecciated intervals, and large, sediment-filled

cavities (Figure 29). Almost all of these features are completely filled with sediment and, to a lesser extent, cement. Later bitumen and corrosive diagenesis are present in these intervals and, in some cases, are even concentrated within or around the karst or dissolution features (Figure 29A). However, when compared to randomly selected cored intervals or to cored intervals of equal thickness where similar features are absent, the intervals containing the karst features tend to have, on average, diminished amounts of bitumen and dissolution, and average reservoir quality in these zones does not appear to be consistently better or worse than elsewhere.

Lost Circulation Zones Lost circulation during drilling in the rim and flank can be shown to occur in association with increased open-fracture density or with solution enlargement of fractures in instances where FMI control, or in some cases, core control, exists. Recent drilling without catastrophic losses and with more complete core recoveries in intervals dominated by vuggy porosity further suggests that perhaps the distribution of fractures or solution-enlarged fractures are the most significant controls on lost circulation in general. Because solution-enlarged fractures are commonly associated with areas of dissolution, corrosion, and bitumen, an important question is whether fracture distribution was the primary control on where solution enlargement and lost circulation occurred, or whether the distribution of corrosion and dissolution, constrained by other factors, determined where fracture enlargement and lost circulation occurred. Ultimately, the distribution of lost circulation zones in the rim and flank may indicate specific pathways through the reservoir taken by corrosive fluids or map out areas of elevated fracture density. As previously outlined, different fracture sets in the Tengiz rim and flank can be distinguished by orientation relative to core, by types of fill, and by the amount of dissolution. Possible origins are varied, including syndepositional instability, burial stresses created by differential compaction of the rim and platform areas (and, to a lesser extent, the rim and flank areas), differential stresses resulting from overburden loading during salt diapirism, and stresses associated with various tectonic events affecting the Precaspian Basin (D. G. Carpenter, 2005, personal communication). Fracture density controlled by the mechanical properties of different facies is one possible influence on the current distribution of lost circulation zones. Both core and FMI data show that microbial

88 / Collins et al.

FIGURE 29. Core examples with sediment-filled features of possible karst origin. (A) Cavities filled with laminated, greenish-colored shale and carbonate formed in outer-platform crinoid skeletal packstone-rudstone. Cavity walls (yellow dashed lines) are surrounded by a zone of matrix corrosion and bitumen (white lines), but matrix reservoir quality overall is marginal. (B) Cavity developed in outer-platform to upper-slope boundstone filled with laminated green-gray marl and boundstone breccia. Because of the unique setting of the T-4635 well (see Figure 15A), this feature could be either a karst feature or a cavity that opened during rim detachment and subsequently filled in a deeper water setting. (C – E) Examples of smaller, geopetally filled fenestral cavities. The sediment fill sequence, which includes green and red clay (C and D), banded (marine) cement and cream-colored micrite (D), and dark-colored sediment (D and E), suggests a complex history of cavity formation and fill. All examples are associated with relatively poor matrix reservoir quality, however.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 89

FIGURE 30. Distribution of lost circulation zones (LCZ) in the rim and flank. Most wells in the rim and flank have lost circulation somewhere in the reservoir section. The charts represent 48 LCZ in 13 wells (T-463, T-4346, T-4635, T-4556, T-5056, T-5059, T-5435, T-5454, T-5857, T-6261, T-6457, T-7252, and T-7453) that have been quantified by sustained loss rate, depth, and thickness. The distribution by distance in meters from the top of the section does not show a clear trend (left), but when the distribution is normalized to indicate relative proportion in the rim interval regardless of well position on the rim profile (right), the pattern supports anecdotal evidence that LCZ are more common in the upper part of the section, regardless of facies present there.

boundstones and tightly packed breccias have a greater fracture density compared to rudstones, packstones, and matrix-supported breccias, suggesting that lost circulation zones might be more common in upper- to middle-slope facies. However, based on well occurrences, it can be demonstrated that lost circulation zones are more common in the upper half to one-third and slightly more common near the base of the rim and flank section, regardless of position in the field (Figure 30). Such a distribution is inconsistent with a primary facies control on lost circulation and, like many other observations, points to a diagenetic overprint. Current well spacing is still insufficient to clearly demonstrate geographical distribution patterns for lost circulation zones, a situation made even worse because most rim and flank wells drilled to date have encountered lost circulation somewhere in the reservoir section. More generally, facies in which fracture permeability was dominant compared to residual matrix permeability may be more prone to lost circulation because later dissolution was concentrated along fractures. Facies with higher residual matrix permeability (for example, poorly cemented

grainy facies and loosely packed breccias) may have allowed corrosive fluids to spread away from fractures and, thus, may ultimately prove less prone to lost circulation. This may be a reason why, in the transitional reservoir zone where permeable grainy facies are most abundant, a relationship between lost circulation zones, bitumen, and corrosion is less obvious. Combined with reduced fracture density in the transitional reservoir zone, this factor may also be why lost circulation zones are less frequent in the outer platform region. Whatever diagenetic processes prove ultimately responsible for dissolution and lost circulation, their effects have produced a reservoir with large-scale interwell connectivity across the transitional reservoir zone, rim, and flank, despite a major stratal surface separating them and a complex facies distribution within each.

SUMMARY AND CONCLUSIONS Based on currently available data, the Tengiz rim and flank are interpreted to have accumulated in two stages. The early stage was dominated by instability and relatively continuous large-scale failure and mass wasting, resulting in deposition of debris aprons around the base of the buildup consisting of coarse boundstone breccias and grainy platform-derived sediments. In the second stage, a wedge consisting of shallow- and outer-platform facies, upper-slope

90 / Collins et al.

microbial boundstone, middle-slope boundstone breccia, and lower-slope breccia and platform-derived sediment prograded out over the early debris apron. This stage was more successful along the eastern margin of the buildup because of favorable platform profile geometry, or elevated microbial productivity, or reduced tectonic influence in those areas, or some combination of those factors. Along the western margin of the buildup, progradation was only locally or partially successful, and rim failure predominated throughout the Serpukhovian and Bashkirian. This study does not provide a process-based, fully predictive model of the distribution of reservoir properties in the Tengiz outer platform, rim, and flank. It raises as many questions as it answers, and some critical uncertainties remain. Some conclusions have alternatives and will likely require modification or change as drilling continues. Future surprises are probable, and subsequent generations of models will no doubt be different from the present one in some manner. However, the following observations indicate factors that are likely to have a significant impact on reservoir quality and prediction in the rim and flank: 1) Reservoir quality is dominantly controlled by fractures and large-scale dissolution and is a function of earlier porosity destruction because of prebitumen cementation versus later dissolution associated with bitumen formation. 2) Porosity destruction caused by sediment infill and prebitumen cementation was significant and affected primary voids, early dissolution voids, early fractures, and early karst features. In particular, early fractures and large-scale dissolution features or karsts are largely filled and do not have a major impact on reservoir quality. 3) Corrosive dissolution and bitumen are a latediagenetic overprint and are responsible for significant porosity modification in a wide variety of features including later fractures, microbial fabrics, micritic skeletal grains, breccia clasts, certain cements, and stylolites. 4) Facies control of fracture density results in both open and closed types being more common in mechanically rigid facies, such as microbial boundstone, cemented grainstone, and clast-supported breccia, and less abundant in facies such as rudstone, packstone, and matrix-supported breccia. 5) Lost circulation zones appear to be more commonly associated with solution-enlarged fractures than with vuggy intervals and are therefore likely

controlled by either fracture distribution or the distribution of late dissolution fairways. 6) Enhanced matrix porosity caused by late dissolution and bitumen cement is commonly better developed where solution-enlarged fractures are present; however, extensive matrix corrosion may be associated with intervals that are less likely to produce severe lost circulation. 7) Reservoir-quality variability exists at large and small scales, resulting in considerable well-to-well differences in regions that are similar in terms of facies, stratigraphy, and depositional history. This suggests that facies is a weak secondary control on reservoir quality.

ACKNOWLEDGMENTS This study involved the work of a large group of geoscientists and specialists. The authors recognize the following individuals for insightful contributions to this manuscript: Kevin Nahm (Chevron), who believed early on that mechanically isotropic microbial boundstones could be mapped with seismic analysis; Brent Francis (ExxonMobil), who progressed this idea through development of the compaction model and who provided seismic interpretations for the debris apron concept; and Tom Heidrick (retired from Chevron), who engaged in valuable discussions regarding the minimal effects of early karst and the potential for later burial fractures. Wayne Narr (Chevron), Jim DeGraff (ExxonMobil), and Dan Carpenter (ExxonMobil) are acknowledged for building on this perception through rigorous description and analyses of fracture characteristics and distribution. Perhaps most importantly, Michael Clark (Chevron) is thanked for initiating an intense coring and core analysis program that formed the basis for most of our geologic studies and for his strong support of the sequence-stratigraphic framework. We also gratefully thank the following individuals from ExxonMobil who made significant contributions of supporting data or data analysis: Gareth Jones and Yitian Xiao (numerical simulation of Tengiz diagenetic processes); Sean Guidry (fluid-inclusion analyses); Paul Hicks (temperature histories from basin modeling); Shin-Ju Ye and Peter Hillock (facies analysis of FMI logs). Early versions of the manuscript were improved by helpful reviews from Niall Toomey, Kelly Bergman, and Gareth Jones. Finally, we thank TengizChevroil and its shareholder companies for permission to publish this study.

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 91

APPENDIX 1: RIM AND FLANK FACIES (MODIFIED FROM WEBER ET AL., 2003)

Lithofacies and Microfacies Crinoid-brachiopod skeletal rudstone to packstone Outer platform breccias and conglomerates

Brachiopod rudstone (coquina) to boundstone

Microbial boundstone

Description Medium-dark gray; large crinoids and/or fragmented thick-shelled brachiopods dominate; sometimes with microbial fabrics and cements Grainstone floatbreccia to packbreccia: light-medium brown; subangular to subrounded clasts of platform-derived grainstone and packstone; skeletal wackestone to grainstone matrix, sometimes with microbial fabrics and cements Grainstone packbreccia: light brown, gray, and black; angular clasts of grainstone and rudstone with mixed lithofacies types Coral rudstone to conglomerate: medium brown to gray; colonial coral fragments; coarse-grained skeletal matrix with abundant crinoids Light gray-brown to medium brown-gray; brachiopod shell fragments and whole shells; some articulated, in feeding or growth position; abundant micritic matrix, some microbial fabrics and calcite cement; minor coarse skeletal debris, crinoids Type C (skeletal-micritic boundstone): light-medium brown, massive to peloidal with abundant fenestrae; crinoids, brachiopods, fenestellid bryozoans, algae, forams, and other skeletal grains bound together by clotted microbial fabric and multiple cement generations (as much as 50% cement) Type B (laminar-cemented boundstone): light brown; irregular concentric or wavyaminated fabric consisting of microbial filaments and early cement (botryoidal, fibrous and radiaxial; inclusion rich; as much as 75%); minor skeletal grains, including fenestellid bryozoa, crinoids, ostracods, thin-walled pelecypods and forams Type A (peloidal-micritic boundstone): light gray-light brown; generally massive or featureless, with minor growth structures and abundant irregular fenestrae or tubules; clotted peloidal fabric and early cement (as much as 25%); sparse fauna, including fenestellid bryozoans, ostracods, and thinwalled pelecypods

Environment Deep subtidal outer platform, possibly below wave base; abundant in central to upper parts of late Visean and Serpukhovian outer-platform cycles Deep outer platform, possibly lowstand; basal parts of Serpukhovian outer-platform cycles

Shallow outer platform; lag or possibly karst? Tops of Serpukhovian outer-platform cycles Shallow outer-platform; basal parts of Serpukhovian outer-platform cycles; possible patch-reef detritus Shallow outer-platform; basal parts of Serpukhovian outer-platform cycles; mud-mounds, mound detritus, and intermound deposits

Serpukhovian deep, subtidal, outer-platform to upper-slope platform break

Serpukhovian upper slope

Serpukhovian upper to middle slope

92 / Collins et al.

APPENDIX 1: (CONT.)

Lithofacies and Microfacies Boundstone breccia

Crinoid-lithoclast rudstone-packstone

Skeletal-lithoclastic rudstone-grainstonefloatstone

Allochthonous grainstone

Lime wackestonemudstone

Description Welded to mosaic breccia: light-medium brown and gray; subangular centimeter- to meter-size clasts of upper-slope boundstone types; tight, clast-supported fabric, sometimes with stylolitic clast boundaries (welded breccia), sometimes with blocky calcite cement (mosaic breccia); minor platformderived matrix (skeletal grains, large crinoids); occasional platform-derived grainstone or packstone lithoclasts Float- to packbreccia: light-dark brown and gray; subangular to subrounded, millimeterto meter-size clasts of upper slope-derived boundstone types; grainstone-packstone matrix with platform-derived skeletal grains; large crinoids and fragmented brachiopod shells; occasional platform-derived grainstone or packstone lithoclasts Brown-light gray; poorly to moderately sorted, crinoid-rich rudstone (platformderived); fragmented brachiopod shells; sand-size skeletal grains and millimeter- to centimeter-size lithoclasts (upper slope and platform derived) Medium-dark gray; thin to thick bedded, wavy-bedded to laminated, graded; locally siliceous to slightly dolomitic; platformderived grains (forams, crinoids, algae, brachiopods, crinoids); occasional millimeter- to centimeter-size boundstone or slope-derived clasts, volcanic siltstone or sandstone clasts Light brown-medium gray; moderately sorted ooid or skeletal grainstone; massive to well bedded Brown, gray, and black; argillaceous, siliceous, and commonly dolomitic; thin laminated to thin bedded; sponge spicules, minor skeletal grains, and sparse radiolarians; commonly interbedded with volcanic silt and sand; potential source rock

Environment Serpukhovian upper to middle slope and debris apron

Serpukhovian middle to lower slope and debris apron

Serpukhovian middle to lower slope and debris apron

Serpukhovian lower slope and debris apron

Lower slope and debris apron; Serpukhovian and Bashkirian Lower slope, debris apron and basinal; Visean, Serpukhovian, and Bashkirian

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 93

APPENDIX 2: FLUID-INCLUSION RESULTS

Well

Depth (m)

T-4635 T-4635 T-4635 T-4635 T-4635 T-4635 T-4635 T-4635 T-4635 T-47 T-5056 T-5056 T-5056 T-5056 T-5056 T-5056 T-5056 T-5056 T-5857 T-5857 T-6337 T-6337 T-6337 T-6337 T-8 T-8 T-8 T-8

4466.25 4466.25 4482.63 4482.63 4482.63 5055.25 5055.25 5055.25 5055.25 5392.5 4194.2 4194.2 4200.47 4200.47 5453.55 5453.55 5453.55 5476.22 4816.93 4816.93 4894.37 4945.64 4945.64 4945.64 3972 4044.5 4044.5 4056.8

Platform Sequence Serpukhovian Serpukhovian Serpukhovian Serpukhovian Serpukhovian Famennian Famennian Famennian Famennian Famennian Serpukhovian Serpukhovian Serpukhovian Serpukhovian Famennian Famennian Famennian Famennian Serpukhovian Serpukhovian Visean D Visean D Visean D Visean D Bashkirian Serpukhovian Serpukhovian Serpukhovian

rim rim rim rim rim

rim rim rim rim

rim rim

Phase

Type

# incl.

Thomog (Water) (8C)

Equant CC Equant CC Equant CC Equant CC Equant CC Calcite Calcite Calcite Calcite Quartz vein Equant CC Equant CC Fluorite Fluorite Zoned CC Zoned CC Equant CC Calcite Prismatic CC Prismatic CC Calcite Equant CC Equant CC Equant CC Coarse CC Zoned CC Equant CC Equant CC

Primary Primary Primary Primary Primary Unspecified Unspecified Unspecified Unspecified Unspecified Primary Primary Primary Primary Primary Primary Unspecified Primary Primary Primary Primary Primary Primary Primary Unspecified Primary Primary Primary

2 2 1 3 2 5 6 >10 4 No assemblage 1 5 4 4 2 3 5 4 2 3 2 4 4 3 4 4 2 2

85 – 90 90 – 95 88 80 – 85 85 – 90 105 – 110 105 – 115 110 – 120 90 – 95 115 – 125 80 – 85 85 – 90 110 – 125 110 – 125 100 95 – 100 85 – 95 95 – 105 80 – 85 85 – 90 85 – 90 85 – 90 85 – 90 85 – 90 100 – 107 80 – 95 85 – 95 85 – 90

Data for hydrocarbon-water or vapor-filled inclusions: Phase refers to cement types described in the text (see Figure 24). ‘‘Calcite’’ indicates the cement phase preceding bitumen formation and, in most cases, refers to equant calcite. Prismatic calcite represents the cores of zoned cement crystals (see Figure 20), and zoned calcite represents subsequent inclusion-rich growth zones (either calcite or dolomite). Coarse calcite is a postbitumen cement phase. An inclusion type of ‘‘unspecified’’ indicates an assemblage or train whose orientation was not specified during analysis, but was not designated as cutting across growth planes and may be primary. The column ‘‘# incl.’’ indicates the number of individual inclusions in a train or assemblage.

REFERENCES CITED Bahamonde, J. R., C. Vera, and J. R. Colmenero, 2000, A steep-fronted Carboniferous carbonate platform: Clinoformal geometry and lithofacies (Picos de Europa, NW Spain): Sedimentology, v. 47, p. 645 – 664. Blendinger, W., 2001, Triassic carbonate buildup flanks in the Dolomites, northern Italy: Breccias, boulder fabric and the importance of early diagenesis: Sedimentology, v. 48, no. 5, p. 919 – 933. Brenckle, P. L., and N. V. Milkina, 2003, Foraminiferal timing of carbonate deposition on the Late Devonian (Famennian)–middle Pennsylvanian (Bashkirian) Tengiz

platform, Kazakhstan: Revista Italiana di Paleontologia e Stratigrafia, v. 109, no. 2, p. 131– 158. Cander, H., 1995, Interplay of water-rock interaction efficiency, unconformities, and fluid flow in a carbonate aquifer: Floridan aquifer system, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 103 – 124. Choquette, P. W., and N. P. James, 1987, Introduction, in N. P. James and P. W. Choquette, eds., Paleokarst: New York, Springer-Verlag, p. 1 – 21. Della Porta, G., J. A. M. Kenter, J. R. Bahamonde, A. Immenhauser, and E. Villa, 2003, Microbial boundstone

94 / Collins et al. dominated carbonate slope (Upper Carboniferous, N Spain): Microfacies, lithofacies distribution and stratal geometry: Facies, v. 49, p. 175–208. Esteban, M., and J. L. Wilson, 1993, Introduction to karst systems and paleokarst reservoirs, in R. D. Fritz, J. L. Wilson, and D. A. Yurewicz, eds., Paleokarst related hydrocarbon reservoirs: SEPM Core Worshop 18, p. 1 – 9. Ford, D. C., 1995, Paleokarsts as a target for modern karstification: Carbonates and Evaporites, v. 10, no. 2, p. 138 – 147. Heydari, E., 2000, Porosity loss, fluid flow and mass transfer in limestone reservoirs: Application to the Upper Jurassic Smackover Formation, Mississippi: AAPG Bulletin, v. 84, no. 1, p. 100 – 118. Hill, C. A., 1987, Geology of Carlsbad Cavern and other caves in the Guadalupe Mountains: New Mexico Bureau of Mines and Mineral Resources, New Mexico and Texas Bulletin 117, 150 p. Hill, C. A., 1990, Sulfuric acid speleogenesis of Carlsbad Cavern and its relationship to hydrocarbons, Delaware basin, New Mexico and Texas: AAPG Bulletin, v. 74, no. 11, p. 1685 – 1694. Hill, C. A., 1995, H2S-related porosity and sulfuric acid oilfield karst, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 301 – 313. Jones, G. D., F. F. Whitaker, P. L. Smart, and W. E. Sanford, 2004, Numerical analysis of seawater circulation in carbonate platforms: II. The dynamic interaction between geothermal and brine reflux circulation: American Journal of Science, v. 304, p. 250 – 284. Kenter, J. A. M., P. M. Harris, and G. Della Porta, 2005, Steep microbial boundstone-dominated platform margins — Examples and implications: Sedimentary Geology, v. 178, p. 5 – 30. Kenter, J. A. M., P. M. Harris, J. F. Collins, L. J. Weber, G. Kuanysheva, and D. J. Fischer, 2006, Late Visean to Bashkirian platform cyclicity in the central Tengiz buildup, Precaspian Basin, Kazakhstan: Depositional evolution and reservoir development, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir chacracterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 7 – 54. Kerans, C., and P. M. Harris, 1993, Outer shelf and shelf crest, in D. G. Bebout and C. Kerans, eds., Guide to the Permian reef geology trail, McKittrick Canyon, Guadalupe Mountains National Park, west Texas, Guidebook 26: Austin, Bureau of Economic Geology, p. 32 – 43. Kerans, C., N. F. Hurley, and P. E. Playford, 1986, Marine diagenesis in Devonian reef complexes of the Canning basin, Western Australia, in J. H. Schroeder and B. H. Purser, eds., Reef diagenesis: Berlin, SpringerVerlag, p. 357 – 380. Kirkland, B. L., J. A. D. Dickson, R. A. Wood, and L. S. Land, 1998, Microbialite and microstratigraphy: Encrustations in the middle and upper Capitan Formation,

Guadaupe Mountains, Texas and New Mexico, U.S.A.: Journal of Sedimentary Research, v. 68, p. 956 – 969. Kohout, F. A., H. R. Henry, and J. E. Banks, 1977, Hydrology related to geothermal conditions of the Floridan Plateau, in K. L. Smith and G. M. Griffin, eds., The geothermal nature of the Floridan Plateau: Florida Department of Natural Resources Bureau Geology Special Publication 21, p. 1 – 34. Lisovsky, N. N., G. N. Gogonenkov, and Y. A. Petzoukha, 1992, The Tengiz oil field in the Pre-Caspian Basin of Kazakhstan (former USSR) — Supergiant of the 1980s, in M. T. Halbouty, ed., Giant oil and gas fields of the decade 1978 – 1988: AAPG Memoir 54, p. 101 – 122. Lucia, F. J., 1995, Lower Paleozoic cavern development, collapse, and dolomitization, Franklin Mountains, El Paso, Texas, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 279 – 300. Lucia, F. J., 1999, Carbonate reservoir characterization: New York, Springer-Verlag, 222 p. Machel, H. G., 1987, Some aspects of diagenetic sulphatehydrocarbon redox reactions, in J. D. Marshall, ed., Diagenesis of sedimentary sequences: Geological Society (London) Special Publication 36, p. 15 – 38. Machel, H. G., 1989, Relationships between sulphate reduction and oxidation of organic compounds to carbonate diagenesis, hydrocarbon accumulations, salt domes, and metal sulphide deposits: Carbonates and Evaporites, v. 4, no. 2, p. 137 – 151. Machel, H. G., and J. H. Anderson, 1989, Pervasive subsurface dolomitization of the Nisku Formation in central Alberta: Journal of Sedimentary Petrology, v. 59, no. 6, p. 891 – 911. Melim, L. A., and P. A. Scholle, 1995, The forereef facies of the Permian Capitan Formation: The role of supply versus sea-level changes: Journal of Sedimentary Research, v. B65, no. 1, p. 107 – 118. Morrow, D. W., 1998, Regional subsurface dolomitization: Models and constraints: Geoscience Canada, v. 25, p. 57 – 70. Mylroie, J. E., and J. L. Carew, 1995, Karst development on carbonate islands, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 55 – 76. Playford, P. E., 1984, Platform-margin and marginal-slope relationships in Devonian reef complexes of the Canning basin, in P. G. Purcell, ed., The Canning basin, Western Australia: Geological Society of Australia and Petroleum Exploration Society of Australia, Canning basin Symposium, Perth 1984, p. 189 – 234. Ross, C. A., and J. R. P. Ross, 1987, Late Paleozoic sea levels and depositional sequences: Cushman Foundation for Foraminiferal Research Special Publication 24, p. 137 – 149. Ross, C. A., and J. R. P. Ross, 1988, Late Paleozoic transgressive-regressive deposition, in C. K. Wilgus, B. S. Hastings, C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes:

The Outer Platform, Rim, and Flank of the Tengiz Buildup / 95 An integrated approach: SEPM Special Publication 42, p. 227 – 247. Saller, A. H., 1984, Petrologic and geochemical constraints on the origin of subsurface dolomite, Enewetak Atoll: An example of dolomitization by normal sea water: Geology, v. 12, p. 217 – 220. Sanford, W. E., F. F. Whitaker, P. L. Smart, and G. D. Jones, 1998, Numercial analysis of seawater circulation in carbonate platforms: I. Geothermal circulation: American Journal of Science, v. 298, p. 801 – 828. Stephens, N. P., and D. Y. Sumner, 2003, Famennian microbial reef facies, Napier and Oscar Ranges, Canning basin, Western Australia: Sedimentology, v. 50, p. 1283– 1302. Stoessell, R. K., 1992, Effects of sulfate reduction on CaCO3 dissolution and precipitation in mixing-zone fluids: Journal of Sedimentary Petrology, v. 62, no. 5, p. 873 – 880. Tinker, S., 1998, Shelf-to-basin facies distributions and sequence stratigraphy of a steep-rimmed carbonate margin: Capitan depositional system, McKittrick Canyon, New Mexico and Texas: Journal of Sedimentary Research, v. 68, p. 1146 – 1174. Tinker, S. W., J. R. Ehrets, and M. D. Brondos, 1995, Multiple

karst events related to stratigraphic cyclicity: San Andres Formation, Yates field, West Texas, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 213– 237. Wagner, P. D., D. R. Tasker, and G. P. Wahlman, 1995, Reservoir degradation and compartmentalization below subaerial unconformities: Limestone examples from west Texas, China, and Oman, in D. A. Budd, A. H. Saller, and P. M. Harris, eds., Unconformities and porosity in carbonate strata: AAPG Memoir 63, p. 301 – 313. Weber, L. J., B. P. Francis, P. M. Harris, and M. Clark, 2003, Stratigraphy, lithofacies, and reservoir distribution, Tengiz field, Kazakhstan, in W. M. Ahr, P. M. Harris, W. A. Morgan, and I. D. Somerville, eds., PermoCarboniferous carbonate platforms and reefs: SEPM Special Publication 78 and AAPG Memoir 83, p. 351 – 394. Wendte, J., H. Qing, J. Dravis, S. L. O. Moore, L. D. Stasiuk, and G. Ward, 1998, High temperature saline (thermoflux) dolomitization of Swan Hills platform and bank carbonates, Wild River area, west-central Alberta: Bulletin of Canadian Petroleum Geology, v. 46, p. 210 – 266.

3

Lindsay, R. F., D. L. Cantrell, G. W. Hughes, T. H. Keith, H. W. Mueller III, and S. D. Russell, 2006, Ghawar Arab-D reservoir: Widespread porosity in shoaling-upward carbonate cycles, Saudi Arabia, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/ SEPM Special Publication, p. 97 – 137.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles, Saudi Arabia Robert F. Lindsay, Dave L. Cantrell, Geraint W. Hughes, Thomas H. Keith, Harry W. Mueller III, and S. Duffy Russell Saudi Aramco, Dhahran, Saudi Arabia

ABSTRACT

G

hawar field is the world’s largest, most prolific field, producing 30–318 API oil from the Arab-D carbonate reservoir. The field is more than 250 km (155 mi) long, as much as 30 km (18.5 mi) wide, and has more than 300 m (1000 ft) of structural closure. The Arab-D reservoir, limestone with some dolostone horizons, stratigraphically comprises the D member of the Arab Formation and the upper part of the Jubaila Formation. Based on ammonite and benthonic foraminiferal evidence, the reservoir formations are Upper Jurassic, Kimmeridgian, in age. The reservoir has an average thickness of more than 60 m (200 ft), an average porosity of more than 15%, and a permeability up to several darcys. The upper half of the reservoir is dominated by exceptionally high reservoir quality; the lower half contains interbeds of high and relatively lower reservoir quality. Early correlation of well logs and cores, before the advent of sequence stratigraphy, subdivided the reservoir from top to base into zones 1, 2, 3, and 4. Zones 2 and 3 have been subsequently subdivided into zones 2a and 2b and zones 3a and 3b, with a more detailed zonation scheme for reservoir management. The reservoir is composed of two composite sequences. One composite sequence is the Arab-D Member of the Arab Formation, with the upper boundary at the top of Arab-D carbonate and below the C-D evaporite, with the sequence boundary locally marked by pods of collapse breccia. The second composite sequence forms the upper part of the Jubaila Formation, for which the sequence boundary between the Arab-D Member and Jubaila Formation is located in zone 2b and is marked by a flood of slightly deeper water cycles over the more graindominated cycles in upper zone 3a and lower zone 2b. Several high-frequency sequences (HFSs) have been identified, each comprising several cycle sets (parasequence sets). Each cycle set is composed of approximately five individual carbonate cycles (parasequences), and each cycle is composed of one to three beds. Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215875M88576

97

98 / Lindsay et al.

These carbonates were deposited approximately 58 south of the equator on a broad, arid, storm-dominated carbonate ramp. From upslope to downslope, the ramp consisted of the following subenvironments: (1) inner ramp; (2) ramp-crest shoal; (3) proximal middle ramp; (4) distal middle ramp; and (5) outer ramp. The inner ramp was a lagoon with localized intertidal islands composed of grainstones and packstones and a highly diverse, shallow-marine benthonic foraminiferal microfauna. The distal or seaward part of this regime consists of packstones characterized by dasyclad and encrusting algae. The ramp-crest shoal is composed of skeletal and oolitic grainstone, mud-lean packstone, and some mud-rich packstone. Skeletal sands of micritized foraminiferal tests and broken skeletal detritus also include larger fragments of transgressive and storm-derived stromatoporoids and corals. The proximal middle ramp is composed of domed and encrusting stromatoporoidcoral mounds and intermound sheltered areas dominated by branched stromatoporoids. The distal middle ramp, deposited below fair-weather-wave base, is composed of micritic to very fine-grained sediment capped by Thalassinoides firmgrounds. These firmgrounds are overlain by storm-derived rudstone and floatstone of innerramp, ramp-crest, and proximal middle-ramp bioclasts. The outer ramp is composed of deeper shallow-marine deposits of micritic to very fine-grained sediment capped by Thalassinoides firmgrounds. In this setting, smaller, benthonic foraminifera are common along with tetraxon and triaxon sponge spicules and coccoliths. From highest to lowest reservoir quality, the lithofacies or rock types consist of (1) skeletal-oolitic grainstone, mud-lean packstone, and some mud-rich packstone; (2) stromatoporoid-red and green algae-coral rudstone and floatstone; (3) Cladocoropsis rudstone and floatstone; (4) dolomite, porous and locally extremely permeable to nonporous; (5) bivalve-coated grain-intraclast rudstone and floatstone; and (6) micritic to very fine-grained deposits. Limestone porosity is a mixture of the following common pore types: interparticle (dominant), moldic (common), intraparticle (common), and microporosity. Less common is porosity associated with Thalassinoides burrows, with vertically oriented tunnels filled by grain-rich sediment. Shelter porosity is uncommon. Dolostone porosity, less common than the major limestone pore types, is a mixture of moldic, intercrystalline, and (least common) intracrystalline porosity. Fractures (least common) do not contribute much porosity but contribute permeability. Diagenetic effects common within Arab-D reservoir carbonates include several dissolution events, recrystallization, and physical compaction. Cementation, episodes of dolomitization, and chemical compaction-stylolitization, although locally important, were less abundant events. The vertical seal for the reservoir is the overlying Arab C-D anhydrite. It is more than 30 m (100 ft) thick and is composed of varvelike laminae to very thin beds of anhydrite (thicker) and carbonate or organic matter (thinner) deposited in a salina. The salina periodically shallowed upward into peritidal and intertidal settings. A few porous-permeable carbonate stringers were deposited when relative sea level rise flooded the evaporitic shelf and temporarily restarted the subtidal carbonate factory, whereas relative sea level fall reestablished the subtidal brine factory and precipitated more evaporites.

INTRODUCTION The Arab-D reservoir of Saudi Arabia is the most prolific oil-producing interval in the world (Bates, 1973; Beydoun, 1988; 1998; 1991; Durham, 2005). Oils were

derived from thermally mature, organic-rich carbonate source rock of the Jurassic-age Hanifa and/or Tuwaiq Mountain Formations (Ayres et al., 1982; Droste, 1990) that subsequently migrated into highly porous and permeable carbonate reservoir rocks, Arab-D Member

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 99

FIGURE 1. Generalized Late Jurassic stratigraphy and lithologies in Saudi Arabia, with Arab and Hith reservoirs named. Modified from Powers (1968) and Meyer et al. (1996).

of Arab Formation and upper Jubaila Formation, in large structural traps (Figure 1). Overlying highly efficient evaporite seals, the Arab C-D anhydrite and, ultimately, the Hith anhydrite, prevented further vertical migration and ensured the containment of oil in these large structural traps. Early correlation of well logs, before the advent of sequence stratigraphy, subdivided the Arab-D reservoir in Ghawar field from top to base into zones 1, 2, 3, and 4. Zones 2 and 3 have been subsequently subdivided into zones 2a and 2b and zones 3a and 3b, respectively, with a more detailed zonation scheme for reservoir management. Arab-D Member of the Arab Formation consists of zone 1, zone 2a, and upper zone 2b, and the remainder of the Arab-D reservoir, consisting of lower zone 2b, zones 3a and 3b, and zone 4, is part of the upper Jubaila Formation (Meyer et al., 2000).

DISCOVERY OF OIL IN ARABIA The first wildcat well drilled in Saudi Arabia was on Dammam dome, a surface structure with four-way closure visible from the island of Bahrain, some 40 km (25 mi) away. Saudi Aramco Dammam 4 discovered oil in Arab Formation carbonates on March 4, 1938. Other surface structures were identified as additional wildcat drilling sites in eastern Saudi Arabia. However, some surface structures were drilled (e.g., Ma’agala and El Alat), only to find they were false structures created by differential solution of lower Eocene

Rus evaporite beds. To address this problem, a shallow drilling program (generally less than 300 m [1000 ft]) collected stratigraphic and structural information beneath the pre-Neogene unconformity, drilling to the brown crystalline limestone of Eocene age beneath the Rus evaporite. Shallow drilling identified true subsurface structures from false structures and outlined an anticlinal trend that became known as Abqaiq, southwest of Dammam dome. Saudi Aramco Abqaiq 1 was drilled in 1940 and discovered oil in porous Arab Formation carbonates. At the same time, field mapping near the Rub’Al Khali desert Empty Quarter, 320 km (200 mi) southwest of Dammam dome, identified Wadi Sahaba, a westto east-oriented wadi that abruptly made a turn to the south before continuing on to the east. Dips taken throughout the area revealed a large, broad, low-relief dome that was named the ‘‘Haradh feature.’’ This was the first indication of the southern tip of the large En Nala anticline, the structural feature on which the shallowest crest of the Ghawar field was ultimately found. As the shallow drilling program continued, it revealed a structural closure at Ain Dar, west of Abqaiq and, ultimately, through 1941, connected that closure with a series of closures all the way to the Haradh closure along the En Nala anticline. In mid-1948, Saudi Aramco Ain Dar 1 established production from the D member of the Arab Formation and the Jubaila Formation and raised the possibility that all of the structurally high areas along the En Nala anticline could be productive. This possibility was confirmed in late 1948 by the successful completion of Saudi Aramco Haradh 1, some 200 km (120 mi) farther south. Between Ain Dar and Haradh, the next closed structure to be drilled was in an area called ‘‘Ghawar,’’ referring to an area below an escarpment that dominated the area. The wildcat, named ‘‘Ithmaniya’’ well 1, now called ‘‘Uthmaniyah well 1,’’ after a small djebel east of the

100 / Lindsay et al.

FIGURE 2. Generalized geologic map of the Arabian Peninsula and the position of the central Arabian arch. Modified from Al-Hinai et al. (1997) and U.S. Geological Survey (1963).

staked location, was yet another discovery in the Arab-D and upper Jubaila. Another structural closure east of Ain Dar, now called Shedgum, produced an additional discovery in the Arab-D and upper Jubaila in 1952. South of Uthmaniyah, the Saudi Aramco Huiyah 1 wildcat, later changed to Hawiyah, was drilled and became yet another discovery in the Arab-D and upper Jubaila. Between 1954 and 1957, successful wildcat and production wells left little doubt that all of the discovery areas, Ain Dar, Haradh, Uthmaniyah, Shedgum, and Hawiyah, were part of one large oil field. To simplify logistics, each area name was used for well names, such as, Ain Dar well 1, Haradh well 1, etc., whereas the entire field became known as Ghawar. Basic tools used to discover Ghawar were fairly simple and involved the Brunton compass, alidade, and plane table for surface mapping; barometer to determine altitude; a shallow drilling program; and good geologic insight. Both the historical outline and the techniques used are expanded by Stegner (1971), Barger (2000), Keith (2005), and T. A. Pledge (2006, personal communication).

STRUCTURAL SETTING The main oil-producing area of Saudi Arabia, including Abqaiq and Ghawar fields, is located on the northeastern part of the central Arabian arch (Figure 2).

To the west is the Arabian shield, a vast complex of Precambrian igneous and metamorphic rocks with some younger igneous rocks (Pollastro et al., 1999). Bordering the shield to the east are interior escarpments where long arcuate belts of Paleozoic, Mesozoic, and lower Tertiary rocks crop out and dip basinward (eastward) at about 18 or less (Powers, 1968). In eastern Arabia, almost flat-lying Tertiary and younger sediments of the interior or Arabian platform effectively cover and mask these older sediments (Powers et al., 1966). The north-northeast-trending En Nala anticline, the structural trap for Abqaiq and Ghawar fields, is dominated by mostly north – south to north-northeast – south-southwest-trending anticlinal trends and flexures that reflect deep-seated basement faults (Figure 3) (Ayres, et al., 1982; Al-Husseini, 2000). Drilling and seismic evidence (Wender et al., 1998; Konert et al., 2001) supports the presence, at depth, of large faults bounding basement blocks under the Ghawar structure that extend from Precambrian metasedimentary basement up through Paleozoic sedimentary rocks and die out in the Mesozoic section (Konert et al., 2001). Broad flexing of post-Paleozoic sedimentary rock is the dominant structural style in eastern Arabia (Powers, 1968; Konert et al., 2001) and is exemplified by the En Nala anticline (Powers et al., 1966), the structural trap for Ghawar and Abqaiq fields. Although there is some evidence of subtle growth of the En Nala structure at least as early as the Permian, growth was most pronounced in the Late Cretaceous, with regional compressive-transpressive stresses being imposed on eastern Arabia in response to the closing of the Neo-Tethys Ocean (Beydoun, 1991; Nicholson, 2000, 2002). The most recent pulses of structural growth occurred during the late Eocene–Oligocene

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 101

FIGURE 3. Major tectonic features of the Arabian plate and Iran. Ghawar field is the large green oil field in the center that trends north-northeast. Modified from Konert et al. (2001).

sion-transtension associated with the continuing vertical growth of these anticlines. In core, oblique to vertically oriented stylolites are evidence of postlithification compressiontranspression associated with the vertical fractures.

STRATIGRAPHIC NOMENCLATURE AND AGE

and during the late Miocene–Pliocene, as the Arabian plate welded onto the Eurasian plate along the Zagros suture or crush zone (Beydoun, 1991; Glennie, 2000) (Figure 3). Minor faulting, fracturing, and jointing are found in the younger formations, especially along crests of large flexures where two vertical joint sets are recognized, including a dominant north-northeast to northeast trend and a subsidiary north-northwest to northwest joint set (Hancock et al., 1984). Image logs suggest that there are also east – west-trending fractures (K. A. T. MacPherson, 2005, personal communication). These features may result from exten-

The uppermost Jubaila Formation, or Ju2 Member (Enay, 1987) as defined in outcrop where the Jurassic section was first described and named, is correlative with the middle and lower part of the Arab-D reservoir in the subsurface (Powers et al., 1966; Powers, 1968) (Figures 1, 4). In outcrop, the top of the Jubaila Formation is placed immediately above the highest occurring stromatoporoids, here interpreted to mean encrusting and domed stromatoporoids, almost invariably not in original growth position, and at the base of the overlying rippled calcarenite (D. Vaslet, 2002, personal communication to G.W. Hughes) (Figure 4). The upper part of the Arab-D reservoir is composed of the D member of the Arab Formation. The Arab-D reservoir is the thickest and oldest of the four reservoirs in the Arab Formation and is overlain by the C-D anhydrite, the upper unit of the Arab-D Member. Ages for the Arab-D – Jubaila succession are well constrained, with the Jubaila Formation considered to be early Kimmeridgian (Arkell, 1952), whereas the overlying Arab Formation is thought to be late Kimmeridgian to early Tithonian (LeNindre et al., 1990; De Matos and Hulstrand,

102 / Lindsay et al.

FIGURE 4. Jubaila Formation west of Riyadh in Wadi Leban. Ju2 member is the upper, resistive beds. Ju1 member is the lower, recessive beds. Ju2 cliff former is 15 m (50 ft) thick.

1995; Hughes, 1997, 2004). The D member is restricted to the Kimmeridgian (Sharland et al., 2001). Although numerous benthonic foraminifera with stratigraphic ranges consistent with a Kimmeridgian age have been described from the Arab-D reservoir, none establishes a more precise date.

PALEOGEOGRAPHIC SETTING AND REGIONAL DEPOSITIONAL ENVIRONMENTS During the Late Jurassic, a vast, shallow carbonate ramp extended from the central Arabian Peninsula, west of the present site of Riyadh, as far east as the present Zagros Mountains of Iran, as far north as central Iraq and south into Oman (Figure 5). A thick succession of carbonate and evaporite sediments was deposited on this shelf (Figure 1). Differential intraplate subsidence, partly enhanced by peripheral aggradation of shallow-marine carbonates, led to the development of intrashelf basins, including the Gotnia, Arabian, and Rub’Al Khali basins (Al-Husseini, 1997; Ziegler, 2001), during deposition of the Callovian Tuwaiq Mountain and Oxfordian Hanifa formations. Organic-rich carbonate source rocks were deposited in these intrashelf basins during part of the Oxfordian (Ayres et al., 1982; Droste, 1990; Carrigan et al., 1994). These intrashelf basins tended to be maintained through the remainder of the Jurassic, influencing the deposition of the overlying Jubaila, Arab, and Hith formations (Figure 6) and bounding broad, relatively stable shallow shelf or platform areas on which the Arab reservoir carbonates were deposited. The peripheral highs around the intrashelf basins are interpreted to have been the loci for the initiation of

stromatoporoid and coral banks and overlying ramp-crest skeletal-oolitic grain shoals, which subsequently prograded over much of the area of the intrashelf basins. Overall, the paleoclimate was probably hot and arid, similar to the present climate on the Arabian Peninsula (Handford et al., 2002). The general distribution of evaporites in the Hith Formation, which formed in response to a restriction of the shelf, and especially of intrashelf basins, from open-marine circulation, was also influenced by these intrashelf basins; evaporites are typically thickest and most halite prone in the intrashelf basins.

ARAB-D LITHOFACIES AND DEPOSITIONAL ENVIRONMENTS The interpretations presented herein are based primarily on observations from core in the Shedgum area of Ghawar field (Figures 7, 8 [the foldout located in the back of this publication]). Previous workers (Mitchell et al., 1988) have developed a classification scheme to organize Arab-D rocks into genetically meaningful packages. This classification divides Arab-D rocks into six depositional lithofacies, which include one anhydrite and five carbonate lithofacies. Carbonate lithofacies are distinguished on the basis of their typical depositional components and include (1) skeletal-oolitic limestones and dolomites (SO); (2) Cladocoropsis limestones and dolomites (CLADO); (3) stromatoporoid-red (green) algae-coral limestones (SRAC) (In this lithofacies, a common organic component is Thaumatoporella parvovesiculifera (Hughes, 1996), a microbial encruster that De Castro (1991, 2002) suggests is a chlorophycean, or green alga. The encrusting form gives it the appearance, both megascopically and under hand lens or binocular microscope, of a red alga and was originally described as a red alga. The ‘‘SRAC’’ acronym is sufficiently widely present in the literature that we continue to use it despite the fact that few red algal fragments actually exist in the lithofacies.); (4) bivalve-coated

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 103

FIGURE 5. Late Jurassic paleogeography of the Arabian Peninsula and surrounding area. Average directions of Arab-D and Jubaila progradation are shown by arrows. Blue band west of Riyadh is the Tuwaiq escarpment. Modified from Handford et al. (2002), Al-Husseini (1997), Ayres et al. (1982), R. B. Koepnick and L. E. Waite (1991, personal communication), Murris (1980), and Scotese (1998). grain-intraclast limestones (BCGI); and (5) micritic limestones and dolomites (MIC). Because dolomitization frequently destroys evidence of original depositional lithofacies, the diagenetic lithofacies dolomite (DOLO) was added (Mitchell et al., 1988) (Figure 9). Although some later workers have sought to revise this classification (Meyer and Price, 1993; Meyer et al., 1996; Handford et al., 2002), these later modifications were not widely accepted and have since fallen out of use. These lithofacies, with the exception of the diagenetic dolomite lithofacies (DOLO), were deposited approximately 58 south of the equator on a broad, arid, storm-dominated carbonate ramp. The ramp consisted of the following subenvironments from upramp to downramp: (1) inner ramp; (2) ramp-crest shoal; (3) proximal middle ramp; (4) distal middle ramp; and (5) outer ramp (Figure 10). Each of these environments is characterized by one or two of the depositional lithofacies. In addition, there was a salina environment that developed after the deposition of the sediments of the Arab-D reservoir, depositing the dominantly anhydrite interval that is the vertical

seal of the reservoir. Because a close relationship exists between the depositional environments and the lithofacies, both will be discussed together below.

Anhydrite Lithofacies; Salina Environment Anhydrite (nonporous) with some interbedded carbonate stringers (some of which, especially where dolomitized, are porous) makes up most of the C-D anhydrite and overlies the Arab-D reservoir (Figure 8, the foldout located in the back of this publication). Typical anhydrite fabrics include nodular, bedded nodular, nodular mosaic, and massive (classification of Maiklem et al., 1969); abundant, vertically aligned (elongate) and uncommon mosaic anhydrite are also observed (Figures 11, 12). In thin section, anhydrite locally exhibits a felted replacive texture of very small, oriented, needlelike anhydrite crystals and clear, coarser, rectangular laths of diagenetic anhydrite. On the crest of the En Nala anticline in Ghawar field, capillary pressure has been strong enough to force light-end hydrocarbons, which look like wispy films, between evaporite crystals and nodules.

104 / Lindsay et al.

FIGURE 6. Regional depositional environments of the Arab and Hith formations. Most of the time, deposition of highenergy grainstones and packstones dominated the broad carbonate shelf. This was succeeded by later widespread deposition of Hith evaporites. Modified from Ziegler (2001).

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 105

FIGURE 7. Shedgum Arab-D type log (open-hole log gamma-ray density-neutron), Ghawar field. Six cored intervals are shown to the right, with a total of 91 m (299 ft) cored and depth adjusted to match open-hole logs. Core coverage is through the complete Arab-D reservoir.

Deposition of the anhydrite occurred after a major fall of relative sea level that subaerially exposed the Arab-D carbonate, marking the top of the Arab-D carbonate. Small sink holes, filled with collapse breccia, have been seen in outcrop (Tuwaiq escarpment) and in Ghawar field in core and on image logs (Figure 13) where anhydrite overlies and fills in the karstic features. Following subaerial exposure, the initial rise of sea level was slow, and the regional carbonate shelf and associated interior basins, such as the Arabian intrashelf basin, became the site of evaporite deposition (probably a combination of mostly gypsum and some anhydrite) (Figures 11, 12). Historically, nodular mosaic to massive anhydrite, typical of the C-D anhydrite, has commonly been interpreted as forming within a sabkha environment (Kinsman, 1964; Shearman, 1978). However, vertically aligned crystals, thought to have been deposited originally as subaqueous gypsum palmate crystals, and bedded anhydrite nodules in the C-D anhydrite suggest that most of the anhydrite probably formed subaqueously in a salina to peritidal to intertidal setting (McGuire et al., 1993; Handford et al., 2002) (Figures 11, 12). Recent detailed descriptions of the anhydrite in two cored wells show that the sediments were deposited in cycles, with each evaporitic cycle starting with subaqueous salina deposition and growth of palmate crystals. Slight shallowing to a peritidal position between the salina proper and the intertidal resulted in palmate crystals growing at an oblique angle. Subsequent intertidal environments produced flat to obliquely oriented gypsum crystals with laminated carbonate mud. Salina and peritidal settings are the most common, with the intertidal setting as the least common. The laminae to very thin beds that are the most common bedding in the evaporite appear to be varvelike and form evaporite couplets that contain thicker evaporite bases, interpreted as summertime deposition, and carbonate- and organic-rich laminae caps, interpreted as wintertime deposition. Uncommon, large transgressive events freshened the highly saline setting to normal-marine salinity, which allowed the subtidal carbonate factory to start up and deposit a cycle of carbonate referred to as a carbonate stringer. Sometimes, these only formed very thin to thin beds, but occasionally deposited a carbonate cycle 1.5–2 m (5–6 ft) thick, with the top portion porous and grain rich. During the following overall relative sea level fall, restriction and evaporation recommenced, and the subtidal evaporite factory again started to deposit subaqueous evaporites.

106 / Lindsay et al.

FIGURE 9. Summary of characteristics and distribution of carbonate lithofacies in the Arab-D reservoir, Ghawar field. Modified from Mitchell et al. (1988).

Inner Ramp The inner ramp, landward and upslope from the landwardmost extent of sediment agitation by significant wave activity and behind the shoal crest, was a lagoon with localized intertidal islands composed of wackestone, packstone, and some grainstone (Figure 10). The sediments of the inner-ramp lagoonal setting are only present at the top of the Arab-D reservoir (zone 1) and are approximately 3 m (10 ft) thick (Figure 8, the foldout located in the back of this publication). This depositional setting contains a restricted fauna, suggesting a lack of normalmarine circulation. Normal-marine fauna, such as the delicate stick-shaped stromatoporoid Cladocoropsis, have been described locally within this setting and may owe their presence to either storm transport or to

a normal-marine passageway, through the ramp-crest shoal, that connected part of the inner ramp with the proximal part of the middle ramp.

Skeletal-oolitic Lithofacies (SO); Ramp-crest Shoal Environment Skeletal-oolitic (SO) limestone and dolomitic limestone is present in the upper part of the Arab-D reservoir (zone 2a) and another interval deeper in the reservoir (zone 2b) (Figure 9). This lithofacies is typically well sorted and contains grainstone, mud-lean packstone, and some mud-rich packstone (Figures 8 [the foldout located in the back of this publication] and 14). Overall, the lithofacies represent the best sorted and, at the megascopic visual level, the most uniformappearing carbonate rock type of the Arab-D. Major

inner ramp, ramp-crest shoal, proximal, and distal middle ramp to outer ramp. Blue is a transgressive mud-rich deposit capped by a firmground that developed during maximum flooding and was burrowed by Thalassinoides. Brown with diamond shapes is distal middle-ramp debris-flow deposits of rudstone and floatstone. Green is proximal middle-ramp biostromes and mounds of stromatoporoids and corals. Blue with fork shapes is proximal middle-ramp Cladocoropsis banks deposited in sheltered areas between and upslope of biostromes and mounds. Red is the cross-bedded ramp-crest grainstone shoal. Purple is inner-ramp lagoonintertidal islands. Those facies and rock types that are abundant in each reservoir zone are underlined by the titles and associated arrows zone 1, zone 2, and zone 3. Distribution of the common to abundant faunal and floral elements (both laterally at any one time and stratigraphically through the reservoir zones) is shown below the figure.

FIGURE 10. Idealized depositional model of the Arab-D reservoir carbonate ramp in Ghawar field. This model depicts deposition of one carbonate cycle from the

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 107

108 / Lindsay et al.

FIGURE 11. Arab C-D evaporite from 2071 – 2073 m (6796 – 6804 ft) in top of core 1 in the Shedgum Arab-D core, Ghawar field. Evaporite crystals were originally gypsum that grew on the floor of a salina. Note the varvelike cyclicity and enterolithic folds. grain types include foraminifera, dasycladacean algae, micritized grains and foraminifera, ooids, and bivalves (Figures 14 – 16). Porosity, from most abundant to least abundant, includes interparticle, moldic, intraparticle, microporosity, and, locally common within dolostones, intercrystal pores (Figures 14–16). Crossbedding and horizontal laminations are common (Figures 8 [the foldout located in the back of this publication] and 14). Other common features include firmgrounds and hardgrounds (Figure 8, the foldout located in the back of this publication), burrows, borings, and coarsening-upward beds. Common types of diagenesis include dissolution and recrystallization, fine isopachous bladed calcite and microspar cementation, and dolomitization. Although scattered dolomite crystals are commonly found, there does not appear to be any complete dolomitization of this lithofacies. The skeletal-oolitic lithofacies were deposited in the ramp-crest shoal environment, consisting of a series of ooid-peloid-intraclast sand shoals that were locally and briefly subaerially exposed as islands in a high-energy very shallow subtidal to intertidal setting in fair-weather-wave base. It was continuously wave washed and occasionally storm swept by the highest

day-to-day energy to have affected the carbonate ramp (Figure 10). Fair-weather-wave base is defined as a zone where waves first touched the sea floor and continued to wash the sea floor with high energy, until friction reduced wave energy, and sediment movement ceased (zone Y of Irwin, 1965). The grain-rich setting is crossbedded and slightly cemented by fine, isopachous, marine, and microspar cement. Skeletal detritus includes storm-derived fragments and larger to nearly whole Cladocoropsis, stromatoporoids, and corals from farther seaward. Microbial encrustations are fairly common. The ramp-crest shoal depositional setting and associated skeletal-oolitic lithofacies form the upper Arab-D reservoir (most of zone 2a) (Figures 8 [the foldout located in the back of this publication] and 10), as well as the uppermost part of the lower sequence (uppermost Jubaila Formation).

Cladocoropsis Lithofacies (CLADO); Sheltered Proximal Middle-ramp Environment Cladocoropsis (CLADO) limestone and dolostone contain greater than 10% of the branching stromatoporoid Cladocoropsis mirabilis (Champetier and Fourcade, 1966) and occur in the upper and middle Arab-D

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 109

FIGURE 12. Arab C-D evaporite from 6804 to 6809 ft (2073 to 2075 m) in top of core 1 in the Shedgum Arab-D core, Ghawar field. Evaporite crystals were originally palmate gypsum that grew on the floor of a salina. Note varvelike cyclicity. Carbonate at the base has been dolomitized. Cores are 2.5 in. (6.35 cm) wide.

reservoir (zones 2a and upper 2b) (Figures 8 [the foldout located in the back of this publication] and 17). Cladocoropsis rudstones and floatstones contain a matrix of grainstone, mud-lean packstone, and mud-rich packstone. They are not as well sorted as those of the skeletal-oolitic lithofacies and are therefore more heterogeneous (Figure 17). Cladocoropsis fragments, dasycladacean algae, foraminifera, and micritized grains and forams are common (Figure 17). Pore types, from most common to least common, are interparticle, intraparticle, moldic, and microporous, with all types abundant (Figure 17), and when dolomitized, intercrystal and moldic porosity are the dominant pore types. Sedimentary structures are subtle, but close examination of cores shows that crossbedding and horizontal laminations are present in highstand parts of cycles (parasequences), whereas soft-body bioturbation and firmgrounds and hardgrounds are more common in transgressive parts of cycles (Figure 8, the foldout located in the back of this publication). Diagenesis is commonly limited to dolomitization, dissolution, recrystallization, and uncommon equant calcite cementation. When completely

dolomitized, Cladocoropsis fragments typically undergo dissolution to form moldic porosity and, if originally in contact with each other, develop super-K (high-permeability) intervals. The Cladocoropsis (CLADO) lithofacies, along with the stromatoporoid-red algaecoral (SRAC) lithofacies, described below, was deposited in a proximal middle-ramp environment, with deposition just within fair-weatherwave base, where constant daily circulation occurred. This environment was characterized by grain-rich, domed, and encrusting stromatoporoid and coral biostromes and mounds. Intermound sheltered areas and sheltered areas immediately upslope of the biostromes and mounds were populated by Cladocoropsis (Figures 8 [the foldout located in the back of this publication] and 10) (Leinfelder et al., 2005). This interpretation is based, in part, on comparison with domed and branched corals in modern environments ( James, 1983), where domed forms occur in biostrome and mound environments with moderate wave energy and low rates of sedimentation, whereas branched forms, such as Cladocoropsis, occupy lower energy sites with higher rates of sedimentation. Cladocoropsis is commonly accompanied by the foraminifera Kurnubia palastiniensis and Nautiloculina oolithica (Figures 15, 16). In the shallower marine moderate-energy SRAC lithofacies, the microbiota is dominated by a dasyclad alga attributed to Clypeina sulcata; the green alga T. parvovesiculifera is common and is joined by a background population of the forams Kurnubia sp., Nautiloculina sp., Redmondoides lugeoni, and some biserial agglutinated forms (Figures 15, 16). Microbial

110 / Lindsay et al.

FIGURE 13. A. Examples of breccia-filling sinkholes in the Arab-D in Ghawar field and in outcrop. The upper photograph is a small sinkhole in the Arab Formation filled with fitted-fabric and chaotic collapse breccia in Wadi Birk, 150 km (93 mi) south of Riyadh. Red baseball cap is at the base of the breccia, and each geologist stands at the side of the breccia with rock hammer (left) and hand (right) on the breccia-country rock contact. B. Formation MicroImager image log in a horizontal well in the south part of Ghawar, with rounded chaotic breccia just beneath the Arab-D composite sequence boundary and overlying Arab C-D evaporite (left). (Scale is in feet.) C. Photograph from the central part of Ghawar in a vertical well. Rounded chaotic breccia has red and brown (terra rosa) geopetal between breccia clasts, along with plant rootlets. Overlying Arab C-D evaporite contains palmate gypsum crystal ghosts. (Photo is 6.4 cm [2.5 in.] wide.)

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 111

tified Shuqria, an earlier relative of Cladocoropsis, to have populated sheltered intermound areas between stromatoporoid and coral biostromes and mounds. Cladocoropsis probably populated a similar depositional setting. Storm events reworked Cladocoropsis and have strewn them within the proximal middle ramp, and more powerful storms reworked fragments back into the ramp crest (Figure 10).

Stromatoporoid-Red Algae-Coral Lithofacies (SRAC); Biostrome-mound Proximal Middle-ramp Environment

FIGURE 14. Ramp crest shoal skeletal-oolitic lithofacies composed of cross-bedded, intraclast-peloid-coated grainooid grainstone with abundant interparticle porosity and some moldic, intraparticle, and microporosity. Cementation was by very fine-bladed calcite and microspar cement. Note minor physical and chemical compaction in the upper and more prevalent compaction in lower photomicrographs. Photomicrograph widths 1.1 mm (upper) and 2.25 mm (lower), Shedgum area.

encrustations are also common (Figure 8, the foldout located in the back of this publication). Considerable facies variability exists within and between the biostromes and mounds versus sheltered parts of this depositional environment. These two lithofacies contain diverse, normal-marine biota, which include common to abundant foraminifera, algae, and microbial encrustations, along with marine cement, which are evidence of deposition under normalmarine conditions. Fairly common cross-bedding also argues for a moderate to potentially high-energy setting. Hermatypic corals imply deposition in the photic zone, whereas the presence of such stenotopic organisms (organisms that have a limited tolerance for changes in environmental conditions) as coral indicates normal-marine conditions. Cladocoropsis have also been reworked with no inplace, upright-standing organisms being identified and very few actual intact branches being preserved. Fieldwork in the Tuwaiq Mountain Formation outcrops in the Tuwaiq escarpment, west of Riyadh, iden-

Stromatoporoid-red algae-coral (SRAC) limestones and dolomitic limestones occur in the middle Arab-D reservoir (zone 2b) and represent the most heterogeneous rock types (Figures 8 [the foldout located in the back of this publication] and 18). Locally, this lithofacies is present in the lower Arab-D reservoir as allochthonous debris-flow deposits (Figures 8 [the foldout located in the back of this publication] and 10). This lithofacies is, in general, extremely poorly sorted and contains both pores and grains (Figure 18) that exhibit a wide range of sizes and shapes. Common grain types include domed, digitate, and encrusting stromatoporoids, corals, foraminifera, micritized grains, and microbial encrustations (Figure 18). Scleractinian corals found in the Arab-D reservoir were originally composed of metastable aragonite and were susceptible to dissolution during diagenesis (Figure 18) and form large, irregular molds. Stromatoporoids and corals form floatstone to rudstone fabrics (Figure 18), with a matrix of grainstone, mud-lean packstone, and mud-rich packstone. Moldic and intraparticle pores are most common in this lithofacies, with interparticle pores subordinate and micropores common. Typical diagenetic modifications include dissolution, bioerosion, recrystallization, dolomitization, anhydrite emplacement, fine isopachous bladed calcite and equant calcite cementation, and microspar cementation. High-energy storm events buffeted biostromes and mounds and reworked them locally, with stronger storms reworking bioclasts and grains downslope as debris flows into the distal middle-ramp setting (Figure 10) and commonly, but less abundantly, transporting fragments upslope into the ramp-crest and inner-ramp environments. The former effect is seen in outcrops of the Jubaila Formation in the Tuwaiq escarpment west of Riyadh where stromatoporoid and coral biostromes and mounds are virtually all reworked. Only one in-place stromatoporoid-coral biostrome, 1 m (3 ft) thick and 30 m (100 ft) in width (Figure 19), has been identified in Wadi Hilwah (Hughes, 2004). All other

112 / Lindsay et al.

FIGURE 15. Shedgum area Arab-D and Jubaila significant biocomponents. Width of view is in millimeters. 1 = brachiopod valves (2 mm); 2 = cerithid gastropod (4 mm); 3 – 4 = Trocholina palastiniensis (1 mm); 5–6 = Quinqueloculina sp. (1 mm); 7 – 8 = Pfenderina salernitana (1 mm); 9 – 10 = Mangashtia viennoti (1 mm); 11 = Alveosepta jacardi (1 mm); 12 – 13 = Redmondoides lugeoni (1 mm); 14 = Nautiloculina oolithica (1 mm); 15 = Kurnubia palastiniensis (1 mm); 16 – 17 = Clypeina sulcata (1 and 2 mm); 18 = Heteroporella jaffrezoi (1 mm); 19 = coral (8 mm); 20 = Cladocoropsis mirabilis (4 mm); 21 = Thaumatoporella parvovesiculifera (2 mm); 22 = tetraxon sponge spicule; 23 –24 = Lenticulina sp. (1 mm).

outcrops show stromatoporoids and corals to have been at least locally reworked in the proximal middle ramp, or they have been reworked farther downdip into the distal middle ramp (Figure 10).

Bivalve-coated Grain-intraclast Lithofacies (BGCI); Distal Middle-ramp Environment Bivalve-coated grain-intraclast (BCGI) limestone and dolomitic limestone occur mostly in the lower Arab-D reservoir (zones 3a, 3b, and 4) as rudstones and floatstones, with grainstone, mud-lean packstone, mud-rich packstone and uncommon wackestone matrix (Figures 8 [the foldout located in the back of this publication] and 20). Bivalves, coated grains, intraclasts, micritized grains, and foraminifera are dominant grain types. Porosity is dominantly interparticle

(Figure 20) and moldic, commonly from dissolution of metastable bivalves, but includes a significant admixture of microporosity and intraparticle pores. Diagenetic modifications noted in this lithofacies include dissolution, recrystallization, dolomitization, stylolitization, and equant calcite cementation. The BCGI lithofacies, along with the micritic lithofacies described below, are the typical sediments of the distal middle ramp, where deposition occurred beneath and seaward of fair-weather-wave base. The dominant sediment is micritic to very fine-grained sediment that is capped by firmgrounds (Figure 20) characterized by Thalassinoides burrow systems (Figure 10) and overlain by storm-derived rudstone and floatstone of inner-ramp, ramp-crest, and proximal middle-ramp bioclasts, most commonly, the BCGI lithofacies but locally including stromatoporoids and corals (SRAC) (Figure 20). Burrowing organisms (infauna) were common, and either these created vertically oriented Thalassinoides burrows in micritic to very fine-grained muds, or softbody burrowers have homogenized the sediment. Preserved sedimentary structures are uncommon, other than those created by bioturbation. These factors

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 113

FIGURE 16. Biozonation and key biocomponents of the Arab-D Member of the Arab Formation and Jubaila Formation in the Shedgum Arab-D core, Ghawar field.

FIGURE 17. Proximal middle-ramp Cladocoropsis rudstone with a packstone matrix containing interparticle, moldic, intraparticle, and microporosity. All Cladocoropsis fragments have been reworked by storms. Note the small micropellets in the matrix (upper). Matrix is composed of intraclasts, micritized forams, micropellets, dasyclad algae (Clypeina sulcata), and stromatoporoid fragments (lower). Photomicrograph widths are 1.1 mm (upper) and 7.4 mm (lower), Shedgum core.

FIGURE 18. Proximal middle-ramp stromatoporoid-coral rudstone, containing a mud-rich packstone matrix and interparticle, moldic, intraparticle, and microporosity. Stromatoporoids have been reworked by storms and bioeroded by bivalves, still visible. Coral mold, with casts of coralites, is beneath 223 (lower left). Note the micropellets that acted like microball bearings during storm reworking. Both photomicrograph widths are 1.1 mm, Shedgum ArabD core on display.

114 / Lindsay et al.

FIGURE 19. Upper photo is Arab-D – Jubaila contact in Wadi Hilwah. Geologist (encircled) is standing by the only in-place coral-stromatoporoid biostrome found in outcrop. All others have been reworked by storms. Arab-D – Jubaila contact is about 15 ft (4.5 m) above the geologist. Top of Arab-D is at the hill top. Lower photo is Arab-D – Jubaila contact in Wadi Nisah. Arab-D is the thin, platy outcrops above geologists, whereas Jubaila is the thicker beds beneath their feet. Note the large storm ripple in Jubaila.

suggest that the sediment accumulated in quiet, lowenergy conditions beneath fair-weather-wave base away from strong wave or current action. Firmgrounds formed on the flooding surfaces of individual carbonate cycles. Firmgrounds are typified by vertical Thalassinoides burrows, along with side gallery living areas with an average vertical depth of penetration of 1 m (3 ft) and a range of vertical penetration from only a few tenths of a meter (<1 ft) to as much as 2.4 m (8 ft) (Figure 21). Thalassinoides burrowers easily mined downward through soft muddy sediment, but ceased excavation once a grain-rich substrate, containing larger grains and skeletal detritus, was intersected. Bivalve-coated grain-intraclast (BCGI) limestone and, to a lesser extent, stromatoporoids and corals of the SRAC lithofacies were deposited as rudstone and floatstone by storm-induced high-energy debris flows that transported debris and matrix downslope into a deeper, shallow-marine distal middle-ramp

setting (Figure 10). Microbial encrustations, coated grains, and ooids indicate that original, coarse-sediment formation was farther upslope in the photic zone and fair-weatherwave base, in shallow-marine water of sufficient turbulence to move grains around and achieve relatively even coatings of grains. Muddy intraclasts, ripped up from areas of mud deposition down the ramp, is another indicator of brief, highenergy storm events. The matrix associated with rudstone and floatstone tends to be mudrich packstone, with mud-lean packstone and grainstone less common. The muddy matrix contains very small micropellets, many of which may have been created by the fruiting bodies of microbial encrustations (R. R. Leinfelder, 2003, personal communication). The entrained mud-rich matrix and micropellets acted like microball bearings to mediate transport of storm-generated debris flows much farther downslope. The relative proportion of mud incorporated into debris flows was responsible for two types of deposits. One is a cohesive debris flow with a mud-rich matrix (Figure 20), and the other is a noncohesive debris flow with a less mud-rich matrix (Figure 20). Additional mud matrix assisted debris flows in transporting sediment farther down the ramp slope. Weaker storm events were responsible for reworking bivalve-coated grain-intraclast (BCGI), whereas larger, more powerful storms were required to dislodge and transport stromatoporoids and corals because they formed biostromes and mounds and were better anchored than were BCGI lithofacies.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 115

FIGURE 20. Distal middle-ramp bivalve-coated grain-

the surrounding lime mudstone matrix is only partially dolomitized. As described above, micritic (MIC) limestones, bivalve-coated grain-intraclast (BCGI) limestones, and, to a lesser extent, stromatoporoid-red algae-coral (SRAC) limestones, are typically interbedded in the distal middle-ramp environment. In the outer-ramp environment, beyond the reach of storm-generated debris flows (Figure 10), lime mudstones (micritic lithofacies) were the only sediments deposited. Each cyclic package is capped by a firmground (Figures 8 [the foldout located in the back of this publication] and 10) that are typified by Thalassinoides burrows that penetrated the muddy carbonate layer. Although a lagoon paleoenvironment has been suggested for this lithofacies (Enay, 1987), the presence of the forams Lenticulina spp., Nodosaria spp., and Pseudomarsonella spp., along with monaxon and tetraxon sponge spicules, coccoliths, and juvenile costate brachiopods, is more consistent with open, normal-marine conditions below fair-weather-wave base, in perhaps 25 – 30 m (80– 100 ft) of water depth. Modern assemblages, including Lenticulina and Nodosaria, are confined to water depths greater than 100 m

intraclast rudstone. The upper part is the cohesive debris flow (mud rich) and the lower part is the noncohesive debris flow (grain rich). Debris flows were created by storms and rest atop firmgrounds burrowed by Thalassinoides. Photomicrograph widths are 7.4 mm (upper) and 2.2 mm (lower), Shedgum Arab-D core.

Micritic Lithofacies (MIC); Distal Middle-ramp and Outer-ramp Environments Micritic (MIC) limestone and dolostone (Figures 8 [the foldout located in the back of this publication] and 21) are common and tend to dominate the lower half of the Arab-D reservoir (zones 3a, 3b, and 4), less common in the middle Arab-D reservoir (zone 2b), and least common to nonexistent in the upper Arab-D reservoir (zone 2a and 1). In contrast to the other depositional lithofacies, micritic limestones are mostly mud supported. Grains present in this lithofacies include micritized grains, bivalves, ostracods, foraminifera, and intraclasts. Although porosity is generally low, typical pore types include microporosity, moldic, and intraparticle pores. In addition, interparticle and occasional intercrystal porosity is preserved in the grain-rich sediment that filled vertically oriented Thalassinoides burrow tubes (Figure 21). Diagenetic alterations include dolomitization, dissolution, recrystallization, and stylolitization. Burrows may be highly dolomitized, whereas

FIGURE 21. Outer ramp mudstone, bioturbated by Thalassinoides burrows and by soft-body burrowers, that contains only low amounts of microporosity. In the upper photomicrograph, note a few dolomite crystals and one quartz silt grain (center) in the pelleted matrix. In the lower photomicrograph is the contact of a Thalassinoides burrow tube that was filled with fine sediment and partially dolomitized and the mud-rich matrix containing a few windblown quartz silt grains. Photomicrograph widths are 1.1 mm (upper) and 7.4 mm (lower), Shedgum Arab-D core.

116 / Lindsay et al.

FIGURE 22. Proximal middle-ramp dolomitized, crossbedded, Cladocoropsis, peloid rudstone (upper) and dolomitized, cross-bedded mud-lean packstone (lower), containing moldic and intercrystal porosity. The upper photomicrograph contains saddle dolomite with ghosts of peloids. Photomicrograph widths are 2.25 mm (upper) and 1.1 mm (lower), Shedgum Arab-D core. (330 ft), but the Mesozoic equivalents are known to have occupied depths shallower than this. The outer ramp was probably affected by storms. Using the Persian Gulf as a modern analog, yearly storms have been found to stir bottom sediment to a depth of 50 m (164 ft), whereas a 100-yr storm can stir bottom sediment to 110 m (360 ft) (Saudi Aramco, oceanographic department database). However, bioturbation from soft-body burrowing organisms and Thalassinoides probably churned the muddier sediment and erased most traces of storm events. No rudstone or floatstone deposits were reworked and carried by debris flows this far down the ramp from the inner ramp, ramp crest, and proximal middle ramp (Figure 10), so distinction between burrow fill and the surrounding matrix is subtle except in those instances where differential partial dolomitization resulted from slight depositional differences.

Dolomite Lithofacies (DOLO) Dolomite (DOLO) (Figures 8 [the foldout located in the back of this publication], 22, 23) generally occurs as thin to thick (0.3–4.6-m, 1–14-ft), sheetlike beds throughout most of the Arab-D reservoir. Dolomite

typically consists of anhedral to euhedral crystals (30– 500 mm in size) that can be sucrosic or mosaic, depending on whether intercrystal porosity is present. Relict burrows and firmground surfaces are locally visible (Figure 23). Intercrystal porosity is dominant, with moldic porosity common (Figures 22, 23) and fracture porosity less common. Thin layers of sucrosic dolomite locally have sufficiently high porosity that they have anomalously high permeability referred to as super-K. Uncommon intracrystal porosity developed when dolomite crystals were partially dissolved to form dolo-donuts (Figure 23). Dolomite crystals commonly have cloudy, inclusion-rich centers and clear, limpid rims (Figures 22, 23). Uncommonly, threeto four-generation dolomite crystals have been identified that contain inclusion-rich centers, surrounded by a clear, limpid layer, with an additional inclusionrich layer, followed by a clear, limpid, outermost layer. Dolomitization of Cladocoropsis lithofacies has locally resulted in the dissolution of the Cladocoropsis fragments and the development of moldic porosity.

FIGURE 23. Dolomitized, cross-bedded, Cladocoropsis, skeletal, peloid rudstone, with a mud-rich packstone matrix and containing intercrystal and moldic porosity and very high permeability (super-K). Long laths are anhydrite crystals. The lower part is dolomitized, peloid mud-lean packstone at a potential subaerial exposure surface, with plant rootlets and/or Skolithos – Arenicolites burrow traces. The pore space is lined with insoluble residue and one dolo-donut intraparticle pore. Photomicrograph widths are 2.24 mm (upper) and 1.1 mm (lower), Shedgum Arab-D core.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 117

If individual moldic pores are touching or if the matrix dolomite has sufficient intercrystal porosity, they develop thin intervals with low porosity but very high permeability, also referred to as super-K (Figure 23). The least common dolomite occurrence is saddle (baroque) dolomite with crystals 500 mm up to a few thousand micrometers in size (Figure 22). This type of dolomite has been interpreted to result from hydrothermal emplacement along widely spaced fracture systems (Cantrell et al., 2004). Distribution is areally restricted, but may extend vertically through much of the reservoir.

ARAB-D RESERVOIR VERTICAL SUCCESSION AND SEQUENCE STRATIGRAPHY In general, from bottom to top, the Arab-D reservoir consists of (1) outer-ramp, mud-rich carbonate cycles (parasequences); (2) distal middle-ramp, mudrich, and rudstone- and floatstone-capped cycles; (3) proximal middle-ramp biostromes and mounds of stromatoporoids and corals and sheltered Cladocoropsis bank cyclic deposition; (4) ramp-crest shoal, highenergy cyclic deposition; and (5) inner-ramp lagoon and localized intertidal islands that form a few cycles at the top of the section (Figures 8 [the foldout located in the back of this publication] and 10). This shoaling-upward succession can be subdivided into 11 informal biozones, with the depths for the Shedgum Arab-D well shown (Figure 16). In this figure, the term ‘‘outer ramp’’ has to be understood in the context of normal, tropical carbonate deposition to indicate water depths of perhaps as much as 30 m (100 ft). In each of the biozones, there is a characteristic association of biotic elements indicative of the water depth and associated paleoenvironment shown (Figure 16). Outer-ramp and distal middle-ramp cyclic deposition was seaward and beneath fair-weather-wave base, but within or at least influenced by transport from sediments that were within storm-wave base. Proximal middle-ramp and ramp-crest shoal deposition was within fair-weather-wave base, and innerramp deposition was landward of wave base. This succession of carbonate deposition on a carbonate ramp recorded an overall shallowing-upward history related to a long-term base-level fall (Mitchell et al., 1988; Meyer and Price, 1993; Handford et al., 2002). This overall shallowing-upward, long-term baselevel fall and loss of accommodation space was punctuated by a series of smaller relative sea level rises and falls (Figure 8, the foldout located in the back of this publication). These rises and falls of sea level de-

posited the Arab-D reservoir as a series of submeterto meter-scale and thicker carbonate cycles (parasequences), with each carbonate cycle composed of one to three beds. Approximately five individual carbonate cycles combine to form a cycle set (parasequence set), with two to six cycle sets forming a HFS and four to five HFSs forming a composite sequence. Two composite sequences form the Arab-D reservoir. One composite sequence forms the Arab-D Member of the Arab Formation; the other forms the upper part of the Jubaila Formation (Figure 24). It is tempting to relate these different cycle scales to Milankovitch cycles (Milankovitch, 1930, 1941).

Carbonate Cycles (Parasequences) On the individual carbonate cycle (parasequence) scale, the Arab-D reservoir is composed of a transgressive mudstone, wackestone, or packstone base, deposited below fair-weather-wave base, and overlain by highstand cross-bedded grainstone, mud-lean packstone, or mud-rich packstone, deposited within fair-weather-wave base and/or storm-wave base (Figures 8 [the foldout located in the back of this publication], 10, 24). A total of 120 cycles (parasequences) have been identified in the Shedgum Arab-D core presented here (Figures 8 [the foldout located in the back of this publication] and 24). A Fischer plot of these cycles has been constructed and annotated to show which cycles were deposited in outer-ramp, distal to proximal middle-ramp, ramp-crest, and inner-ramp depositional settings (Figure 24). Individual carbonate cycles (parasequences), from bottom to top, in the Arab-D reservoir consist of the cycles discussed below.

Outer-ramp Cycles Outer-ramp carbonate cycles are submeter to meter scale and occasionally thicker and consist of mudstone, wackestone, and local mud-rich packstone that have an uppermost firmground with Thalassinoides burrows, with no highstand grain-rich cap (Figures 8 [the foldout located in the back of this publication], 10, 24), although a thin layer of highstand-equivalent sediment may be preserved at some localities. Thalassinoidesburrowed firmgrounds represent the maximum flooding surface (MFS) of an individual cycle.

Distal Middle-ramp Cycles Distal middle-ramp carbonate cycles are submeter to meter scale and thicker and contain transgressive mud-rich bases, similar to the outer-ramp setting, with the maximum flooding surface (MFS)

118 / Lindsay et al.

FIGURE 24. Fischer plot of the Shedgum Arab-D cored interval. Each carbonate cycle (parasequence) is color coded to show were it was deposited on the carbonate ramp. The small dimple on each cycle represents maximum flooding.

of the cycle at the top of the mud-rich base, where a firmground with Thalassinoides burrows formed. Thalassinoides burrows in Jubaila and Arab Formation outcrops are spaced 0.1 m (4 in.) from each other. Highstand progradation caused by relative sea level fall brought the proximal middle ramp, ramp crest shoal, and inner ramp close enough so that storms could rework inner-ramp, ramp-crest, and proximal middle-ramp mud, grains, and skeletal bioclasts downslope as debris flows that covered the firmground surface with grainstone or rudstone and floatstone (Figures 8 [the foldout located in the back of this publication], 10, 24). In some cases, more than one debris flow (evidenced by two or more stacked, fining-upward packages) covered the firmground, attesting to multiple storms passing through the area. Farthest downslope, in the distalmost parts of the middle ramp, highstand rudstone and floatstone caps are thin and are locally reduced to coarse packstones. In contrast, upslope and closer to the transition from proximal to distal parts of the middle ramp, high-

stand rudstones-floatstones thicken, and transgressive cycle bases thin to a point where they are of equal thickness (Figure 10).

Proximal Middle-ramp Cycles The proximal middle ramp contains a thinner transgressive base, with the maximum flooding surface forming a firmground or the top of a bioturbated interval and a thicker highstand cap composed of biostromes and mounds of stromatoporoids and corals and Cladocoropsis banks. Cladocoropsis banks were deposited in sheltered areas between biostromes and mounds and in sheltered areas immediately upslope of the biostromes and mounds, in slightly deeper parts of the proximal middle ramp (Figures 8 [the foldout located in the back of this publication], 10, 24). Stromatoporoid and coral biostromes and mounds were reworked by storms, although stromatoporoids and corals have been found to be still in place (Figure 19). Cladocoropsis was easily reworked by storms, and none have been found in their original growth position.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 119

Ramp-crest Shoal Cycles The ramp-crest shoal contains thinner transgressive bases that are only locally preserved, with the remainder eroded away or not deposited as wave base stirred the floor of the ramp crest and continuously washed grains with high energy and deposited a thin to thick cross-bedded highstand cycle cap (Figures 8 [the foldout located in the back of this publication], 10, 24). In this setting, transgressive bases of HFSs may contain Cladocoropsis, stromatoporoids, and corals. Storms during the highstand also reworked fragments of Cladocoropsis, stromatoporoids, and corals into cross-bedded, intraclast-peloid-ooid grainstone and mud-lean packstone (Figure 10).

Inner-ramp Cycles A lagoon formed behind the ramp crest and included uncommon, small intertidal islands (Figures 8 [the foldout located in the back of this publication], 10, 24). Transgressive bases contain bioturbated, mudrich to grain-rich strata with Thalassinoides-burrowed firmgrounds overlain by highstand-associated grainrich caps. These form the last few cycles at the top of the Arab-D reservoir. Intertidal islands, thought to range in size from a few well spacings to less than one well spacing (one well space = 1 km [0.6 mi]) in aerial extent, have been identified in a few cored wells. Generally, the last cycle of deposition of Arab-D carbonate was a transgressive lagoon cycle with a Thalassinoidesburrowed firmground that lacked a highstand cap. We interpret this facies to be the upslope deposit of a series of cycles deposited during a forced regression, terminating the deposition of the Arab-D.

Cycle Sets (Parasequence Sets) A total of 28 cycle sets (parasequence sets) has been identified in the Shedgum Arab-D core (Figures 8 [the foldout located in the back of this publication] and 24). These cycle sets are generally made up of approximately five individual cycles (parasequences), ranging from two to six, which record an increase in accommodation space that progressively declined with subsequent deposition. In the Arab-D reservoir, the last few shallowing-upward cycles may thicken, which implies that the earlier deposited cycles did not shallow up to sea level and only partly filled accommodation space (Figure 24). An overall increasing accommodation space trend that is matched by sediment fill at the beginning of a cycle set resulted in shallowing to near sea level, resulting in the last few cycles being deposited in higher energy within fairweather-wave base. In the lower part of the reservoir,

in a deeper shallow-marine setting, shallowing upward only extended into the influence of storm-wave base (Figure 10).

High-frequency Sequences In the Shedgum Arab-D core on display, eight HFSs have been identified; the top of a possible ninth HFS occurs at the base of the cored interval (Figures 8 [the foldout located in the back of this publication] and 24). Because the entire reservoir interval was deposited in an overall shallowing-upward depositional system with progressive loss of accommodation space, each successive HFS has a somewhat different and, therefore, unique, lithofacies succession from those above and below it. The general stacking pattern of HFSs is a transgressive systems tract (TST) of mud-rich cycles and a highstand systems tract (HST) of grain-rich cycles, with the maximum flooding surface (MFS) at the top of the mud-rich or muddier cycle (Figures 8 [the foldout located in the back of this publication], 10, 24). In the lower part of the Arab-D reservoir (zone 3), HFSs are 46 – 52 ft thick, with thick transgressive systems tracts (TSTs) and thin highstand system tracts (HSTs). Toward the top of the lower part of the reservoir, transgressive systems tracts (TSTs) become thinner, and highstand system tracts (HSTs) thicken in an expected trend. In the upper part of the Arab-D reservoir (zones 2 and 1), HFSs progressively thin upsection from 46, 40, 29, 19, and finally, to 9 ft (14, 12, 9, 6, and finally, to 2.7 m) (Figure 8, the foldout located in the back of this publication). This thinning of HFSs demonstrates an overall shallowing-upward history and base-level fall with the progressive loss of accommodation space. Transgressive systems tracts (TSTs) are initially thick and dramatically thin, whereas HSTs become the thicker and dominant overall deposition in the ramp crest (Figure 8, the foldout located in the back of this publication). In the lower part of the reservoir, transgressive systems tracts (TSTs) tend to be composed of outer-ramp and distal middle-ramp, mud-rich cycles (Figure 10). At the base of the cored interval, HSTs are composed of thin, storm-generated rudstone and floatstone with a mud-rich matrix (Figure 8, the foldout located in the back of this publication). Upsection, transgressive systems tracts (TSTs) become less mud rich and more grain rich and are composed of distal and proximal middle-ramp cycles (Figures 8 [the foldout located in the back of this publication], 10, 24). Highstand systems tracts are thicker, grain rich, and cross-bedded as they shallow upward into fair-weather-wave base and develop a ramp-crest shoal (Figure 8, the foldout

120 / Lindsay et al.

located in the back of this publication). Two paleosol horizons (described below) have been identified in the lower part of the reservoir (zone 3), with no paleosols identified in the upper part of the reservoir (zones 1 and 2) (Figure 23). In the ramp-crest shoal, the HSTs become thick and dominated deposition in the upper part of the reservoir (Figure 8, the foldout is located in the back of this publication). Base-level fall led to the deposition of several seaward-stepping ramp-crest shoal complexes with low-angle clinoform geometries. Downlapping clinoform progradation was both northward and southward from an area in the vicinity of the Shedgum Arab-D core. Perhaps, several HFSs prograded and downlapped to fill accommodation space throughout the length of Ghawar field.

Composite Sequences Two composite sequences have been identified. The upper composite sequence forms the Arab-D Member of the Arab Formation and is herein referred to as the Arab-D composite sequence and covers a vertical interval from 6809.2 to 6945.75 ft (2075.4 to 2117.06 m) (zone 1, zone 2a, and the upper half of zone 2b) (Figures 8 [the foldout located in the back of this publication] and 24). The top of the composite sequence is located at the top of the Arab-D reservoir, where the contact between the uppermost Arab-D carbonate and the overlying Arab C-D evaporite forms the sequence boundary. Along the composite sequence boundary, uncommon, small, shallow, karst-generated sink holes (not present in this well) filled with a few feet of collapse breccia are present. These can be seen in core, in outcrop, and on FMI Fullbore Formation MicroImager logs (Figure 13). In outcrop, small teepee structures characterize the top of the Arab-D composite sequence boundary in Wadi Hilwah (Figure 19). The Arab-D composite sequence in the Shedgum Arab-D core consists of 55 cycles (parasequences), 13 cycle sets (parasequence sets), and 5 HFSs (Figures 8 [the foldout located in the back of this publication] and 24). If the minimal amount of time for each cycle to be deposited is 20,000 yr (assuming association with 20,000-yr astronomical precession cycle), then the Arab-D Member may represent approximately 1.1 m.y. of time. Most of the Arab-D Member deposition was in the ramp-crest shoal, which is composed of approximately 22 cycles, 5 cycle sets (parasequence sets), and 2 HFSs, with a hardground separating the ramp crest from the inner ramp (Figure 24). The inner ramp consists of five cycles that compose one cycle set that forms a 9-ft (2.7-m)-thick, potential, HFS (Figure 24).

The remaining 28 cycles, incorporated into 7 cycle sets making up 2 HFSs and part of a third, are composed of proximal middle-ramp cycles, with only one distal middle-ramp cycle and no outer ramp cycles (Figure 24). The lower composite sequence is the upper part of the Jubaila Formation (6945.75-ft [2117.06-m] core depth to below the base of the core — lower half of zone 2b and all of zone 3) (Figures 8 [the foldout located in the back of this publication] and 24). This composite sequence is referred to as the Jubaila composite sequence. The top of the composite sequence was briefly subaerially exposed and contains uncommon meniscus and flowstone (gravity) cements. Two HFSs in other cored wells in the southern part of Ghawar shallowed upward and developed short-lived paleosols. One paleosol is a Psamment paleosol, with plant rootlets that were in a shoreface to localized island setting and the other deeper in the section contained mangrovelike rootlets in a paleosol at the zone 3a–3b contact (paleosols were identified by G. Retallack, 2003, 2004, personal communication). The lower half of the Jubaila composite sequence also contains a minor amount of windblown quartz silt (<1%) discussed below. The Jubaila composite sequence in the example core consists of 65 cycles, 15 cycle sets, and 4 HFSs (Figure 24). The ramp-crest shoal is composed of only three cycles within one cycle set in the uppermost HFS (Figures 8 [the foldout located in the back of this publication] and 24). No inner-ramp lagoon and intertidal island setting was deposited in this vicinity. However, sediments containing the shallow-marine benthonic foraminifera Trocholina alpina (Figures 15, 16) were reworked from a nearby lagoon and mixed into the ramp crest at this level. Outer-ramp and distal middle-ramp depositional settings represent most of the deposition, with 55 cycles in 10 cycle sets that are in 2 HFSs (Figures 8 [the foldout located in the back of this publication] and 24). The proximal middle ramp is represented by 12 cycles and 3 cycle sets in the uppermost HFS (Figures 8 [the foldout located in the back of this publication] and 24). Meyer et al. (1996, 2000) and Al-Dhubeeb (2005) used biostratigraphy to correlate the contact between the Arab-D Member of the Arab Formation and the underlying Jubaila Formation in outcrop sections in the Tuwaiq escarpment into the subsurface through the Khurais and Ghawar fields. Hughes (2004) defined nine paleonvironmentally influenced biozones (D1 to D9) in Ghawar field (Figure 25). Al-Dhubeeb (2005) has demonstrated that three of those biozones (D3 to D5) are not represented in Khurais field, and that three slightly younger zones (D5 to D7) are not

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 121

FIGURE 25. Biostratigraphic zonation of the Arab-D Member and upper Jubaila Formation from outcrops in the Tuwaiq escarpment into the subsurface through Khurais and Ghawar fields. Three zones belonging to the Arab-D Member are missing in outcrop. In Khurais field, three zones are also missing (upper and middle in the Arab-D and the lower from uppermost Jubaila). In Ghawar field, all zonations are present. The distance from the Tuwaiq escarpment to Ghawar field is 300 km (186 mi). From A. G. Al-Dhubeeb (2005, personal communication).

represented in the Tuwaiq escarpment outcrop west of Riyadh (Figure 25). In outcrop, the Arab-D Member is composed of inner-ramp lagoonal deposits of thin, platy, mud-rich beds, which rest upon reworked proximal middle-ramp stromatoporoid and coral rudstone and floatstone debris flows of the Jubaila Formation (Figure 19) (Meyer, 2000). In outcrop, the composite sequence boundary separating the overlying Arab-D Member and the underlying Jubaila Formation is located at a datum where stromatoporoids and corals disappear upsection and are overlain by thin, platy, and muddy inner-ramp lagoon beds (Figure 19) (Meyer et al., 1996; D. Vaslet, 2002, personal communication). In Ghawar field, with the complete biostratigraphic succession and complete inner-ramp to outerramp succession present, identifying the composite sequence boundary is more difficult. However, the same general criteria, upward decrease in the abundance and consistent presence of stromatoporoids and corals, coupled with an increase in abundance of the branched stromatoporoid Cladocoropsis, can be used to identify this boundary. Although the top of the Jubaila composite sequence is difficult to identify, a few additional pieces of information have helped identify the composite sequence boundary. First, at one outcrop locality in the Tuwaiq escarpment, 1 m (3 ft) of sandstone was found resting upon the composite sequence boundary that separates the Jubaila Formation from the

overlying Arab-D Member (Steineke et al., 1958). In the subsurface in Ghawar field, thin sections from cored wells contain quartz silt in the upper part of zone 2b (Figure 8, the foldout located in the back of this publication). The quartz silt is interpreted to have been reworked by eolian processes off the subaerially exposed western shelf (Tuwaiq escarpment area) Jubaila composite sequence boundary and blown into the basin (Figure 26). In the Arabian intrashelf basin, the ultimate resting site for the windblown silt is within the transgressive systems tract (TST) of the first HFS of the Arab-D composite sequence. This HFS is interpreted to onlap onto the Jubaila composite sequence boundary regionally, but did not toplap it (Figures 8 [the foldout located in the back of this publication] and 27). The second HFS of the Arab-D composite sequence only received windblown silt in the lower part of the transgressive systems tract (TST) (Figure 8, the foldout located in the back of this publication). This second HFS is interpreted to have also initially onlapped onto the Jubaila composite sequence in the vicinity of the present-day Tuwaiq escarpment and, during maximum flooding (MFS), toplapped the composite sequence (Figure 27). It has been estimated that approximately 47 ft (14 m) of topographic relief existed between the subject well in the Arabian intrashelf basin and outcrops in the Tuwaiq escarpment (Figure 27). The topographic relief implied by this onlap-toplap series of events is interpreted to have been 14 m (47 ft) because the first two HFSs of the Arab-D Member (some 14 m [47 ft]

122 / Lindsay et al.

FIGURE 26. Windblown quartz and dolomite silt set in a dolomitized matrix that contains intercrystal and moldic porosity. Quartz silt was blown off the subaerially exposed Jubaila Formation composite sequence boundary (Tuwaiq escarpment) into the Arabian intrashelf basin. Black crystals near the top of the photomicrograph are pyrite cubes. Photomicrograph width is 0.55 mm (6903.7 ft), Shedgum Arab-D core.

thick in the Shedgum Arab-D core) onlapped and eventually toplapped onto the Jubaila composite sequence boundary in the Tuwaiq escarpment (Figure 27). Based on this interpretation, the onlapping, HFSs imply that the western shelf in the Tuwaiq escarpment was subaerially exposed. The interpretation of subaerial exposure is further supported by several features. First, deep incisions have been found in the uppermost Jubaila shelf margin in an outcrop in the vicinity of the Diplomatic Quarter (DQ) on the southwest side of Riyadh. The last and deepest incision forms the top of the Jubaila Formation and the top of the Jubaila composite sequence (Figure 28). These observations could easily have been produced under

highstand conditions of the Jubaila composite sequence as overall accommodation space was lost. Second, Powers et al. (1966) described a mudrich interval in outcrop that could be regionally mapped throughout the subsurface. This interval may represent maximum flooding (MFS) of the Arabian intrashelf basin during Arab-D Member deposition and may be the dolomitized interval separating zones 2a and 2b or, more likely, the next lower, somewhat persistent, dolomitized interval (upper zone 2b) (Figure 8, the foldout located in the back of this publication). Dolomitization in Ghawar is generally in mud-rich transgressive systems tracts (TSTs) (Figure 8, the foldout located in the back of this publication) (although significant dolomitization locally occurs in grainy intervals, most notably in the CLADO facies, where dissolution of the Cladocoropsis colonies may produce superpermeability).

FIGURE 27. Conceptual model of basal Arab-D Member onlap and toplap onto the Jubaila composite sequence boundary, using the Shedgum Arab-D core (Ghawar field) in the Arabian intrashelf basin compared to outcrops in the Tuwaiq escarpment, west of Riyadh, a distance of 310 km (189 mi).

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 123

FIGURE 28. Photographs of the DQ section along the Makkah Highway, west side of Riyadh, where the Jubaila composite sequence boundary, three quarters up the road cut, has been downcut into by erosion and channelized. Storms have reworked corals and stromatoporoids along the boundary and into the channels. About 3 – 5 m (10 – 15 ft) of Arab-D overlies the Jubaila at this section. Vertical height of the bottom photo is 4.6 m (15 ft).

In outcrop, the upper part of the Jubaila Formation contains a rich assemblage of reworked stromatoporoids and corals with only a few whole pieces of Cladocoropsis present (Meyer, 2000). In the subsurface in Ghawar field, a rich assemblage of stromatoporoids and corals is also present, whose upsection give way to Cladocoropsis bank assemblages that are not represented in outcrop (Figure 8, the foldout located in the back of this publication). Stromatoporoid and coral biostromes and mounds are most common in the upper Jubaila Formation, whereas Cladocoropsis banks are most common in the lower part of the Arab-D Member. Exclusion of extensive Cladocoropsis in the Arab-D outcrop can be also explained by unsuitable, possibly too shallow environmental conditions, such as inner-ramp deposition. In the subsurface at Ghawar field, a combination of information can be used to identify the top of the Jubaila composite sequence boundary: (1) the stratigraphic level above which stromatoporoids and corals are less abundant; (2) the stratigraphic level above which Cladocoropsis is more abundant; (3) the interval through which windblown quartz silt is present in the overlying onlapping, HFSs; and (4) the stratigraphic level above which overlying transgressive mud-rich intervals have been dolomitized.

ARAB-D LATERAL FACIES VARIATIONS Although impossible to interpret from a single core from the Shedgum area, significant lateral variability in the distribution of lithofacies does occur in the field (Figure 29). Overall, a general pattern of thinning of the carbonate section and thickening of overlying evaporite occurs from north to south across Ghawar. This

thickness change has been alternatively interpreted as a facies change from carbonate to evaporite at the top of the Arab-D reservoir (Mitchell et al., 1988) or as a transition from a paleotopographic high in northern Ghawar into an intrashelf basin to the south (Handford et al., 2002). In addition, smaller scale heterogeneity occurs internally within the Arab-D reservoir and varies in a systematic way both temporally and geographically in the field. Lateral variability is least pronounced in the lower Arab-D reservoir, where meter-scale, coarsening-upward cycles consist of transgressive micritic wackestone and mudstone overlain by stormtransported, bivalve-coated grain-intraclast (BCGI) rudstone and floatstone and occasionally by stromatoporoid-red algae-coral (SRAC) deposits. They represent either episodic storm deposition or cyclic variation in sea level. The facies contained therein appear to be relatively continuous and create a highly stratified reservoir interval (Mitchell et al., 1988) that are interpreted to very gently downlap. In contrast, the middle and upper Arab-D reservoir is characterized by a much greater diversity of facies that tend to lack lateral continuity. Locally, thick accumulations of stromatoporoid-red algal-coral rudstone and floatstone with an internal matrix of mud-rich to mud-lean packstone and grainstone are present and have been interpreted as local carbonate buildups or mounds (Mitchell et al., 1988; Handford et al., 2002) or as a stromatoporoid bank (Hughes, 1996; Meyer et al., 1996). Skeletal-oolitic (SO) grain-rich shoals cap the upper Arab-D interval and are thought to have a shingled geometry, probably derived from the southward progradation in the southern 80% of Ghawar and northward progradation in the northern 20%. Highstand systems tracts (HSTs) contain the greatest lateral variability and exhibit potential downlap

FIGURE 29. South to north cross section of Arab-D reservoir lithofacies. Modified from Mitchell et al. (1988) and Cantrell (2004).

124 / Lindsay et al.

surfaces, whereas transgressive systems tracts (TSTs) contain the best and most extensive lateral continuity. Variations in continuity and development of downlapping shingled or clinoform surfaces may be related to relative sea level rise and fall that are not necessarily representative of the Late Jurassic greenhouse setting. Instead, the Late Jurassic appears to be intermediate between icehouse and greenhouse conditions (J. F. Read, 2004, personal communication). If truly greenhouse, tidal-flat caps should be common, and if icehouse, numerous exposure surfaces should be common (Read, 2004), although perhaps obscured if little meteoric water was available during subaerial exposure. What is described with respect to the Arab-D reservoir is considered to represent intermediate or transitional conditions between the greenhouse and icehouse extremes.

DIAGENESIS Diagenesis has modified the original sedimentary textures and fabrics of the Arab-D reservoir and is a significant factor in shaping the final reservoir quality of these rocks. Major diagenetic processes active in the Arab-D reservoir include dolomitization, dissolution and formation of microporosity, cementation, and compaction (Mitchell et al., 1988). The most significant of these processes are dolomitization, dissolution, and microporosity formation because they created the most variation in reservoir quality.

Dolomitization Dolomitization has long been recognized as a significant component of the Arab-D reservoir in Ghawar field (Powers, 1962; Mitchell et al., 1988; Meyer et al., 1996) and contributes to vertical and lateral heterogeneity to the reservoir in terms of textural, mineralogical, and pore-type alteration (Cantrell et al., 2001, 2004; Swart et al., 2005). At least three dolomite phases exist in the Arab-D, all of which are petrographically, geochemically, and stratigraphically distinguishable (Table 1). Fabric-preserving dolomite (Cantrell, 2004) consists of very fine crystals of dolomite (10 – 50 mm), in which the original limestone fabric is well preserved. Fabric-preserving dolomite typically occurs in the uppermost Arab D (zone 1) as thin, sheetlike, or stratiform layers that are always intimately associated with overlying laminated and bedded anhydrite. Reservoir quality is generally fair to poor, with porosity typically less than 10% and permeability ranging from a few millidarcys to a few tens of millidarcys. Because of

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 125

Table 1. Summary of Arab-D reservoir dolomite types.* Fabric-preserving Dolomite Style of dolomitization

Fabric-preserving

Petrography

Cloudy crystals, 10 – 50 mm in size

Cathodoluminescence Blue-light flourescence Distribution Average d18O ratio Average 87Sr/86Sr ratio Time formed

Nonluminescent Flouresces strongly Stratiform 2.08% PDB 0.70681 Very early, almost penecontemporaneously(?) Hypersaline

Water compositions

NFP Dolomite

Baroque

Nonfabric preserving, typically obliterates all traces of the original limestone fabric Crystals may be clear or internally zoned, 50 to more than 150 mm in size Nonluminescent Flouresces weakly Stratiform 2.66% PDB 0.70682 Early

Obliterative, with saddleshaped crystals that display undulose extinction

Hypersaline

Crystals are typically somewhat cloudy, 100 – 700 mm in size Patchy luminescence Flouresces weakly Nonstratiform 7.37% PDB 0.70712 Later Hot, mineral-laden fluids from depth

*From Cantrell (2004).

its relatively heavy oxygen isotope values, intimate association with the overlying anhydrite, and fabricpreserving nature, fabric-preserving dolomite is interpreted to have formed early in the diagenetic history of the sediment by dense, highly evaporated, magnesium-rich brines percolating down into the underlying sediment from the overlying salina. The early dolomitization interpretation is in agreement with many previous studies of fabric-preserving or mimetic dolomites (Dawans and Swart, 1988; Pleydell et al., 1990; Land, 1991; Kimbell, 1993), and it is generally recognized that fabric-preserving texture will not form if the precursor limestone has already stabilized to low-Mg calcite (Sibley, 1982, 1991). It should be noted that there is dolomite lower in the section (zones 3a and 3b) where some aspects of the original sediment may be interpreted (e.g., grain size and shape, cross-bed inclination, etc.), but where fine details are not preserved. These dolomites are included with the non-fabric-preserving dolomites (NFPs) discussed below. Non-fabric-preserving dolomite consists of mediumsize crystals (50 to >150 mm), nonsaddle (nonbaroque) dolomite, in which the original limestone fabric has been lost, and in many cases, all fabric evidence has been completely obliterated (Table 1) (Cantrell, 2004). Typically, NFP dolomite occurs as stratiform layers scattered throughout the reservoir. It generally has very poor reservoir quality. Intercrystal or vuggy porosity is typically less than 10%, and permeability is less than 1 md. Locally, intercrystal porosity

may exceed 20%, and permeability may approach several darcys. NFP dolomite contains heavy oxygen isotope values (Table 1). Because of its general geochemical similarity to fabric-preserving dolomites, it is interpreted to have formed from hypersaline fluids derived from the overlying salina anhydrite. In a few uncommon examples, some NFP dolomite contains light oxygen isotope values, has good reservoir quality (similar to above), and is thought to represent a transitional form, with the third dolomite type known as saddle or baroque dolomite. Strontium isotopic ratios (Table 1) suggest that both the fabric-preserving and NFP dolomite formed early or shortly after deposition of the original sediment or during early burial (Cantrell, 2004). Saddle or baroque dolomite is a coarse-crystal dolomite with saddle-shaped crystals displaying undulose extinction in thin section. It is uncommon in the reservoir and appears to be limited to wells that contain abnormally thick sections of dolomite. In extreme cases, saddle dolomite is vertically pervasive and crosscuts stratigraphy. Geochemically, saddle dolomite is distinctive with high iron and very low (light) oxygen isotopic compositions (Table 1) and is interpreted to have formed from high-temperature fluids during burial diagenesis. Reservoir quality is fair to poor. The interpretation that saddle dolomite formed relatively late in the diagenetic history of a rock from high-temperature fluids during deeper burial diagenesis is consistent with many previous studies (Folk and Assereto, 1974; Radke and Mathis, 1980) and is in agreement with the fluid-inclusion

126 / Lindsay et al.

FIGURE 30. Cross sections contrast the lateral persistence and continuity of stratigraphic dolomite (section A) with vertically pervasive, nonstratigraphic dolomite (section B). Modified from Cantrell et al. (2004).

results from the Arab-D (mean homogenization temperature = 97.48C, Table 1) (Cantrell, 2004). In addition, quantification and mapping of the dolomite content of the Arab-D across Ghawar revealed that dolomite does not occur randomly or uniformly across the field. Fabric-preserving and most NFPs typically occur as bed-parallel, thin, lenticular or sheetlike units that range in thickness from 0.3 to 4.6 m (1 to 15 ft). These stratiform dolomite bodies commonly occur within fairly well-defined stratigraphic intervals in the Arab-D (Figure 30A). As such, these dolomites may vertically stratify the reservoir and result in local impediments to vertical flow. Note, however, that thin intervals (locally as little as 0.3 m [1 ft] thick) of high-porosity and high-permeability NFP dolomite may contribute a high proportion of the flow from an individual well, creating a high-flow or super-K zone. Detailed core descriptions throughout the field have begun to show a preference for dolomitization to be more pervasive in the transgressive systems tract (TST) and at the maximum flooding surface (MFS), with less dolomite associated with the HST. The apparent reason for this phenomenon is that the TST

is more mud rich. Descending brines sourced from the overlying C-D evaporite will percolate through the more grainrich HST and be dramatically slowed to preferentially dolomitize the TST. In contrast, the more uncommon saddle dolomite and some isotopically light forms of NFP dolomite (more coarsely crystalline than most NFP dolomite) are nonstratiform in nature, in that they typically occur as abnormally thick dolomitized intervals that crosscut stratigraphic reservoir zones (Figure 30B). Mapping reveals that this type of dolomite occurs as linear northeast–southwest-oriented fracture and fault trends across Ghawar (Figure 31). Dolomite along these trends typically makes up 40– 60% (and locally as much as 100%) of the entire Arab-D reservoir interval, whereas offtrend dolomite typically makes up less than 20% of the reservoir. These fractureand fault-related saddle dolomite trends appear to be very limited areally (1 – 2 km [0.6 – 1.2 mi] in width) but extend obliquely across the field for distances greater than 60 km (37 mi). Because of their narrow width (which is on the order of one to two well spacings), these trends are interpreted to be vertical in orientation and to potentially create horizontally separated compartments in the reservoir and would generally act as vertically oriented, elongate, lowpermeability barriers that would drastically effect horizontal flow in the reservoir. When porous, these trends of saddle (baroque) dolomite can contain high permeability and act as sites of high fluid flow, or super-K, in the reservoir (the occurrence of which is currently unpredictable). A model of dolomitization (Figure 32) is proposed to explain the occurrence of both stratiform and nonstratiform types of dolomite. Stratiform dolomite

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 127

originated via several mechanisms that occurred early in the diagenetic history of the sediment and are related to downward percolation of hypersaline brines as the overlying C-D evaporite was being deposited (Figure 32). In contrast, nonstratiform, predominantly saddle dolomite formed from hot, mineralladen fluids that ascended upsection and emplaced into the reservoir from depth along a series of faults and/ or fractures and preferentially dolomitized the section adjacent to faults and fractures (Figure 32). In some instances, the fractures were first solution widened prior to emplacement of the saddle dolomite. Saddle dolomite has been observed in faulted blocks of Jubaila – Arab-D carbonates along the north flank of Wadi Nisah, which provides some evidence for fault-related dolomitization of the saddle dolomite.

Implications

FIGURE 31. The left is a map of percent dolomite in the Arab-D reservoir, Ghawar field. The right is a map of percent dolomite in zone 2b, Arab-D reservoir, Ghawar field. Modified from Cantrell et al. (2004).

With the exception of saddle dolomite, which is thought to be associated with the movement of fluids along fracture and fault zones during later burial, the occurrence of dolomite in the Arab-D reservoir reflects specific physical and chemical conditions that occurred during deposition to early burial of the Arab-D. As a result, placing these dolomites in the framework of the overall depositional and stratigraphic architecture of the Arab-D is essential to understand the proper temporal and spatial

128 / Lindsay et al.

FIGURE 32. Dolomitization model for the Arab-D reservoir from Cantrell et al. (2004).

context in which dolostones formed and to provide a predictive framework for the occurrence of similar dolostones in other intervals. It is clear that the partitioning of dolomite types occurs in the Arab-D, with NFP dolomite occurring in both composite sequences of the Arab-D carbonate, whereas fabric-preserving dolomite occurs only in zone 1, at the top of the upper composite sequence

and nearest to the source of descending brines, where brine concentrations were highest and reaction rates were rapid. In addition, although NFP dolomite occurs in both Arab-D composite sequences, its distribution is not uniform. No significant dolomitization of any kind occurs in the skeletal-oolitic (SO) grainstone-dominated upper Arab-D (zone 2A), but NFP dolomite is common in the middle part and the upper part of the lower Arab-D reservoir (zones 2B and 3A). This NFP dolomite occurs in association with stromatoporoid-red algal-coral (SRAC) and Cladocoropsis (CLADO) rudstone and floatstone (packstone and grainstone matrix) in the basal part zone 2B and with mudstone and wackestone interbedded with bivalve-coated grain-intraclast (BCGI) rudstone and floatstone (grainstone, mud-lean and mud-rich packstone matrix) in zone 3A. Little or no dolomite occurs in zone 3B or below in the northern half of Ghawar, but some zone 3B mudstone is dolomitized in the southern half. Although fabric-preserving and NFP dolomites are interpreted to have similar origins, their different dolomite in Ghawar field. Modified fabrics reflect their different stratigraphic contexts and potentially slightly different times of formation. Fabricpreserving dolomite formed early (penecontemporaneously), prior to extensive stabilization of the sediment to low-Mg calcite, whereas NFP formed later (but still early, as suggested by the Sr isotope data), subsequent to mineralogical stabilization of the sediment, but after initial compaction. Overall, the vertical distribution of dolomite reflects both the vertical succession of the Arab-D as well

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 129

FIGURE 33. Arab-D reservoir rock classification for Ghawar field. Modified from Cantrell and Hagerty (2003).

as the hydrodynamics of the top-down dolomitization mechanism proposed for both types of dolomite. Dense, hypersaline brines that migrated downward from the overlying salina initially moved into zone 1 shortly after deposition and produced fabric-preserving dolomite in unstabilized sediment. As deposition of salina evaporites continued, brines percolated downward and moved into the older Arab-D reservoir composite sequences below. Initially, these fluids moved unrestricted through well-sorted skeletal-oolitic (SO) grainstones in the Arab-D (zone 2A) but slowed significantly as they passed through the highly heterogeneous, poorly sorted, and more mud-prone sedimentrock at the zone 2A – 2B contact and in zones 2B and 3A. There, increased residence time of the fluids resulted in increased probability of dolomitization of these mud-prone sediments. In addition, the finer grain size in the muddier parts (TST) of these sequences promoted dolomitization by providing a greater number of nucleation sites for dolomitization (Sibley and Gregg, 1987) than would the less mud-prone, more grain-rich sediment-rock above. Ultimately, these fluids continued moving downsection (into lower zone 3A and zone 3B) and eventually equilibrated with the local host sediment-rock-water system and were no longer capable of dolomitizing. As a result, the amount of dolomite generally decreases downsection below the middle Arab-D reservoir. Dolomitization thus induced heterogeneity in a variety of ways that are generally predictable in the Arab-D reservoir. In a stratigraphic sense, dolomite occurrence is predictable based on the following criteria: (1) proximity to a thick, laterally extensive evaporite unit (C-D evaporite), the deposition of which provided a source of downward-migrating hypersaline brines; (2) a major sequence boundary that separated the overlying evaporite from underlying carbonate; and (3) cyclic carbonates that contain poorly sorted, mud-prone sediments that are prone to retain

dolomitizing fluids. The upper composite sequence boundary at the top of the Arab-D served to help partition fabricpreserving from NFP dolomite; mud-rich transgressive systems tracts in HFSs also help localize dolomite. The presence of certain sedimentary fabrics, such as poorly sorted and mud-prone rocks in these cycles, appears to act as the primary control over the occurrence of nonsaddle dolomite. Provided that these three controlling factors are present, it is possible to understand and predict patterns of dolomitization in the Arab-D, as well as in other similar carbonate-evaporite cycles that formed elsewhere. Vertically oriented hightemperature saddle (baroque) dolomite, in fractureand fault-related structural trends, across the Arab-D reservoir have the potential to influence horizontal flow in the reservoir.

RESERVOIR ROCK CLASSIFICATION An integrated petrographic and petrophysical study of Arab-D carbonates in the Ghawar field has provided a new carbonate reservoir rock classification (Figure 33) (Cantrell and Hagerty, 2003). Each reservoir rock type recognized in this classification has a distinct pore network as defined by porosity-permeability relationships and pore-throat radii connectivity expressed as pore-throat radii size distribution and J-function curves. This classification divides Arab-D carbonates into seven limestone and four dolomite rock types (Figure 33). The amount of matrix (lime mud) and pore types are the primary defining parameters for limestones. Reservoir rock type for dolomite is based on crystal texture. Note that numerous pigeon holes in the classification scheme (Figure 33) have been left blank, indicating that those criteria are of no significance in defining that particular rock type. The seven limestone reservoir rock types are based on the values of five petrographic parameters, applied in the following order (that is, starting from the top of the table in Figure 33 and working toward the bottom): (1) abundance of cement; (2) abundance of matrix (lime mud); (3) sorting (Krumbein and

130 / Lindsay et al.

FIGURE 34. Typical core-plug photograph examples of the major Arab-D reservoir limestone rock types. Divisions at top of photographs are in millimeters. Modified from Cantrell (2004).

The four dolomite reservoir rock types are classified according to dolomite crystal texture, although stratigraphic position and abundance of porosity can also be effective in their classification. The four textures are fabric preserving (Vfp), sucrosic (Vs), intermediate (Vi) and mosaic (Vm). As noted previously, the Vfp dolomite is only found in zone 1 of the Arab-D, where it is the major dolomite type. Vs dolomite occurs in dolomites with more than 12% porosity; Vm is less than 5% porosity, and Vi is between 5 and 12% porosity. Vfp dolomites have pore systems similar to their precursor limestone, but the pore systems of the other dolomite types are unique.

Implications

Pettijohn, 1988); (4) dominant pore type; and (5) size of the largest molds. Of the five, the amount of matrix is the most important parameter, although the amount of cement is the first criterion for subdividing the limestone rock types (Figure 33). In general terms, six of these seven types fall into two broad families, A and B, each of which can then be subdivided into three members (types I, II, and III) according to their matrix content. The first family, A, is a fairly coarsegrained, poorly sorted rock with relatively large molds. The second family, B, is a generally fine- to mediumgrained, well-sorted rock with few or small molds. The seventh rock type contains more than 10% cement, which modifies the pore-size distribution enough to warrant a separate reservoir rock type. Each of the reservoir rock types exhibits a distinctive pore-throat size distribution and, in turn, a Leverett J-function. The seven types are also characterized by distinctive porosity-permeability relationships (Figures 34, 35).

The distribution of these rock types is not random in the reservoir and reflects the stratigraphy and vertical succession of rock types previously described in the Arab-D reservoir (Figure 8, the foldout located in the back of this publication). As noted previously, zone 2A is dominated by relatively uniform, wellsorted, skeletal-oolitic grainstone that falls into reservoir rock type IB, whereas more heterogeneous Cladocoropsis (CLADO) and stromatoporoid-red algalcoral (SRAC) rudstone and floatstone that contain a matrix of packstone and grainstone occur in zone 2B and fall into reservoir rock type IA or IIA. Zone 3A tends to be dominated by heterogeneous and relatively mud-prone, bivalve-coated grain-intraclast (BCGI) rudstone and floatstone that contain a matrix of packstone that tend to fall in reservoir rock type IIB, whereas zone 3B typically contains abundant micritic wackestone to mudstone that would be classified in reservoir rock type IIIB. The occurrence of these reservoir rock types is thus predictable in the reservoir.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 131

FIGURE 35. Porosity-permeability crossplots of individual limestone reservoir rock types. Outliers are labeled according to their nonuniformity or textural characteristics (Bu = burrows; Cg = coarse grained; Fr = microfractures; Tx = textural variations). Green are for zones 1 and 2, and red are for zones 3 and 4. Modified from Cantrell (2004).

RESERVOIR CHARACTERIZATION USING IMAGE LOGS Background

This new classification scheme is significant, in that it focuses on observed heterogeneities in the rocks and groups them according to impact on flow properties. Thus, this classification is quite different from other, more widely used carbonate classifications. Most other classifications have relied on either textural (Dunham, 1962; Powers, 1962; Lucia et al., 2001) or lithofacies information (Mitchell et al., 1988; Meyer and Price, 1993). Figure 36 compares two standard classifications with the scheme used in this study. Although collection of texture and lithofacies information can provide the information required to develop the depositional and stratigraphic framework of the reservoir and its gross reservoir quality, it does not provide sufficient information to define pore network types in the reservoir. These textural and lithofacies classifications only indirectly relate to pore types and, in most instances, are unsuccessful in subdividing the reservoir rocks into groupings that are significant relative to their flow characteristics. The recognition and codification of the present classification, producing reservoir rock groupings that have significantly similar flow characteristics, invites development of a coordinated scheme for the prediction of occurrence of these groupings in the reservoir.

Understanding the heterogeneity of porosity and permeability is critical to evaluating production behavior of Ghawar field (Figure 37). Development scenarios have been implemented that employ long-radius, multilateral horizontal wellbores, termed ‘‘maximum reservoir contact’’ (MRC) wells, which target the uppermost Arab-D reservoir while avoiding deeper potential highpermeability zones. Because nearly all of these wells are not cored, few available data characterize reservoir quality in multilateral wells, with the exception of image and conventional logs. Several challenges exist to characterizing reservoir porosity and permeability variations in the Ghawar field. First, in some instances, a complex secondary diagenetic overprint (such as dolomitization) has altered the primary texture. Quantifying petrophysical properties of the reservoir and correlation of wire-line-log signatures to depositional facies become challenging. In these instances, porosity and permeability are controlled much more by diagenesis than by depositional facies, although depositional facies influenced the degree of diagenetic overprint. Second, conventional wire-line logs lack the fundamental vertical and azimuthal resolution to provide an adequate characterization of the reservoir heterogeneity. Significant textural and stratigraphic variations between adjacent thin layers hold the key to understanding the sequence stratigraphy. Third, standard core plugs, taken at vertical intervals of 6 in. (15 cm) or more, may not be adequate to characterize the reservoir quality.

132 / Lindsay et al.

permeability prediction (Russell et al., 2002, 2004), and extrapolation of image log-derived permeability data into wells without image logs using sonic-derived vuggy porosity measurements (Xu et al., 2006). Porosity and permeability in vuggy carbonate reservoirs is quantified through an optimized workflow of borehole image and conventional log processing that is calibrated to core data. Permeability from cores, which commonly has a poor correlation with conventional log-derived permeability, is found to have an exponential correlation to the vug porosity component of the total porosity computed from borehole image porosity analysis.

Results

FIGURE 36. Distribution of limestone samples by standard carbonate rock classifications and reservoir rock types. Dunham textures: GRN = grainstone; MLP = mud-lean packstone; PCK = packstone; WCK = wackestone; MUD = mudstone. Depositional facies: SO = skeletal-oolitic; MIC = micritic; BCGI = bivalve-coated grain-intraclast; SRAC = stromatoporoid-red algae-coral; CLAD = Cladocoropsis. Modified from Cantrell (2004).

Methodology Carbonate rocks are characterized by complex porosity systems (Lucia, 1995; Kazatchenko and Mousatov, 2002). Small-scale heterogeneity of porosity and permeability can only be resolved in wellbore logs with image log data (Newberry et al., 1996). Microresistivity image logs have ideal, small-scale vertical resolution as well as azimuthal borehole coverage for detailed reservoir heterogeneity analysis. Core data are necessary to calibrate all the log results, especially in heterogeneous rocks. In wells without borehole images, openhole lithology logs are required to determine reservoir rock types for heterogeneity analysis. Image logs that detect relative microresistivity changes (e.g., pores invaded by conductive drilling mud that contrast with resistive rock matrix) in the wellbore may aid in the identification of rock texture. Therefore, image logs have been used for various quantitative analyses in carbonate reservoir characterization, such as partitioning of matrix and vuggy porosity through borehole porosity image analysis (Newberry et al., 1996; Chitale et al., 2004), vug connectivity through image texture analysis leading to

Primary depositional texture exerts a strong influence on reservoir quality in the Arab-D because of the effect of lime mud on permeability. Muddominated textures, such as mudstone and wackestone, have generally low porosity and permeability, with the exception of diagenetically altered dolostone that has relatively high permeability in intermediate porosity. Packstone and floatstone have higher porosity and permeability, and grain-supported textures, such as grainstone and rudstone, have the best reservoir quality. Traditional log-based porositypermeability transforms do not adequately predict permeability in different textures with similar total porosity values because the vuggy porosity component of the total porosity is not measured on conventional wire-line logs, which provide porosity values averaged more than 0.6 – 1.2 m (2 – 4 ft) vertically. Stromatoporoid floatstone and rudstone contain significant vuggy porosity in image logs, and the clearly multimodal porosity spectrum measured azimuthally in image logs contributes to anomalously high permeability that matches production data. Sucrosic dolostone, commonly characterized by large intercrystalline vugs, accounts for extremely high flow from very thin intervals (<0.3 m [<1 ft] thick) because of permeability that is measured from image log analyses at two to three orders of magnitude higher than conventional log transforms. This dramatic increase of permeability with vug porosity partitioning without total porosity change can result in the inverse correlation between porosity and permeability (Russell et al., 2002). Conventional wire-line-log porosity-permeability transforms can neither predict nor resolve such heterogeneity of permeability. Finally, well-sorted, highly permeable oolitic, peloidal grainstones of the upper Arab-D are commonly very thin (less than 0.3 m [1 ft]) and account for high flow

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 133

FIGURE 37. Cross-bedded peloid grainstone in zone 2A. Image logs resolve thin beds of grainstone (black lines on left) and commonly show hints of cross-bedding (as indicated by the enlarged view of the red outlined interval shown below).

rates (Figure 37). These grainstones are well sorted and exhibit a high degree of interconnectedness of intergranular pores. Image-derived permeability data strongly correlate with core-derived permeability in these facies. Image logs clearly resolve the thin grainstone layers, and calibration of image facies with highresolution core description data provides a methodology for extrapolating grainstone layers into uncored wells using image logs. A methodology is shown for predicting flow behavior from image logs alone using image-derived permeability. After calibration in vertical cored wells, the classification of image textures and facies is the basis for estimating grainstone, stromatoporoid, and dolostone facies, cyclicity, and their associated permeabilities in image logs from horizontal and uncored wells.

CONCLUSIONS 1) The Arab-D reservoir in the Ghawar field was deposited on a tropical carbonate ramp that extended from outer ramp (below fair-weather-wave

base) to a depth of approximately 30 m (100 ft) water depth characterized by micritic cycles with Thalassinoides burrows. Updip, the distal middle ramp is characterized by micritic cycles capped by Thalassinoides firmgrounds with overlying floatstone to rudstone of dominantly bivalve-coated grain-intraclast sediment. The proximal middle ramp is characterized by domed and encrusting stromatoporoid and coral biostromes and mounds with delicately branched Cladocoropsis stromatoporoids in sheltered areas. The ramp-crest shoal is characterized by skeletal-oolitic grainstone to packstone, and the inner-ramp lagoon is characterized by more micritic sediment with local intertidal islands. 2) The reservoir is made up of two composite sequences. The lower composite sequence is the upper part of the Jubaila Formation. The sequence boundary occurs at the top of the rampcrest shoal, which is overlain by middle-ramp lithofacies. The upper composite sequence is the D member of the Arab Formation, the sequence

134 / Lindsay et al.

boundary at its top marked by local karst-collapse breccias and flooded over by subaqueous evaporites interpreted as the TST of the following composite sequence. 3) Major diagenetic features of the Arab-D reservoir include dolomitization, dissolution, cementation, and compaction. Dolomitization was mostly non-fabric-preserving and stratiform, occurring in more micritic TST sediments and probably formed by downward seepage of Mg-rich brines during deposition of the overlying C-D anhydrite. Locally, saddle dolomite occurs in bodies that crosscut stratigraphy, perhaps resulting from late fluids moving up local fracture zones. Dissolution is widespread, especially in the grainier parts of the reservoir, but the cause is less obvious given the evaporitic nature of the overlying hydrographic system. 4) Porosity includes interparticle, moldic, intraparticle, and microporosity. Permeability is mostly dominated by interparticle pores, which are typically only lightly cemented. Although moldic pores are abundant, they typically only contribute significantly to permeability in dolomite, where the delicately branched stromatoporoid Cladocoropsis has been leached and the matrix has been replaced by tightly intergrown dolomite. Dolomite is porous to nonporous (tight), with porosity a mixture of moldic, intercrystal, and intracrystal pores. 5) Image logs, in conjunction with other logs, show significant promise for the identification of lithofacies, especially coarse-grained lithofacies, in the Arab-D reservoir. Image log data clearly resolve thin layers of connected vugs in sucrosic dolostone and stromatoporoid rudstone and floatstone that contribute to high permeability and flow rates. Permeability from image log analyses correlates strongly with core permeability, yet image-derived permeabilities are two to three orders of magnitude higher than conventional log transforms in intervals with high vuggy porosity. Predicted image-derived flow behavior correlates strongly with actual flowmeters.

ACKNOWLEDGMENTS The authors thank the management of Saudi Aramco for granting permission to publish this contribution. In particular, the authors thank Abdulla Al-Naim, Aboud Al-Afifi, Sa’id Al-Hajri, Adnan AlSharif, and George Grover for their support and review

of the manuscript. We thank Nassir Alnaji, David Bacchus, and the cartography section of Saudi Aramco for their help with the creation of the digital and scanned images.

REFERENCES CITED Al-Dhubeeb, A. G., 2005, Biofacies as a tool for calibrating the Jurassic Jubaila – Arab formational contact from outcrop in Riyadh area to subsurface in eastern Saudi Arabia: Master’s thesis, King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia, 219 p. Al-Hinai, K. G., A. E. Dabbagh, W. C. Gardner, M. Khan, and S. Saner, 1997, Shuttle imaging radar views of some geological features in the Arabian Peninsula: GeoArabia, v. 2, p. 165 – 178. Al-Husseini, M. I., 1997, Jurassic sequence stratigraphy of the western and southern Arabian Gulf: GeoArabia, v. 2, p. 361 – 382. Al-Husseini, M. I., 2000, Origin of the Arabian plate structures: Amar collision and Najd rift: GeoArabia, v. 5, no. 4, p. 527 – 542. Arkell, W. J., 1952, Jurassic ammonites from Jebel Tuwaiq, central Arabia: Royal Society of London Philosophical Transactions, Series B, v. 236, p. 241 – 313. Ayres, M. G., M. Bilal, R. W. Jones, L. W. Slentz, M. Tartir, and A. O. Wilson, 1982, Hydrocarbon habitat in main producing areas, Saudi Arabia: AAPG Bulletin, v. 66, p. 1 – 9. Barger, T. C., 2000, Out in the blue: Vista, Selwa Press, 320 p. Bates, B. S., 1973, Oscar for an oilfield: Saudi Aramco World, v. 24, no. 6, p. 14 – 15. Beydoun, Z. R., 1988, The Middle East: Regional geology and petroleum resources: Beaconsfield, United Kingdom, Scientific Press, 292 p. Beydoun, Z. R., 1991, Arabian plate hydrocarbon geology and potential — A plate tectonic approach: AAPG Studies in Geology 33, 77 p. Beydoun, Z. R., 1998, Arabian plate oil and gas: Why so rich and so prolific?: Episodes, v. 21, p. 74 – 81. Cantrell, D. L., 2004, Carbonate heterogeneity during global greenhouse time: Examples from the Jurassic of the Middle East: Ph.D. thesis, University of Manchester, Manchester, England, United Kingdom, 368 p. Cantrell, D. L., and R. M. Hagerty, 2003, Reservoir rock classification, Arab-D reservoir, Ghawar field, Saudi Arabia: GeoArabia, v. 8, p. 435 – 462. Cantrell, D. L., P. K. Swart, C. R. Handford, C. G St. C. Kendall, and H. Westphall, 2001, Geology and production significance of dolomite, Arab-D reservoir, Ghawar field, Saudi Arabia: GeoArabia, v. 6, p. 45 – 60. Cantrell, D. L., P. K. Swart, and R. M. Hagerty, 2004, Genesis and characterization of dolomite, Arab-D reservoir, Ghawar field, Saudi Arabia: GeoArabia, v. 9, p. 11 – 36. Carrigan, W. J., G. A. Cole, E. L. Colling, and P. J. Jones, 1994,

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 135 Geochemistry of the Upper Jurassic Tuwaiq Mountain and Hanifa formations petroleum source rocks of eastern Saudi Arabia: in B. J. Katz, ed., Petroleum source rocks, casebooks in earth sciences series: Berlin, SpringerVerlag, p. 67–87. Champetier, Y., and E. Fourcade, 1966, A propos de Cladocoropsis mirabilis Felix dans le Jurassique superieur de Sud-Est de l’Espagne: Estudios Geologicos, v. 22, p. 101 – 111. Chitale, D. V., J. Quirein, T. Perkins, G. B. Lambert, and J. C. Cooper, 2004, Application of a new borehole imager and technique to characterize secondary porosity and net-to-gross in vugular and fractured carbonate reservoirs in Permian basin: 45th Society of Professional Well Log Analysts Annual Logging Symposium, Noordwijk, June 6 – 9, 2004. Dawans, J. M., and P . K. Swart, 1988, Textural and geochemical alternations in late Cenozoic Bahamian dolomites: Sedimentology, v. 35, p. 385 – 404. De Castro, P., 1991, On the cortical layer of Thaumatoporellaceans (green algae) (abs.): 5th International Symposium on Fossil Algae, Capri, April 7 – 12, p. 16 – 17. De Castro, P., 2002, Thaumatoporella parvovesiculifera (Raineri): Typification, age and historical background (Senonian, Sorrento Peninsula, southern Italy): Bollettino della Societa Paleotologica Italiana, v. 41, no. 2 – 3, p. 121 – 129. De Matos, J. E., and R. F. Hulstrand, 1995, Regional characteristics and depositional sequences of the Oxfordian and Kimmeridgian, Abu Dhabi, in M. I. AlHusseini, ed., Middle East Geosciences Conference, GEO’94: Bahrain, Gulf PetroLink, v. 1, p. 346 – 356. Droste, H. H. J., 1990, Depositional cycles and source rock development in an epeiric intra-platform basin, the Hanifia Formation of the Arabian Peninsula: Sedimentary Geology, v. 69, p. 281 – 296. Dunham, R. J., 1962, Classification of carbonate rocks according to depositional texture, in W. E. Ham, ed., Classification of carbonate rocks: AAPG Memoir 1, p. 108 – 121. Durham, L. S., 2005, Saudi Arabia’s Ghawar field: The elephant of all elephants: AAPG Explorer, January 2005, p. 4 and 7. Enay R., 1987, Le Jurassique d’Arabie Saoudite Centrale: Geobios Memoir Special 9, 314 p. Folk, R. L., and R. Assereto, 1974, Giant aragonite rays and baroque white dolomite in tepee-fillings, Triassic of Lombardy, Italy (abs.): SEPM Annual Meeting Programs with Abstracts, v. 164, p. 34 – 35. Glennie, K. W., 2000, Cretaceous tectonic evolution of Arabia’s eastern plate margin: A tale of two oceans: in A. S. Alsharhan and R. W. Scott, eds., Middle East models of Jurassic/Cretaceous carbonate systems: SEPM Special Publication 69, p. 9 – 20. Hancock, P. L., A. Al-Kadhi, and N. A. Sha’at, 1984, Regional joint sets in the Arabian platform as indicators of intraplate processes: Tectonics, v. 3, p. 27 – 43.

Handford, C. R., D. L. Cantrell, and T. H. Keith, 2002, Regional facies relationships and sequence stratigraphy of a super-giant reservoir (Arab-D Member), Saudi Arabia, in J. M. Armentrout, ed., Sequence stratigraphic models for exploration and production: Evolving methodology, emerging models and application histories, 22nd Annual Bob F. Perkins Research Conference: Gulf Coast Section SEPM, p. 539 – 564. Hughes, G. W., 1996, A new bioevent stratigraphy of late Jurassic Arab-D carbonates in Saudi Arabia: GeoArabia, v. 1, p. 417 – 434. Hughes, G. W., 1997, The Great Pearl Bank barrier of the Persian Gulf as a possible Shu’aiba analogue: GeoArabia, v. 2, p. 279 – 304. Hughes, G. W., 2004, Middle to Upper Jurassic Saudi Arabian carbonate petroleum reservoirs: Biostratigraphy, micropaleontology and paleoenvironments: GeoArabia, v. 9, no. 3, p. 79– 114. Irwin, M. L., 1965, General theory of epeiric clear water sedimentation: AAPG Bulletin, v. 49, p. 445 – 459. James, N. P., 1983, Reef environment, in P. A. Scholle, D. G. Bebout, and C. H. Moore, eds., Carbonate depositional environments: AAPG Memoir 33, p. 345 – 440. Kazatchenko, E., and A. Mousatov, 2002, Primary and secondary porosity estimation of carbonate formation using total porosity and the formation factor: Presented at 2002 Society of Petroleum Engineers Annual Technical Conference and Exhibition, San Antonio, Texas, SPE Paper 77787, 6 p. Keith, T. H., 2005, Finding the super-giant: The discovery of Ghawar field: GeoFrontier, Dhahran Geoscience Society, v. 2, no. 1, p. 29 – 44. Kimbell, T. N., 1993, Sedimentology and diagenesis of late Pleistocene fore-reef calcarenites, Barbados, West Indies: A geochemical and petrographic investigation of mixing zone diagenesis: Ph.D. dissertation, University of Texas, Dallas, 296 p. Kinsman, D. J. J., 1964, Recent carbonate sedimentation near Abu Dhabi, Trucial Coast, Persian Gulf: Ph.D. thesis, University of London, London, 302 p. Konert, G., A. M. Al-Afifi, S. A. Al-Hajri, and H. J. Droste, 2001, Paleozoic stratigraphy and hydrocarbon habitat of the Arabian plate: GeoArabia, v. 6, p. 407 – 442. Krumbein, W. C., and F. J. Pettijohn, 1988, Manual of sedimentary petrography: SEPM Reprint Series 13, 549 p. Land, L. S., 1991, Dolomitization of the Hope Gate Formation (north Jamaica) by seawater: Reassessment of mixing-zone dolomite, in H. P. Taylor, J. R. O’Neill, and I. R. Kaplan, eds., Stable isotope geochemistry: A tribute to Samuel Epstein: Geochemical Society (London) Special Publication 3, p. 121 – 133. Leinfelder, R. R., F. Schlagintweit, W. Werner, O. Ebli, M. Nose, D. U. Schmid, and G. W. Hughes, 2005, Significance of stromatoporoids in Jurassic reefs and carbonate platforms — Concepts and implications:

136 / Lindsay et al. Facies: International Journal of Paleontology, Sedimentology and Geology, v. 51, p. 287 – 325. LeNindre, Y.-M., J. Manivit, H. Manivit, and D. Vaslet, 1990, Stratigraphie sequentielle du Jurassique et du Cretace en Arabie Saoudite: Bulletin de la Societe Geologique de France, series 8, 6, p. 1025 – 1034. Lucia, F. J., 1995, Rock-fabric/petrophysical classification of carbonate pore space for reservoir characterization: AAPG Bulletin, v. 79, p. 1275 – 1300. Lucia, F. J., J. W. Jennings, M. Rahnis, and F. O. Meyer, 2001, Permeability and rock fabric from wireline logs, Arab-D reservoir, Ghawar field, Saudi Arabia: GeoArabia, v. 6, p. 619–646. Maiklem, W. R., D. G. Bebout, and R. P. Glaister, 1969, Classification of anhydrite; a practical approach: Bulletin of Canadian Petroleum Geology, v. 17, no. 2, p. 194 – 233. McGuire, M. D., G. Kompanik, M. Al-Shammery, M. AlAmoudi, R. B. Koepnick, J. R. Markello, M. L. Stockton, and L. E. Waite, 1993, Importance of sequence stratigraphic concepts in development of reservoir architecture in upper Jurassic grainstones, Hadriya and Hanifa reservoirs, Saudi Arabia: Proceedings of 8th Middle East Oil Show, Society of Petroleum Engineers Paper 25578, p. 489 – 499. Meyer, F. O., 2000, Carbonate sheet slump from the Jubaila Formation, Saudi Arabia: Slope implications: GeoArabia, v. 5, no. 1, p. 144 – 145. Meyer, F. O., and R. C. Price, 1993, A new Arab-D depositional model, Ghawar field, Saudi Arabia: 8th Middle East Oil Show and Conference Proceedings, Bahrain, p. 465 – 474. Meyer, F. O., R. C. Price, I. A. Al-Ghamdi, I. M. Al-Goba, S. M. Al-Raimi, and J. C. Cole, 1996, Sequential stratigraphy of outcropping strata equivalent to Arab-D reservoir, Wadi Nisah, Saudi Arabia: GeoArabia, v. 1, p. 435 – 456. Meyer, F. O., G. W. Hughes, and I. Al-Ghamdi, 2000, Jubaila Formation, Tuwaiq Mountain escarpment, Saudi Arabia: Window to lower Arab-D reservoir faunal assemblages and bedding geometry (abs.): GeoArabia, v. 5, no. 1, p. 143. Milankovitch, M., 1930, Mathematische klimalehre and astronomische theorie der klimaschwankungen, in W. Koppen and R. Geiger, eds., Handbuch der Klimatologie, I (A): Borntraeger, Berlin, Gebruder, 176 p. Milankovitch, M., 1941, Kanon der erdbestrahlung und seine anwendung auf das eiszeitenproblem: Akad Royale Serbe, v. 133, 633 p. Mitchell, J. C., P. J. Lehmann, D. L. Cantrell, I. A. Al-Jallal, and M. A. R. Al-Thagafy, 1988, Lithofacies, diagenesis and depositional sequence; Arab-D Member, Ghawar field, Saudi Arabia, in A. J. Lomando and P. M. Harris, eds., Giant oil and gas fields— A core workshop: SEPM Core Workshop, v. 12, p. 459 – 514. Murris, R. J., 1980, Middle East stratigraphic evolution and oil habitat: AAPG Bulletin, v. 64, p. 597 – 618. Newberry, B. M., L. M. Grace, and D. D. Stief, 1996, Analy-

sis of carbonate dual porosity systems from borehole electrical images: Society of Petroleum Engineers Paper 35158, 7 p. Nicholson, P. G., 2000, Compressional, fault-related folds and Saudi Arabia’s major hydrocarbon fields (abs.): GeoArabia, v. 5, p. 152 – 153. Nicholson, P. G., 2002, A 700 million year tectonic framework for hydrocarbon exploration and production in Saudi Arabia (abs.): GeoArabia, v. 7, p. 284. Pleydell, S. M., B. Jones, F. J. Longstaffe, and H. Baadsgaard, 1990, Dolomitization of the Oligocene – Miocene Bluff Formation on Grand Cayman, British West Indies: Canadian Journal Earth Sciences, v. 27, p. 1098 – 1110. Pollastro, R. M., A. S. Karshbaum, and R. J. Vigor, 1999, Maps showing geology, oil and gas fields and geologic provinces of the Arabian Peninsula: U.S. Geological Survey Open-file Report, 97-470B, version 2.0, one CD-ROM. Powers, R. W., 1962, Arabian Upper Jurassic carbonate reservoir rocks, in W. E. Ham, ed., Classification of carbonate rocks: AAPG Memoir 1, p. 122– 192. Powers, R. W., 1968, Saudi Arabia: Lexique Stratigraphique International, 3: Paris, Centre National de la Recherche Scientifique, 171 p. Powers, R. W., L. R. Ramirez, C. D. Redmond, and E. L. Elberg, 1966, Sedimentary geology of Saudi Arabia: Geology of the Arabian Peninsula: U.S. Geological Survey Professional Paper 560-D, 150 p. Radke, B. M., and R. L. Mathis, 1980, On the formation and occurrence of saddle dolomite: Journal of Sedimentary Petrology, v. 50, p. 1149 – 1168. Read, J. F., 2004, Carbonate sequence stratigraphy, short course notes: Saudi Aramco, Dhahran, Saudi Arabia, unpaginated. Russell, S. D., M. Akbar, B. Vissapragada, and G. Walkden, 2002, Rock types and permeability prediction from dipmeter and image logs: AAPG Bulletin, v. 86, no. 10, p. 1709 – 1732. Russell, S. D., K. Sadler, W. H. Weihua, P. Richter, R. Y. Eyvazzadeh, and E. A. Clerke, 2004, Applications of image log analyses to reservoir characterization, Ghawar and Shaybah fields, Saudi Arabia: GEO 2004, Middle East Geoscience Conference and Exhibition, abstract, p. 125. Scotese, C. R., 1998, Quick time computer animations, Paleomap Project: Department of Geology, University of Texas at Arlington. Sharland, P. R., R. Archer, D. M. Casey, R. B. Davies, S. H. Hall, A. P. Heward, A. D. Horbury, and M. D. Simmons, 2001, Arabian plate sequence stratigraphy: GeoArabia Special Publication 2, 371 p. Shearman, D. J., 1978, Evaporites of coastal sabkhas, in W. E. Dean and B. C. Schreiber, eds., Marine evaporites: SEPM Short Course 4, p. 6 – 42. Sibley, D. F., 1982, The origin of common dolomite fabrics: Clues from the Pliocene: Journal of Sedimentary Petrology, v. 52, p. 1087 – 1100.

Ghawar Arab-D Reservoir: Widespread Porosity in Shoaling-upward Carbonate Cycles / 137 Sibley, D. F., 1991, Secular changes in the amount and texture of dolomite: Geology, v. 19, p. 151 – 154. Sibley, D. F., and J. M. Gregg, 1987, Classification of dolomite textures: Journal of Sedimentary Petrology, v. 57, p. 967 – 975. Stegner, W., 1971, Discovery: The search for Arabian oil: Beirut, Middle East Export Press, 190 p. Steineke, M., R. A. Bramkamp, and N. J. Sander, 1958, Stratigraphic relations of Arabian Jurassic oil, in L. G. Weeks, ed., Habitat of oil: AAPG Symposium, p. 1294–1329. Swart, P. K., D. L. Cantrell, H. Westphal, C. R. Handford, and C. G. St. C. Kendall, 2005, Origin of dolomite from Ghawar field, Saudi Arabia: Evidence from petrographic and geochemical constraints: Journal of Sedimentary Petrology, v. 75, p. 476 – 491.

U.S. Geological Survey, 1963, Geologic map of the Arabian Peninsula, scale 1:2,000,000, and geologic quadrangle maps, scale 1:500,000: 1 map. Wender, L. E., J. W. Bryant, M. F. Dickens, A. S. Neville, and A. M. Al-Moqbel, 1998, Paleozoic (pre-Khuff ) hydrocarbon geology of the Ghawar area, eastern Saudi Arabia: GeoArabia, v. 3, p. 273 – 302. Xu, C., S. D. Russell, J. Gournay, and P. Richter, 2006, Porosity partitioning and permeability quantification in vuggy carbonates using wireline logs, Permian Basin, West Texas: Petrophysics, v. 47, no. 1, p. 13 – 22. Ziegler, M. A., 2001, Late Permian to Holocene paleofacies evolution of the Arabian plate and its hydrocarbon occurrences: GeoArabia, v. 6, p. 445 – 504.

FIGURE 8. Core description of the SDGM Arab-D core. Six cores were taken, with a total cored interval of 299 ft (91 m), through the complete Arab-D reservoir.

4

Strohmenger, C. J., L. J. Weber, A. Ghani, K. Al-Mehsin, O. Al-Jeelani, A. Al-Mansoori, T. Al-Dayyani, L. Vaughan, S. A. Khan, and J. C. Mitchell, 2006, High-resolution sequence stratigraphy and reservoir characterization of Upper Thamama (Lower Cretaceous) Reservoirs of a giant Abu Dhabi oil field, United Arab Emirates, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 139 – 171.

High-resolution Sequence Stratigraphy and Reservoir Characterization of Upper Thamama (Lower Cretaceous) Reservoirs of a Giant Abu Dhabi Oil Field, United Arab Emirates Christian J. Strohmenger, Ahmed Ghani, Omar Al-Jeelani, Abdulla Al-Mansoori, and Taha Al-Dayyani,

Lee Vaughan ExxonMobil Exploration Company, Houston, Texas, U.S.A.

Sameer A. Khan

Abu Dhabi Company for Onshore Oil Operations, Abu Dhabi, United Arab Emirates

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

L. Jim Weber

John C. Mitchell

ExxonMobil Exploration Company, Houston, Texas, U.S.A.

ExxonMobil Exploration Company, Houston, Texas, U.S.A.

Khalil Al-Mehsin Abu Dhabi National Oil Company, Abu Dhabi, United Arab Emirates

ABSTRACT

I

mportant hydrocarbon accumulations occur in platform carbonates of the Lower Cretaceous Kharaib (Barremian and early Aptian) and Shuaiba (Aptian) formations (upper Thamama Group) of Abu Dhabi. The Kharaib and Lower Shuaiba formations contain three reservoir units separated by three low-porosity and low-permeability dense zones. From base to top, the thickness of the reservoir intervals range from approximately 80, 170, to 55 ft (24, 51, to 16 m),

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215876M883271

139

140 / Strohmenger et al.

respectively, for the Lower Kharaib, Upper Kharaib, and Lower Shuaiba Reservoir Units. Core and well-log data of a giant oil field of Abu Dhabi, as well as outcrop data from Wadi Rahabah in the Emirate of Ras Al-Khaimah were used to establish a sequence-stratigraphic framework and a lithofacies scheme, applicable to all three reservoir units and the three dense zones. The Lower and Upper Kharaib Reservoir Units, as well as the lower, middle, and upper dense zones are part of the late transgressive sequence set of a secondorder supersequence, made up of two third-order composite sequences. The overlying Lower Shuaiba Reservoir Unit belongs to the late transgressive sequence set and the early highstand sequence set of this second-order supersequence and is made up of one third-order composite sequence. The three third-order composite sequences are composed of 19 fourth-order parasequence sets that show predominantly aggradational and progradational stacking patterns, typical of greenhouse cycles. Conventionally, composite sequence boundaries are placed at or near the base of the three dense zones. As an alternative scenario, the possibility that the major composite sequence boundaries actually occur on top of these dense zones is discussed. On the basis of faunal content, texture, sedimentary structures, and lithologic composition, 13 reservoir lithofacies and 8 nonreservoir (dense) lithofacies are identified from core. Similar lithofacies are identified in time-equivalent rock exposures studied in Wadi Rahabah. Depositional environments of reservoir units range from lower ramp to shoal crest to near-back shoal open-platform deposits. Dense zones were deposited in an inner-ramp, restricted shallow-lagoonal setting. Intensively bioturbated wackestone and packstone, and interbedded organic- and siliciclastic-rich limestone, characterize the dense zones. Locally, mud cracks, blackened grains, and rootlets are observed. Outcrop analogs of subsurface reservoirs allow for a detailed investigation of facies architecture and structure of carbonate bodies. Integration of subsurface and outcrop data (e.g., low-angle clinoforms that cannot be seen in core data) leads to more insightful and realistic geological models of subsurface stratigraphy. Geological model realizations based on core, outcrop, well-log, and seismic data constrain fluid flow-simulation models. Results mimic known behavior in analogous producing fields, and the process of going from rock data to simulation provides a useful training tool for reservoir characterization methods and techniques.

INTRODUCTION Large hydrocarbon accumulations have been discovered and produced from platform carbonates of the Lower Cretaceous Kharaib and Shuaiba formations (Thamama Group) in Abu Dhabi (Alsharhan, 1989; Alsharhan and Nairn, 1993, 1997). The focus of this study is on the sequence stratigraphy and sedimentology of a giant Abu Dhabi oil field referred to herein as Field B (Figure 1). Field B was discovered in 1965 and went on production in 1973. It is a faulted anticline (Figure 2) with production of about 408 API oil from the Kharaib Formation and planned future production from

the Lower Shuaiba Formation. The recoverable oil reserves are estimated to be in billions of barrels. More than 350 production and water-injection wells have been drilled to date. The Kharaib Formation is of Barremian and early Aptian age (Vahrenkamp, 1996; Granier, 2000; Pittet et al., 2002; Granier et al., 2003; C. Liu, T.-C. Huang, Y.-Y. Chen, 2005, personal communication) (Figure 3) and contains two reservoir units (Lower Kharaib Reservoir Unit and Upper Kharaib Reservoir Unit) separated and encased by three zones of very low porosity and permeability, subsequently referred to as dense zones (lower, middle, and upper dense zones; Figure 4). The thickness of the Lower Kharaib Reservoir

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 141

FIGURE 1. Location map showing the major oil fields of Abu Dhabi (green) and Field B (red) where core and well-log data of the Kharaib and Lower Shuaiba formations were studied. Also shown is the location of the studied outcrops at Wadi Rahabah, Ras Al-Khaimah.

Unit is about 80 ft (24 m), and the thickness of the Upper Kharaib Reservoir Unit ranges between 150 ft (45 m; downflank) and 190 ft (58 m; crestal area). The thickness of the lower dense zone, underlying the Lower Kharaib Reservoir Unit, is approximately 50 ft (15 m); the thickness of the middle dense zone, separating the Upper and Lower Kharaib Reservoir Units, is approximately 45 ft (14 m), and the thickness of the upper dense zone, overlying the Upper Kharaib Reservoir Unit, is approximately 40 ft (12 m). The thickness of the Lower Shuaiba Reservoir Unit (overlying the upper dense zone; Figures 3, 4) is approximately 55 ft (17 m). Highly permeable beds (>1 d) that mainly consist of coated-grain grainstone and rudist floatstonerudstone facies are recognized in the upper parts of the Upper and Lower Kharaib Reservoir Units (Figure 4). Generally, reservoir quality degrades from the crest to the flank throughout the field. This is

because of pressure solution (compaction) and cementation caused by the interaction of the carbonates with formation water and increasing burial depth (Gro ¨ tsch et al., 1998a). In contrast to the water leg, the oil leg (mostly the crestal area) is not as affected by pressure-dissolution processes (Burgess and Peter, 1985). The presence of hydrocarbons seems to inhibit the development of stylolites. The distribution of stylolites therefore shows an apparent dichotomy between crest (less stylolites) and flank (more stylolites), interpreted not to be related to differential burial effects between crest and flank, but instead, to the hydrocarbon infilling of the structure. Crestal wells (oil leg) are slightly thicker and show excellent porosity and permeability compared to downflank wells (water leg), which exhibit reduced thickness and reduced porosity and permeability (Gro ¨ tsch et al., 1998a). Traditionally, the subdivision of the Kharaib and Lower Shuaiba formations into reservoir units and

142 / Strohmenger et al.

FIGURE 2. Seismic line (west – east) running through the center of Field B shown in Figure 1. The seismic cross section shows the structure of Field B and the interpreted main stratigraphic horizons from Upper Jurassic to Upper Cretaceous. LKRU = Lower Kharaib Reservoir Unit; UKRU = Upper Kharaib Reservoir Unit; LSRU = Lower Shuaiba Reservoir Unit.

subunits is based on lithostratigraphic correlation, using the vertical distribution of stylolites. Investigation of stylolite distribution in the Kharaib Formation shows that such an approach is not tenable and, therefore, not recommended for building geological (static) and, ultimately, reservoir (dynamic) models. The origin and the degree of stylolitization depend on the heterogeneity of the sedimentary column (e.g., preexisting surfaces, lithological contrasts, facies changes, and type of bedding; Park and Schot, 1968; Nelson, 1984). Moreover, some facies will be less affected than others by stylolites. For example, algal lithofacies will be more resistant to compaction during burial because of calcite cement trapping in the algal structure. In addition, early diagenetic rim cement in interparticle pore space of grain-dominated facies minimizes compaction. Previous Abu Dhabi Company for Onshore Oil Operations (ADCO) internal and published studies conducted on the Kharaib and Lower Shuaiba Reservoir Units of Field B and outcrop analogs of Oman

focused on reservoir characterization (Gro ¨ tsch et al., 1998a; Melville et al., 2004), facies analyses, and sequence stratigraphy (Gro¨tsch et al., 1998a; Borgomano et al., 2002; Pittet et al., 2002; van Buchem et al., 2002; Hillga¨rtner et al., 2003; Immenhauser et al., 2004) and agree closely with the results presented in this chapter. The results of published and unpublished studies conducted at Field B were used to build three-dimensional (3-D) geological and reservoir models that adequately describe the reservoir (Gro¨tsch et al., 1998a; Melville et al., 2004) in its midterm production state, but are based on only limited core data. The results presented in this chapter integrate core and well-log data from Field B, as well as outcrop data (Strohmenger et al., 2004a, b, c; Suwaina et al., 2004) from Ras Al-Khaimah (United Arab Emirates; Figure 1). For the first time, detailed, sequence stratigraphykeyed descriptions of approximately 15,000 ft (4580 m) of core material from 50 wells of Field B and 4 vertical outcrop sections of Wadi Rahabah (Ras Al-Khaimah) have been conducted, considerably improving the

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 143

FIGURE 3. Upper Thamama Group (Lower Cretaceous) sequencestratigraphic framework.

aries, parasequence set boundaries, and parasequence boundaries (Boichard et al., 1994; Rebelle et al., 2004; Strohmenger et al., 2004b). This can be explained by early diagenetic cementation occurring at the top of the shallowing-upward cycles, creating a contrast in rock properties with the overlying rock type. During burial, these heterogeneous intervals undergo preferential stylolitization. Calcite that is dissolved during pressure solution reprecipitates in pores adjacent to the stylolites (Park and Schot, 1968; Koepnick, 1984), forming a nonporous zone adjacent to the chronostratigraphic boundary. If the boundary separates graindominated lithofacies (below) from mud-dominated lithofacies (above), cementation will be located dominantly below the boundary. In the case of grain-dominated overlying material, cementation may occur above and below the chronostratigraphic boundary (Rebelle et al., 2004; Strohmenger et al., 2004b). Early diagenetic processes thus follow the sequencestratigraphic framework and, therefore, can be predicted away from well control.

preexisting facies and sequence stratigraphy models. The newly developed high-resolution sequencestratigraphic framework and facies scheme allow a better prediction of the vertical and lateral distribution of reservoir quality and reservoir continuity throughout Field B and have become standard for all other ADCO fields. Comparing the location of stylolites or denser, stylolite-bearing zones with the established highresolution sequence-stratigraphic framework of the Kharaib Formation reveals that some stylolites are related to third-, fourth-, and fifth-order sequence bound-

THAMAMA GROUP SEQUENCE-STRATIGRAPHIC FRAMEWORK The Thamama Group can be described by two second-order supersequences. The older supersequence corresponds to the Habshan Formation (Berriasian and early Valanginian). The younger supersequence encompasses the interval between the top Habshan and top Shuaiba formations (Sharland et al., 2001, 2004; Davies et al., 2002; Droste and van Steenwinkel, 2004; Strohmenger et al., 2004a, b, c; Haq and Al-Qahtani, 2005) (Figure 3). The transgressive

144 / Strohmenger et al.

FIGURE 4. Type well B-1, showing the established highresolution sequence-stratigraphic framework for the Kharaib and Lower Shuaiba formations. ULRU 1 = Upper Lekhwair Reservoir Unit 1. See Figures 9 – 29 for color code of lithofacies types.

sequence set of this supersequence includes, from oldest to youngest, the Lekhwair (Valanginian to Barremian), and the Kharaib (Barremian and early Aptian) formations; including the Hawar (upper dense zone). The Shuaiba Formation is of Aptian age (Vahrenkamp, 1996; Gro ¨ tsch et al., 1998b; Granier, 2000; Pittet et al., 2002; Granier et al., 2003; C. Liu, T.-C. Huang, Y.-Y. Chen, 2005, personal communication) and is part of the late transgressive sequence set (early Aptian) and the highstand sequence set (Late Aptian) of this longer term depositional cycle.

Based on facies stacking patterns (shallowing-upward trends), sedimentary structures (e.g., Glossifungites burrows), and surface morphologies (e.g., erosive surfaces), 4 third-order composite sequence boundaries (SB), 3 composite maximum flooding surfaces (MFS), and 13 marine flooding surfaces (FS), which may correspond to third- or higher order flooding surfaces and sequence boundaries (flooding surface/sequence boundary [FS/SB]) were identified from core material of the Kharaib and Lower Shuaiba formations and tied to wire-line logs (Figure 4). Where possible, third-order SBs, MFSs, and parasequence set boundaries (FS) were tied to the stratigraphic framework established by Sharland et al. (2001, 2004) and Davies et al. (2002) for the Arabian plate (Figure 4). The nomenclature proposed by Sharland et al. (2001, 2004) is based on genetic stratigraphic sequences (Galloway, 1989), bounded by MFSs. Each of their identified MFSs is assigned a name, a number, and an age, like K60 (123 Ma) and K70 (120 Ma). In contrast, our sequence-stratigraphic subdivision follows the sequence-stratigraphic approach established by ExxonMobil (Mitchum, 1977; Vail et al., 1977, 1991; Vail, 1987; Van Wagoner et al., 1987, 1988; Haq et al., 1988; Sarg, 1988; Sarg et al., 1999), where depositional sequences are bounded by sequence boundaries. To tie our identified chronostratigraphic boundaries back to the already existing stratigraphic framework of the Arabian plate, we had to update the existing nomenclature proposed by

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 145

Sharland et al. (2001, 2004) and Davies et al. (2002) in a way that our boundaries are assigned to the already existing names (e.g., K60 and K70) but also append information whether these boundaries correspond to sequence boundaries (e.g., K60_SB), maximum flooding surfaces (e.g., K60_MFS: corresponding to the K60 MFS of Sharland et al., 2001, 2004) or flooding surfaces (e.g., K60_FS100).

Kharaib and Lower Shuaiba Sequence-stratigraphic Framework The Kharaib Formation, including the Hawar (upper dense zone), corresponds to the late transgressive sequence set (TSS) of a second-order supersequence. It is described by a second-order sequence set (base lower dense zone to base upper dense zone or Hawar), made up of two third-order composite sequences (Figure 3). The lower third-order composite sequence starts at the base of the lower dense zone and is capped by a pronounced sequence boundary (exposure surface) about 20 ft (6 m) below the middle dense zone (Figure 4). It is overlain by a third-order composite sequence that is bounded on top by a regionally correlative sequence boundary below the upper dense zone (Figure 4). Fourteen fourth-order parasequence sets and several fifth-order parasequences that make up the two third-order composite sequences show predominantly aggradational and progradational stacking patterns, typical for greenhouse cycles (Sarg et al., 1999) (Figure 4). The Lower Shuaiba Formation corresponds to the late transgressive (TSS) and early highstand sequence set (HSS) of the same second-order supersequence and is made up of a third-order composite sequence (Yose et al., 2004a, b, 2006) (Figure 3), subdivided by five fourth-order parasequence sets (Figure 4).

Lower Kharaib Third-order Composite Sequence Intensively burrowed and bioturbated wackestone and packstone and interbedded organic- and siliciclastic-rich limestone characterize the lower part of the Lower Kharaib sequence. Thalassinoides burrows are abundant, and burrow fills are commonly dolomitized. Well-developed firmground surfaces (Glossifungites burrows) cap several parasequences (Sattler et al., 2005). Locally, mud cracks, blackened grains, and paleosols have been observed. Diversity of fauna and flora is low. Other characteristics of this interval are centimeter- and decimeter-scale discontinuous beds. However, the overall interval, referred

to as the lower dense zone (Figure 4), can be correlated for tens to hundreds of kilometers across the platform. This interval is interpreted as the early transgressive systems tract. Low- to moderate-energy, skeletal wackestone and packstone that overlie the lower dense zone are interpreted to correspond to the late transgressive systems tract (Figure 4). Overlying the transgressive systems tract (TST), the highstand systems tract (HST) exhibits a normal and diverse marine fauna and flora characterized by high-energy coated-grain, algal, skeletal grainstone and rudstone with sparse rudist fragments (Figure 4). The depositional environment is interpreted as a tidal-influenced, high-energy bioclastic shoal environment. Water depth probably did not exceed 10 m (33 ft) even during the highest rates of relative sea level rise. Normal-marine conditions with good water circulation prevailed, as indicated by the generally diverse faunal content. Nutrients were sufficiently abundant to sustain the development of the algal buildups and rudists. Water turbidity was low as indicated by the abundance of Lithocodium/Bacinella (Dupraz and Strasser, 1999; Immenhauser et al., 2005). Four fourth-order parasequence sets are capped by two marine flooding surfaces (K50_FS100 and K50_FS600), a third-order maximum flooding surface (K50_MFS; Figure 5A), and a third-order sequence boundary (K60_SB; Figure 5B). Several small-scale parasequences are identified in the Lower Kharaib sequence (K50_SB to K60_SB; Figure 4). These parasequences are superimposed on a larger third-order trend. Successive parasequences show slightly different characteristics. Oldest parasequences are muddominated lagoonal sediments and show intense bioturbation and firmgrounds at bounding discontinuity surfaces (Sattler et al., 2005). Firmgrounds formed during periods of subaerial exposure and subsequent rapid increase in accommodation and reduced sedimentation rates. Subsequent parasequences are thicker and are made up of lagoonal wackestone to packstone and algal, skeletal, peloid floatstone-rudstone. Sedimentation occurred in an open lagoonal environment with increased water circulation during loss of accommodation. Upper parasequences are grain dominated (algal, skeletal, peloid floatstone-rudstone, coatedgrain grainstone, and coated-grain, algal, skeletal rudstone) and thinner, indicating an overall increase in current energy and decrease in accommodation. The uppermost parasequences are thin and dominated by moderate- to high-energy miliolid shoals (peloid, skeletal packstone and peloid, skeletal grainstone). At this time, accommodation was nearly filled.

146 / Strohmenger et al.

FIGURE 5. Core photographs of chronostratigraphic boundaries. (A) Maximum flooding surface K50_MFS showing burrowed and bored hardground (arrows), as well as patchy dolomitization (Do). (B) Sequence boundary K60_SB showing stylolitically overprinted erosive surface, iron mineralization, and cementation below sequence boundary. (C) Maximum flooding surface K60_MFS showing erosive contact between skeletal wackestone (SW, below) and skeletal, peloid packstone (SPP, above), as well as patchy dolomitization (Do). (D) Sequence boundary K70_SB showing erosive and burrowed surface (Glossifungites burrows, Gl), as well as cementation below sequence boundary. (E) Sequence boundary K80_SB showing stylolitically overprinted erosive surface and cementation below sequence boundary. (F) Flooding surface K70_FS100 on top of the upper dense zone (Hawar), which might also correspond to third-order sequence boundary Ap3sb (see Figure 4). The pronounced pedogenic overprint of the carbonates below the erosive surface favors the interpretation of a major subaerial exposure surface.

Upper Kharaib Third-Order Composite Sequence The early transgressive systems tract of the Upper Kharaib sequence is composed of Orbitolina-rich, bioturbated and burrowed skeletal, peloid wackestone and packstone of the uppermost Lower Kharaib Res-

ervoir Unit (Figure 4) and is overlain by intensively burrowed and bioturbated skeletal, peloid wackestone and sparse packstone. This interval is laterally continuous over considerable distance (tens to hundreds of kilometers) and is referred to as the middle

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 147

dense zone (Figure 4). Cycle tops are capped by firmgrounds (Glossifungites burrows), with sediment- and calcite spar-filled burrows (Sattler et al., 2005). The middle dense zone consists of abundant peloids, small echinoderm fragments, and some dasycladacean and red algae. In addition, other green algae, small mollusks, and sponge spicules are common. Locally, mud cracks, blackened grains, and paleosols have been observed. In outcrops, the organic- and siliciclastic-rich nature of the Kharaib middle dense zone results in a friable, nodular weathered outcrop. During the late transgressive systems tract, an open lagoonal environment persisted on the platform. Dolomitized Thalassinoides firmgrounds indicate temporary cessation in sedimentation. Burrowed horizons are continuous over many kilometers (Sattler et al., 2005). The fauna is more diverse, with notable orbitolinids, echinoderm remains, and bivalves. Carbonate rock textures are dominated by skeletal wackestone. At the ocean margin, rudist shoals, microbial mounds, and beaches developed at this time (van Buchem et al., 2002). Skeletal, peloid wackestone-packstone, coated-grain grainstone, coated-grain, algal, skeletal grainstone and rudstone, rudist and chondrodont floatstone and rudstone, and peloid, skeletal packstone and grainstone, rich in miliolids, gradually covered the entire platform during the highstand systems tract. The rudists include caprotinids (Glossomyophorus) and monopleurids (Agriopleura) but no caprinids. Rudists, as well as chondrodonts, reflect deposition in shallow water. These grain-dominated lithofacies were deposited under moderate- to high-energy, normal-marine conditions and exhibit considerable lateral variability. In the uppermost part of the sequence, miliolid-rich, skeletal, peloid packstone and grainstone with planar bedding and low-angle cross-bedding imply moderateto high-energy conditions. In the Upper Kharaib sequence, shoaling did not occur until the uppermost part of the highstand systems tract. An increase in overall accommodation from the Lower Kharaib to the Upper Kharaib sequence may reflect a long-term second-order transgressive trend in relative sea level (Figure 4). Ten fourth-order parasequence sets bounded on top by eight marine flooding surfaces (K60_FS100, K60_FS200, K60_FS300, K60_FS600, K60_FS700, K60_FS800, K60_FS900, and K60_FS1000), a third-order maximum flooding surface (K60_MFS; Figure 5C), and a third-order sequence boundary (K70_SB; Figure 5D), as well as several small-scale parasequences, are identified in the Upper Kharaib sequence (K60_SB to K70_SB; Figure 4).

Lower Shuaiba Third-order Composite Sequence The early transgressive systems tract of the thirdorder Lower Shuaiba sequence corresponds to the upper dense zone (Hawar), overlying the Upper Kharaib sequence (Figure 4). This interval is rich in discoidal orbitolinids. Peloids, small echinoderm fragments, and dasycladacean algae also occur. The upper dense zone is organic and siliciclastic rich, and can be correlated for tens to hundreds of kilometers across the platform with varying thicknesses. Decimeter-scale beds are marked by highly burrowed firmgrounds (Glossifungites burrows). Mud cracks, blackened grains, and paleosols are quite common, indicating frequent exposure. The Lower Shuaiba Reservoir Unit, overlying the upper dense zone, is a shallow-to deeper water limestone consisting of algal (Lithocodium/Bacinella) boundstone (Immenhauser et al., 2005) and skeletal, peloid wackestone to packstone and foraminifera, skeletal wackestone (Figure 4). This succession shows a deepeningupward trend interpreted to correspond to the late transgressive systems tract. The overlying highstand systems tract is dominated by deeper marine foraminifera, skeletal mudstone to wackestone that show a subtle shallowing-upward trend as well as an upward increase in grain richness (graining upward) to skeletal (shell fragments), peloid wackestone to packstone. The Lower Shuaiba composite sequence is bounded on top by sequence boundary K80_SB (Figure 5E). Five fourth-order parasequence sets are capped by three marine flooding surfaces (K70_FS100, K70_FS200, and K70_FS300), a third-order maximum flooding surface (K70_MFS), and a third-order sequence boundary (K80_SB; Figure 5E). Several small-scale parasequences are identified in the lower part of the Lower Shuaiba sequence (K70_SB to K80_SB; Figure 4). The overlying Upper Shuaiba Reservoir Unit is dominated by deeper marine, planktonic foraminifera wackestone-mudstone and intercalated low-porosity and low-permeability, burrowed, and bioturbated wackestone to mudstone (dense intervals).

Alternative Sequence-stratigraphic Interpretation of the Kharaib Formation The identified composite sequence boundaries K60_SB and K70_SB of the Kharaib Formation clearly show indications of significant emersion of the Kharaib platform carbonates (Figure 5B, D) and can regionally be correlated throughout the entire Arabian platform. However, the overall facies successions from shallow-water upper ramp carbonates (reservoir lithofacies types) below the sequence boundaries to inner ramp (dense zones lithofacies types) above the

148 / Strohmenger et al.

FIGURE 6. Schematic models showing two possible sequence-stratigraphic interpretations of the dense zones, using the upper dense zone as an example. (A) Conventional model: upper dense zone deposits correspond to the early transgressive systems tract of the Lower Shuaiba sequence, overlying major third-order composite sequence boundary (K70_SB) on top of the Upper Kharaib Reservoir. Minor higher order flooding surface (K70_FS100) occurs on top of the upper dense zone. (B) Alternative model: upper dense zone deposits correspond to the late highstand systems tract of the Upper Kharaib sequence, displaying major third-order composite sequence boundary (Ap3sb) at the top of the upper dense zone. Minor third- or higher order sequence boundaries occur at the base (Ap2sb) and within the upper dense zone.

sequence boundaries follow a normal shallowingupward and progradational trend. In addition, the socalled dense zones, especially the upper dense zone (Hawar), show frequent omission surfaces (firmgrounds), most likely corresponding to subaerial exposure, and significant thickness variation throughout the Arabian platform. For example, the upper dense zone decreases in thickness from approximately 40 ft (12 m) in onshore Abu Dhabi to approximately 20 ft (6 m) in offshore Abu Dhabi and is not present at the platform margin in Oman (van Buchem et al., 2002). Therefore, as an alternative sequence-stratigraphic interpretation, composite sequence boundaries could be placed at the top of the lower (Barr4sb), middle (Barr6sb), and upper (Ap3sb; Figure 5F) dense zones (Figure 4). The dense zones would thus correspond to late highstand systems tracts, representing inner-

ramp, restricted platform deposits, time-equivalent to outer ramp, prograding carbonates (Figure 6). The omission surfaces in the dense zones are interpreted to represent higher order sequence boundaries, indicating the frequent subaerial exposure of the inner ramp during the late highstand systems tracts (reduced accommodation). Most of the time represented by the dense zones would actually correspond to nondeposition.

KHARAIB AND LOWER SHUAIBA FACIES AND DEPOSITIONAL ENVIRONMENT On the basis of texture, grain types, sedimentary structures, faunal content, and lithologic composition, 21 lithofacies types were defined for the subsurface Kharaib and Lower Shuaiba formations (Strohmenger et al., 2004a, b, c). Thirteen lithofacies

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 149

FIGURE 7. Paleobathymetrical profile showing the interpreted depositional environment of each of the thirteen lithofacies types identified within the Lower and Upper Kharaib Reservoir Units and the Lower Shuaiba Reservoir Unit. types (LF1–LF13) correspond to the three reservoir units: Lower Kharaib Reservoir Unit, Upper Kharaib Reservoir Unit, and Lower Shuaiba Reservoir Unit

(Figure 7), and eight lithofacies types (LF20–LF27) occur in three dense zones: lower dense zone, middle dense zone, and upper dense zone (Figure 8). The

FIGURE 8. Paleobathymetrical profile showing the interpreted depositional environment of each of the eight lithofacies types identified within the lower, middle, and upper dense zone.

150 / Strohmenger et al.

FIGURE 9. Reservoir lithofacies type 1: Rudist, peloid rudstone (RPR). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, porosity in blue). Thin-section photomicrograph shows large chondrodont shells.

lithofacies types and their interpreted depositional environments are listed below. Interpretations described here agree closely with the work of Pittet et al. (2002) and van Buchem et al. (2002).

  

blocky calcite cement as common pore-filling cement associated with subaerial exposure surfaces shallow subtidal, high-energy open platform above fair-weather-wave base shoal to upper-ramp rudist buildups and reworked buildups high porosity and moderate to very high (>1 d) matrix permeability moldic, vuggy, and interparticle porosity molds and vugs that dominate under exposure surfaces Lower (uncommon or not present at Field B) and Upper Kharaib Reservoir Units (see Figure 9)

Lithofacies Types of the Lower Kharaib, Upper Kharaib, and Lower Shuaiba Reservoir Units



The analyzed 13 lithofacies types range from open platform, deeper subtidal, lower ramp, to shallow subtidal to intertidal, upper-ramp environments (Figure 7). A detailed description of the individual lithofacies types is given below. For a summary, see Figures 9–21.



Lithofacies LF1: Rudist, Peloid Rudstone (RPR)

Lithofacies LF2: Rudist, Peloid Floatstone (RPF)





    

rudists, chondrodonts, and other mollusks, foraminifera, and echinoderms peloids and coated grains (ooids, superficial ooids) rudstone texture matrix: grainstone and packstone texture moderate to well sorted bimodal grain-size distribution

 

    

rudists, chondrodonts and other mollusks, foraminifera, echinoderms, and sponge spicules peloids floatstone texture matrix: packstone and wackestone texture poor to moderate sorted bimodal grain-size distribution

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 151

FIGURE 10. Reservoir lithofacies type 2: Rudist, peloid floatstone (RPF). Shown here are summary table, core photograph, and thin section photomicrograph (plane polarized light, porosity in blue).

FIGURE 11. Reservoir lithofacies type 3: Skeletal, peloid packstone (SPP) overlain by algal, skeletal, peloid floatstonerudstone (ASPF). Shown here are summary table, core photographs, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S, porosity in blue). Skeletal, peloid packstone may show fenestral structures (arrows; left core photograph) and commonly occurs on top of shallowing-upward parasequence sets and parasequences (Glossifungites burrows, Gl; right core photograph).

152 / Strohmenger et al.

FIGURE 12. Reservoir lithofacies type 4: Skeletal, peloid grainstone (SPG). Shown here are summary table, core photographs, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S, porosity in blue). Core photograph (left) shows low-angle swash cross-bedding. Thin-section photomicrograph shows abundant miliolids and small echinoderm plates.

FIGURE 13. Reservoir lithofacies type 5: Coated-grain, skeletal grainstone (CgSG). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, porosity in blue).

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 153

FIGURE 14. Reservoir lithofacies type 6: Coated-grain, algal, skeletal rudstone – floatstone (CgASR). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, porosity in blue).

FIGURE 15. Reservoir lithofacies type 7: Algal, skeletal, peloid floatstone – rudstone. Shown here are summary table, core photographs, and thin section photomicrograph (plane polarized light, porosity in blue). Thin-section photomicrograph shows preserved intraparticle porosity in Lithocodium/Bacinella grain and pore-filling cement (white areas).

154 / Strohmenger et al.

FIGURE 16. Reservoir lithofacies type 8: Algal, skeletal floatstone – boundstone (ASFB). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S, porosity in blue). Thin-section photomicrograph shows preserved primary intraparticle (framework) porosity in Lithocodium/Bacinella boundstone.

FIGURE 17. Reservoir lithofacies type 9: Orbitolinid, skeletal packstone (OSP). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S, porosity in blue). Thin-section photomicrograph displays echinoderm plates and densely packed discoidal orbitolinids with preserved primary intraparticle porosity.

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 155

FIGURE 18. Reservoir lithofacies type 10: Skeletal, peloid wackestone – packstone (SPWP). Shown here are summary table, core photograph, and thin-section photomicrograph (stained with alizarin red-S, porosity in blue). Thin-section photomicrograph shows benthonic foraminifera and small echinoderm plates.

FIGURE 19. Reservoir lithofacies type 11: Orbitolinid, skeletal wackestone (OSW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, porosity in blue). Thin-section photomicrograph displays discoidal orbitolinids.

156 / Strohmenger et al.

FIGURE 20. Reservoir lithofacies type 12: Skeletal wackestone (SW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, porosity in blue).

FIGURE 21. Reservoir lithofacies type 13: Foraminifera, skeletal wackestone (FSW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S, porosity in blue). Core photograph shows bioturbation. Thin-section photomicrograph displays planktonic foraminifera and echinoderm plates.

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 157 

     

blocky calcite as common pore-filling cement associated with subaerial exposure surfaces shallow subtidal, moderate-energy open platform above fair-weather-wave base upper ramp reworked rudist buildups low to high porosity and highly variable matrix permeability moldic, vuggy, intraparticle, and microporosity molds and vugs that dominate under exposure surfaces Lower (not present at Field B) and Upper Kharaib Reservoir Units (see Figure 10)

Lithofacies LF3: Skeletal, Peloid Packstone (SPP)           

   

miliolids and other foraminifera, echinoderms, algae, and mollusks miliolids that can be very abundant in the upper part of the Upper and Lower Kharaib sequences peloids, aggregate grains, oncoids, and ooids packstone texture with minor grainstone interlayer moderate to well sorted unimodal grain-size distribution uncommonly planar laminated and low-angle cross-bedded gradational basal contact and erosive and burrowed upper contact (Glossifungites burrows) caps fining- and shallowing-upward cycles desiccation cracks and circumgranular cracks around grains, indicating subaerial exposure shallow subtidal to intertidal, moderate-energy restricted and open platform above fair-weatherwave base inner shoal and upper ramp deposits low to high porosity and low to moderate matrix permeability intraparticle, micromoldic, and microporosity Lower and Upper Kharaib Reservoir Units (see Figure 11)

     

gradational basal contact and sharp and erosive (dominantly) or gradational upper contact shallow subtidal to intertidal, high-energy open platform above fair-weather-wave base upper ramp beach, near shoal crest, and near inner shoal deposits low to high porosity and low to high matrix permeability interparticle, intraparticle, and micromoldic porosity Lower and Upper Kharaib Reservoir Units (see Figure 12)

Lithofacies LF5: Coated Grain, Skeletal Grainstone (CgSG)  

          

miliolids and other foraminifera, echinoderms, green algae, Lithocodium/Bacinella fragments, and mollusks coated grains (ooids, superficial ooids, composite grains, and oncoids), aggregate grains, intraclasts, and peloids grainstone texture with minor packstone interlayer moderate to well sorted unimodal to bimodal grain-size distribution uncommonly cross-bedded sharp basal contact and gradational (dominant) or sharp upper contact shallow subtidal, high-energy open platform above fair-weather-wave base upper-ramp, near-shoal crest deposits moderate to high porosity and low to very high matrix permeability (>1 d) interparticle, intraparticle, moldic, and vuggy porosity molds and vugs that dominate under exposure surfaces Lower and Upper Kharaib Reservoir Units (see Figure 13)

Lithofacies LF4: Skeletal, Peloid Grainstone (SPG)

Lithofacies LF6: Coated-grain, Algal, Skeletal Rudstone-Floatstone (CgASR)





     

miliolids and other foraminifera and echinoderms miliolids that can be very abundant in the upper part of the Upper and Lower Kharaib sequences peloids, coated grains (ooids, superficial ooids and composite grains) grainstone texture with minor packstone interlayer well sorted unimodal grain-size distribution planar laminated and low-angle cross-bedded indicating tidal influence



   

Lithocodium/Bacinella, miliolids and other foraminifera, echinoderms, green algae, rudists, and other mollusks coated grains (composite grains, oncoids, superficial ooids, and uncommon ooids), aggregate grains, intraclasts, and peloids rudstone (predominant) to floatstone texture matrix: grainstone and mud-lean packstone texture poor to moderate sorted bimodal grain-size distribution

158 / Strohmenger et al.       

sharp basal contact and gradational upper contact shallow subtidal, high-energy open platform above fair-weather-wave base upper-ramp, near-shoal crest deposits moderate to high porosity and low to very high (> 1 d) matrix permeability interparticle, intraparticle, moldic, and vuggy porosity molds and vugs that dominate under exposure surfaces Lower and Upper Kharaib Reservoir Units (see Figure 14)

Lithofacies LF7: Algal, Skeletal, Peloid Floatstone-Rudstone (ASPF)

  

Lithofacies LF9: Orbitolinid, Skeletal Packstone (OSP)      

           

Lithocodium/Bacinella, foraminifera, echinoderms, rudists and other mollusks, and sponge spicules peloids, oncoids, aggregate grains, and uncommon coated grains (superficial ooids) floatstone (predominantly) to rudstone texture matrix: packstone to mud-lean packstone texture poor to moderate sorted bimodal grain-size distribution sharp basal contact and gradational finingupward upper contact shallow subtidal, moderate- to high-energy open platform above fair-weather-wave base upper-ramp algal buildups and reworked algal buildups moderate to high porosity and low to high matrix permeability interparticle, intraparticle, and microporosity Lower and Upper Kharaib Reservoir Units (see Figure 15)



   

       

        

Lithocodium/Bacinella, orbitolinids and other foraminifera, mollusks, and echinoderms algal-binding activity forms low-relief digitate growth framework and encrusting masses peloids floatstone to boundstone (bindstone) texture matrix: packstone and wackestone texture poor to moderate sorted bimodal grain-size distribution subtidal, low- to moderate energy open platform near fair-weather-wave base lower, upper-ramp algal buildups and reworked algal buildups

orbitolinids and other foraminifera, mollusks, echinoderms, and sponge spicules peloids packstone texture with minor wackestone interlayer moderate to well sorted bimodal grain-size distribution bioturbation shallow subtidal, low- to moderate-energy restricted platform, and subtidal, open platform near fair-weather-wave base deposits inner ramp, restricted lagoonal deposits, and upper- to middle ramp deposits low to high porosity and low matrix permeability intraparticle, micromoldic, and microporosity Lower and Upper Kharaib Reservoir Units (see Figure 17)

Lithofacies LF10: Skeletal, Peloid Wackestone-Packstone (SPWP)



Lithofacies LF8: Algal, Skeletal Floatstone-Boundstone (ASFB)

moderate to high porosity and low to high matrix permeability intraparticle (framework), shelter, and microporosity Lower Shuaiba Reservoir Unit (see Figure 16)

  

foraminifera, sponge spicules, algae, and echinoderms peloids and uncommon coated grains wackestone to packstone texture moderately sorted unimodal grain-size distribution bioturbation thin dolomitized intervals common subtidal, low-energy open platform near fairweather-wave base upper to middle ramp deposits low to high porosity and low to moderate matrix permeability intraparticle, micromoldic, and microporosity Lower and Upper Kharaib Reservoir Units, and Lower Shuaiba Reservoir Unit (see Figure 18)

Lithofacies LF11: Orbitolinid, Skeletal Wackestone (OSW)   

orbitolinids and other foraminifera, mollusks, echinoderms, and sponge spicules peloids wackestone texture with minor packstone interlayer

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 159     

   

moderate sorted bimodal grain-size distribution bioturbation thin dolomitized layers shallow subtidal, low-energy restricted platform, and deeper subtidal, open platform below fairweather-wave base inner-ramp, restricted lagoonal deposits, and middle-ramp deposits low to high porosity and low matrix permeability microporosity, intraparticle, and micromoldic porosity Lower and Upper Kharaib Reservoir Units (see Figure 19)

Lithofacies LF12: Skeletal Wackestone (SW)             

foraminifera, sponge spicules, echinoderms, and algae peloids wackestone texture moderately sorted well sorted unimodal grain-size distribution bioturbation thin dolomitized layers subtidal, low-energy open platform below fairweather-wave base middle-ramp deposits fine-grained, microporous limestone with low to moderate matrix permeability microporosity, intraparticle, and micromoldic porosity Upper Kharaib Reservoir Unit (see Figure 20)

  

Lithofacies Types of the Lower, Middle, and Upper Dense Zones The analyzed eight lithofacies types represent organic- and siliciclastic-rich, inner-platform, restricted lagoonal deposits (Figure 8). The frequent occurrence of blackened pebbles and blackened grains suggests a coastal, peritidal, oxygen-depleted environment (Strasser and Davaud, 1983; Strasser, 1984). This is supported by the elevated organic content of the carbonates, interpreted from high uranium readings of the spectral gamma-ray log. Sedimentary structures like Glossifungites burrows, desiccation cracks, root marks, and erosive surfaces, as well as paleosols indicate repeated subaerial exposures. A detailed description of the individual lithofacies types is given below. For a summary see Figures 22–29.

Lithofacies LF20: Wispy-laminated, Burrowed, Skeletal Packstone (WBSP)        

Lithofacies LF13: Foraminifera, Skeletal Wackestone (FSW)







        

planktonic and benthonic foraminifera, sponge spicules, and echinoderms peloids pyrite wackestone texture with minor mudstone interlayer moderately sorted well sorted unimodal grain-size distribution bioturbation subtidal, low-energy open platform below fairweather-wave base middle to lower ramp deposits

fine-grained, microporous limestone with low matrix permeability microporosity, intraparticle, and micromoldic porosity Lower Shuaiba Reservoir Unit (see Figure 21)



 

low diversity, abundant, and small-size fauna foraminifera, mollusks, echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains pyrite siliciclastics present packstone texture wispy- and nodular-bedded burrows and bioturbation shallow subtidal, low- to moderate-energy restricted platform inner shoal to inner ramp, restricted lagoonal deposits low porosity and low matrix permeability lower and middle (uncommon) dense zones (see Figure 22)

Lithofacies LF21: Wispy-laminated, Burrowed, Orbitolinid, Skeletal Packstone (WBOSP)  

low diversity and high abundance of fauna discoidal orbitolinids and other foraminifera, green algae (dasycladacean), mollusks, echinoderms, sponge spicules, and ostracods

160 / Strohmenger et al.

FIGURE 22. Nonreservoir lithofacies type 20: Wispy-laminated, burrowed, skeletal packstone (WBSP). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light). Thin-section photomicrograph displays abundant skeletal debris.

FIGURE 23. Nonreservoir lithofacies type 21: Wispy-laminated, burrowed, orbitolinid, skeletal packstone (WBOSP). Shown here are summary table, core photograph, and thin-section photomicrographs (plane polarized light). Thinsection photomicrographs show densely packed discoidal orbitolinids and wispy stylolites.

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 161

FIGURE 24. Nonreservoir lithofacies type 22: Wispy-laminated, burrowed, orbitolinid, skeletal wackestone (WBOSW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light).

FIGURE 25. Nonreservoir lithofacies type 23: Wispy-laminated, burrowed, skeletal wackestone – mudstone (WBSW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light). Thinsection photomicrograph shows dolomite rhombs and pyrite.

162 / Strohmenger et al.

FIGURE 26. Nonreservoir lithofacies type 24: Burrowed, bioturbated, skeletal packstone (BBSP). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light). Core photograph shows intensive burrowing and bioturbation.

FIGURE 27. Nonreservoir lithofacies type 25: Burrowed, bioturbated, orbitolinid, skeletal packstone (BBOSP). Shown here are summary table, core photograph, and thin-section photomicrographs (plane polarized light, stained with alizarin red-S). Thin-section photomicrograph shows densely packed discoidal orbitolinids and glauconite (green grains).

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 163

FIGURE 28. Nonreservoir lithofacies type 26: Burrowed, bioturbated, orbitolinid, skeletal wackestone (BBOSW). Shown here are summary table, core photographs, and thin-section photomicrograph (plane polarized light). Thin-section photomicrographs show glauconite (green grains, left thin section photomicrograph), discoidal orbitolinids, and green algae (right thin-section photomicrograph).

FIGURE 29. Nonreservoir lithofacies type 27: Burrowed, bioturbated, skeletal wackestone (BBSW). Shown here are summary table, core photograph, and thin-section photomicrograph (plane polarized light, stained with alizarin red-S).

164 / Strohmenger et al. 

             

peloids blackened pebbles and blackened grains dolomite rich pyrite rich siliciclastics present packstone texture wispy and nodular bedded burrows and bioturbation maroon lithoclasts, red paleosols, and caliche crusts rhizoliths (root tubes) and desiccation cracks frequent subaerial exposure shallow subtidal, low- to moderate-energy restricted platform inner-ramp, restricted lagoonal deposits low porosity and low matrix permeability upper dense zone (Hawar) (see Figure 23)

          

Lithofacies LF24: Burrowed, Bioturbated, Skeletal Packstone (BBSP)  

Lithofacies LF22: Wispy-Laminated, Burrowed, Orbitolinid, Skeletal Wackestone (WBOSW)

 

 

              

low diversity and high abundance of fauna discoidal orbitolinids and other foraminifera, green algae (dasycladacean), mollusks, echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains dolomite rich pyrite rich siliciclastics present wackestone texture with minor packstone interlayer wispy and nodular bedded burrows and bioturbation maroon lithoclasts, red paleosols, and caliche crusts rhizoliths (root tubes), and desiccation cracks frequent subaerial exposure shallow subtidal, low-energy restricted platform inner-ramp, restricted lagoonal deposits low porosity and low matrix permeability upper dense zone (Hawar) (see Figure 24)

      

 

   

     

low diversity, abundant, and small-size fauna foraminifera, mollusks, echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains dolomite rich pyrite rich

low diversity, abundant, and small-size fauna foraminifera, echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains pyrite packstone texture with minor wackestone interlayer burrows and bioturbation shallow subtidal, low- to moderate-energy restricted platform inner-ramp, restricted lagoonal deposits low to moderate porosity and low matrix permeability lower and middle (uncommon) dense zones (see Figure 26)

Lithofacies LF25: Burrowed, Bioturbated, Orbitolinid, Skeletal Packstone (BBOSP)



Lithofacies LF23: Wispy-laminated, Burrowed, Skeletal Wackestone-Mudstone (WBSW)

siliciclastics present wackestone to mudstone texture with minor packstone interlayer wispy and nodular bedded burrows and bioturbation paleosols and caliche crusts rhizoliths (root tubes) and desiccation cracks frequent subaerial exposure shallow subtidal, low-energy restricted platform inner-ramp, restricted lagoonal deposits low porosity and low matrix permeability lower and middle dense zones (see Figure 25)

    

low diversity and high abundance of fauna discoidal orbitolinids and other foraminifera, green algae (dasycladacean), echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains pyrite glauconite packstone texture with minor wackestone interlayer burrows and bioturbation shallow subtidal, low- to moderate-energy restricted platform inner ramp, restricted lagoonal deposits low porosity and low matrix permeability upper dense zone (Hawar) (see Figure 27)

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 165

Lithofacies LF26: Burrowed, Bioturbated, Orbitolinid, Skeletal Wackestone (BBOSW)  

        

low diversity and high abundance of fauna discoidal orbitolinids and other foraminifera, green algae (dasycladacean), echinoderms, sponge spicules, and ostracods blackened pebbles and blackened grains pyrite glauconite wackestone texture with minor packstone interlayer burrows and bioturbation shallow subtidal, low-energy restricted platform inner-ramp, restricted lagoonal deposits low porosity and low matrix permeability upper dense zone (Hawar) (see Figure 28)

Lithofacies LF27: Burrowed, Bioturbated, Skeletal Wackestone–Mudstone (BBSW)           

low diversity, abundant, and small-size fauna foraminifera, mollusks, echinoderms, sponge spicules, and ostracods peloids blackened pebbles and blackened grains pyrite wackestone to mudstone with minor packstone interlayer burrows and bioturbation shallow subtidal, low-energy restricted platform inner ramp, restricted lagoonal deposits low porosity and low matrix permeability lower and middle dense zones (see Figure 29)

Sequence Stratigraphy-keyed Lithofacies Distribution Lithofacies distribution in the Lower and Upper Kharaib Reservoir Units is closely linked to the overall sequence-stratigraphic framework. The transgressive systems tracts of both the Lower and Upper Kharaib sequences are clearly more mud dominated (high porosity but low to moderate permeability), whereas the grain-dominated lithofacies types that show excellent porosity and permeability are nearly exclusively restricted to the highstand systems tracts (Figure 4). The identified 16 parasequence sets that build the Lower (4 parasequence sets) and Upper (8 parasequence sets) Kharaib Reservoir Units and the Lower Shuaiba Reservoir Unit (4 parasequence sets) define 16 reservoir subunits; the basic building blocks

for the geological (static) and reservoir (dynamic) model. A generalized southwest–northeast cross section shows the vertical and horizontal lithofacies and facies association distribution of the Lower and Upper Kharaib Reservoir Units as well as the Lower Shuaiba Reservoir Unit in the established sequencestratigraphic framework (Figure 30). The lateral and horizontal variations of the grain-dominated facies associations and lithofacies types in the highstand systems tracts of the Lower and Upper Kharaib sequences are evident (Figure 30).

KHARAIB FORMATION OUTCROP ANALOGS AT WADI RAHABAH, MUSANDAM PENINSULA (RAS AL-KHAIMAH) The Musandam Peninsula consists of a series of upthrusted blocks (Hagab thrust) of Permian to Cretaceous strata. The dominant structures in the peninsula are north – south arcuate thrust faults, associated tear faults, and north –south folds. A slight regional northward dip of the peninsula toward Iran exists. The Ruus Al Jibal forms the backbone of the peninsula, consisting of a series of upthrusted blocks (Hudson, 1960). One of these upthrusted blocks is the 4760-ft (1450-m) high Jabal Hagab to the east of the peninsula opposite to the city of Ras Al-Khaimah. Rock exposures at Wadi Rahabah (north of the city of Ras Al-Khaimah) are time equivalent to the Lower Cretaceous upper Thamama Group, the most prolific onshore oil- and gas-producing zones in Abu Dhabi. The Lower Cretaceous Kahmah Group (Oman) and Musandam Group M4 (Musandam Peninsula) are time equivalent to the Thamama Group (United Arab Emirates, Saudi Arabia, and Bahrain), which contains (from oldest to youngest) the reservoirs of the Habshan, Lekhwair, Kharaib, and Shuaiba formations of the subsurface of the United Arab Emirates. Outcrop analogs of subsurface reservoirs allow for a detailed investigation of the facies architecture and structure of these carbonate bodies, keyed to the overall sequence-stratigraphic framework (Alsharhan and Nairn, 1993; Borgomano et al., 2002; van Buchem et al., 2002; Hillga¨rtner et al., 2003; Immenhauser et al., 2004; Strohmenger et al., 2004b; Suwaina et al., 2004). The 21 lithofacies types that were defined for the subsurface Lower and Upper Kharaib and Lower Shuaiba Reservoir Units and the lower, middle, and upper dense zones (Strohmenger et al., 2004a, b, c) are broadly applicable to the Kharaib and Lower Shuaiba

166 / Strohmenger et al.

FIGURE 30. Southwest – northeast cross section through Field B showing the established sequence-stratigraphic framework and the vertical and horizontal distribution of facies associations (lithofacies 1 [RPR] + lithofacies 2 [RPF]; lithofacies 4 [SPG] + lithofacies 5 [CgSG] + lithofacies 6 [CgASR]; and lithofacies 9 [OSP] + lithofacies 11 [OSW]) and lithofacies types (lithofacies 3: SPP; lithofacies 7: ASPF; lithofacies 8: ASFB; lithofacies 10: SPWP; lithofacies 12: SW; and lithofacies 13: FSW). Transgressive systems tract (TST): green color; highstand systems tract (HST): orange color (see Figure 4).

formations rock exposures. The depositional environments do not occur randomly in time and space. They occur systematically in specific systems tracts of the third-order composite sequences (Strohmenger et al., 2004a, b, c; Suwaina et al., 2004; Vaughan et al., 2004). The southeast wall of the wadi shows a carbonate succession that corresponds to the Lower and Upper Kharaib Reservoir Units (Strohmenger et al., 2004a, b, c; Suwaina et al., 2004) (Figure 31A). Meter-scale lithofacies stacking patterns reflect variations in ba-

thymetry and energy. The top of a parasequence is determined by a flooding surface. A sharp change in lithofacies from shallow high-energy shoals to slightly deeper water muddominated lagoonal deposits is recognized across parasequence boundaries. Muddominated rocks grade upward into packstone and floatstone and subsequently into grainstone and rudstone with abundant rudist debris. These rocks indicate a progressive increase in energy and shallowing of the depositional environment. The top of the parasequence is marked either by well-sorted grainstone and packstone with abundant miliolids or by deposits of the intershoalprotected areas that may become restricted and exhibit mud-dominated rocks with abundant foraminifera, green dasycladacean algae, small mollusks, and sparse rudists. The overlying parasequence commonly indicates a change to deeper and quieter depositional environments as described, or it displays a more gradual deepening with initial high-energy deposits above a reactivation surface. Composite sequence boundary K70_SB at the top of the Upper Kharaib sequence (top Upper Kharaib Reservoir Unit) displays pedogenic overprint of

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 167

FIGURE 31. Southeast wall of Wadi Rahabah outcrop. (A) Shown here is the Kharaib Formation measured section (base

lower dense zone to top upper dense zone; 250 ft [76 m]). (B) The top of the Upper Kharaib Reservoir (Upper Kharaib sequence) displays pedogenic overprint of limestone beds (peds) below sequence boundary K70_SB. (C) Base of lower dense zone showing erosive surface and mud cracks, corresponding to sequence boundary K50_SB.

limestone beds below the sequence boundary (paleosol: peds; Figure 31B). Composite sequence boundary K50_SB at the base of the Lower Kharaib sequence (base lower dense zone) shows an erosive surface displaying desiccation cracks (Figure 31C). Outcrop analogs are a valuable source of data for reservoir characterization. They are particularly useful for visualizing interwell variability because subsurface wellbores represent discrete, widely spaced data. In other words, outcrops provide the continuous, large-scale coverage of a seismic line and provide the fine-scale resolution of core measurements. Modeling input parameters such as lateral and vertical continuity (variogram range), object dimensions (aspect ratios), detailed reservoir architecture (layering), facies relationships, and the nature and extent of diagenetic features can be observed and measured. To better understand reservoir properties of the Upper Thamama producing zones in Abu Dhabi, a geologic model was developed from descriptions of time-equivalent, analogous outcrops in the Emirate of Ras Al-Khaimah. Parasequences, maximum flooding surfaces, sequence boundaries, and lithofacies were described in four measured sections and tied to the

subsurface data of Field B. Porosity measurements derived from logs, core porosity, and core permeability data from an Abu Dhabi oil field (Field B) were applied to similar lithofacies observed in outcrop. These subsurface data were used in the outcrop 3-D model to more precisely reflect subsurface reservoir behavior because outcrops in the United Arab Emirates experienced a significantly different burial history relative to the subsurface reservoirs and, thus, have different porosity and permeability (Strohmenger et al., 2004b; Suwaina et al., 2004). The objective of the reservoir modeling was to investigate the impact of reservoir architecture, porosity-permeability relationships, facies, and scale-up on the development of quantitative 3-D geologic and fluid flow-simulation models. The use of continuously exposed outcrops that are time equivalent and have similar geology to the subsurface Upper Thamama reservoirs of Abu Dhabi provide an opportunity for improving geologic models of these reservoirs. For the stratigraphic interval studied in outcrop (base lower dense zone to top upper dense zone), four composite reservoir facies (three permeable reservoir facies and one dense facies) adequately described the flow behavior. These reservoir facies were modeled

168 / Strohmenger et al.

FIGURE 32. Integration of outcrop data into geological, porosity, and permeability (fluid-flow) models. (A) North wall of Wadi Rahabah showing clinoforms in the Upper Kharaib sequence. (B) Information gathered from outcrop studies (occurrence of clinoforms) should be incorporated into geological (static) and reservoir (dynamic) models. (C) Porosity model showing porosity distribution and the influence of clinoforms. High porosity is shown in red. No porosity is shown in pink. (D) Permeability model showing resulting permeability distribution and the influence of clinoforms. Fluids will encounter more baffles and travel more slowly through these dipping layers than through the flat-lying proportional layers. High permeability is shown in red. No permeability is shown in pink. and were scaled up into a simulation model. In addition to the four measured sections treated as wells, six pseudowells were generated to help constrain the mapping and modeling algorithms. These pseudowells represented copies of the measured sections and contained realistic unit thickness changes based on outcrop observations. Multiple scenarios, including various injection patterns, scale-up methods, and locations of thin, high-permeability streaks, were investigated. Results mimic known behavior in analogous producing fields, and the process of going from rock data to simulation provide a useful training tool for reservoir characterization methods and techniques (Vaughan et al., 2004). The north and south wall of the wadi show clinoforms in the Upper Kharaib sequence. Local clusters of fractures and offset of clinoforms by minor faults can also be observed (Figure 32A). Even if progradational patterns (clinoforms) cannot be identified in core material, the information gathered from out-

crop data (Figure 32A) should be used to build realistic geological (static; Figure 32B) and reservoir (dynamic; Figure 32C, D) models (Strohmenger et al., 2004a, b, c; Suwaina et al., 2004; Vaughan et al., 2004). Porosity (Figure 32C) and permeability (Figure 32D) models show that low-angle clinoforms will act as baffles, and fluid flow through these dipping layers will therefore be slower compared to flat-lying proportional layers. Sector model simulation results address issues of scale-up, stratigraphic architecture, and diagenesis, as well as alternative recovery mechanisms and field development scenarios.

CONCLUSIONS The Kharaib Formation (Barremian and early Aptian) contains two reservoir units (Lower and Upper Kharaib Reservoir Unit) separated and encased by three so-called dense zones (lower, middle, and upper dense zones). It is part of the late transgressive

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 169

sequence set of a second-order supersequence, made up of two third-order composite sequences (base lower dense zone to base upper dense zone or Hawar). The overlying Lower Shuaiba Reservoir Unit belongs to the late transgressive sequence set and the early highstand sequence set and is made up of one third-order composite sequence. Nineteen fourth-order parasequence sets were identified in core and on well-log data. Four parasequence sets build into the Lower Kharaib, ten parasequence sets build into the Upper Kharaib, and five parasequence sets build into the Lower Shuaiba third-order composite sequences (Figure 4). These parasequence sets show predominantly aggradational and progradational stacking patterns, typical of greenhouse cycles (Sarg et al., 1999). The identified, chronostratigraphically bounded parasequence sets were used to subdivide the three reservoir units into a total of 16 subunits. As an alternative sequence-stratigraphic interpretation, the major third-order composite sequence boundaries (Figures 4, 6) may actually occur at the tops of the dense zones (lower, middle, and upper dense zones) and not, as conventionally interpreted (van Buchem et al., 2002) and discussed above, below the dense zones (Figure 6). Thirteen lithofacies types correspond to the three reservoir units (Lower and Upper Kharaib Reservoir Unit and Lower Shuaiba Reservoir Unit) and eight lithofacies types occur in the three dense zones (lower, middle, and upper dense zones; Figures 7–29). Muddominated lithofacies types dominate the transgressive systems tracts, whereas grain-dominated, highly porous and permeable lithofacies types make up the highstand systems tracts of the Lower and Upper Kharaib third-order composite sequences (Figure 30). These lithofacies types are broadly applicable to the Kharaib and Shuaiba formations rock exposures (timeequivalent to the Upper Thamama reservoir units) studied in Wadi Rahabah, Ras Al-Khaimah (United Arab Emirates). Many of the so-called stylolitic intervals correspond to major facies changes related to third-, fourth-, and fifth-order sequence boundaries, parasequence set boundaries, and parasequence boundaries. Early diagenetic processes follow the sequence-stratigraphic framework and, therefore, can be predicted away from well control. Outcrops of the Kharaib Formation at Wadi Rahabah are of seismic scale and, thus, provide fieldscale cross sections that help to refine sequencestratigraphic and facies models and aid to the understanding of reservoir geometry distribution and

reservoir continuity in the subsurface (Vaughan et al., 2004) (Figures 31, 32). Outcrops are particularly useful for filling the holes common in subsurface well data sets used as input to geologic (static) and reservoir (dynamic) models.

ACKNOWLEDGMENTS The authors gratefully acknowledge the managements of Abu Dhabi Company for Onshore Oil Operations (ADCO), Abu Dhabi National Oil Company (ADNOC), and ExxonMobil Exploration Company for permission to publish this study. We thank Prasad Patannakara (ADCO) and his team for drafting the figures. Appreciation is extended to Ismail Al-Mansouri, Majeed Haq, Mohammed Hilal, Hassan Ahmed, and Abdulsalam Khoury from ADCO core facility (Mussafah) for their help in preparing and laying out the core material and for preparing thin sections. Sean A. Guidry, Christine I. Gonzalez, Martin A. Herrmann, Lisa A. Roehl, Chengije Liu, Susan M. Agar, Jerry J. Kendall, and Mike G. Kozar (ExxonMobil), as well as Andrew Clark, Srikant Guruswamy, Hafez H. Hafez, Jorge S. Gomes, Michel J.-M. Rebelle, William L. Soroka, and Mohammed Ayoub (ADCO) are thanked for their valuable discussions. We extend special thanks to Omar Suwaina (ADNOC) for his valuable contributions to this study. We appreciate the helpful comments and suggestions of reviewers Jose´ E. de Matos, Ju ¨ rgen Gro ¨ tsch, and Jonathan Kaufman, which greatly improved the manuscript.

REFERENCES CITED Alsharhan, A. S., 1989, Petroleum geology of the United Arab Emirates: Journal of Petroleum Geology, v. 12, no. 3, p. 253 – 288. Alsharhan, A. S., and A. E. M. Nairn, 1993, Carbonate platform models of Arabian Cretaceous reservoirs, in J. A. T. Simo, R. W. Scott, and J.-P. Masse, eds., Cretaceous carbonate platforms: AAPG Memoir 56, p. 173 – 184. Alsharhan, A. S., and A. E. M. Nairn, 1997, Sedimentary basins and petroleum geology of the Middle East: Amsterdam, Elsevier, 843 p. Boichard, R., A. S. Al Suwaidi, and H. Karakhanian, 1994, Sequence boundary types and related porosity evolutions: Example of the Upper Thamama Group in field ‘‘A’’ (offshore Abu Dhabi, UAE): 6th Abu Dhabi International Petroleum Exhibitions & Conference, Abu Dhabi Society of Petroleum Engineers Paper 76, Abu Dhabi, p. 417 – 428. Borgomano, J., J.-P. Masse, and S. Al Maskiry, 2002, The lower Aptian Shuaiba carbonate outcrops in Jebel Akhdar, northern Oman: Impact on static modeling

170 / Strohmenger et al. for Shuaiba petroleum reservoirs: AAPG Bulletin, v. 86, no. 9, p. 1513 – 1529. Burgess, C. J., and C. K. Peter, 1985, Formation, distribution and prediction of stylolites as permeability barriers in the Thamama Group, Abu Dhabi: 4th Middle East Oil Show, Bahrain, Society of Petroleum Engineers Paper 13698, p. 165 – 174. Davies, R. B., D. M. Casey, A. D. Horbury, P. R. Sharland, and M. D. Simmons, 2002, Early to mid-Cretaceous mixed carbonate-clastic shelfal systems: Examples, issues and models from the Arabian plate: GeoArabia, v. 7, no. 3, p. 541 – 598. Droste, H., and M. van Steenwinkel, 2004, Stratal geometries and patterns of platform carbonates: The Cretaceous of Oman, in G. P. Eberli, J. L. Masaferro, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 185 – 206. Dupraz, C., and A. Strasser, 1999, Microbialites and microencrusters in shallow coral bioherms (middle – late Oxfordian, Swiss Jura Mountains): Facies, v. 40, p. 101 – 130. Galloway, W. E., 1989, Genetic stratigraphic sequences in basin analysis: I — Architecture and genesis of flooding-surface bounded depositional units: AAPG Bulletin, v. 73, no. 2, p. 125 – 142. Granier, B., 2000, Lower Cretaceous stratigraphy of Abu Dhabi and the United Arab Emirates — A reappraisal: 9th Abu Dhabi International Petroleum Exhibition & Conference Proceedings, Paper 0918, Abu Dhabi, p. 526 – 535. Granier, B., A. S. Al Suwaidi, R. Busnardo, S. K. Aziz, and R. Schroeder, 2003, New insight on the stratigraphy of the ‘‘Upper Thamama’’ in offshore Abu Dhabi (U.A.E.): Carnets de Ge´ologie, Article 2003/5, p. 1 – 17. Gro ¨ tsch, J., O. Al-Jeelani, and Y. Al-Mehairi, 1998a, Integrated reservoir characterization of a giant Lower Cretaceous oil field, Abu Dhabi, U.A.E.: 8th Abu Dhabi International Petroleum Exhibition & Conference Proceedings, Society of Petroleum Engineers Paper 49454, Abu Dhabi, p. 77 – 86. Gro ¨ tsch, J., I. Billing, and V. Vahrenkamp, 1998b, Carbonisotope stratigraphy in shallow-water carbonates: implications for Cretaceous black-shale deposition: Sedimentology, v. 45, no. 4, p. 623 – 634. Haq, B. U., and A. M. Al-Qahtani, 2005, Phanerozoic cycles of sea-level change on the Arabian platform: GeoArabia, v. 10, no. 2, p. 127 – 160. Haq, B. U., J. Hardenbol, and P. R. Vail, 1988, Mesozoic and Cenozoic chronostratigraphy and cycles of sealevel change, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 71 – 108. Hillga¨rtner, H., F. S. P. van Buchem, F. Gaumet, P. Razin, B. Pittet, J. Gro ¨ tsch, and H. Droste, 2003, The Barremian – Aptian evolution of the eastern Arabian carbonate platform margin (northern Oman): Journal of Sedimentary Research, v. 73, no. 5, p. 756 – 773.

Hudson, R. G. S., 1960, The Permian and Triassic of the Oman Peninsula, Arabia: Geological Magazine, v. 18, no. 4, p. 299 – 309. Immenhauser, A., H. Hillga¨rtner, and E. Van Bentum, 2005, Microbial-foraminiferal episodes in the early Aptian of the southern Tethyan margin: Ecological significance and possible relation to oceanic anoxic event 1a: Sedimentology, v. 52, no. 1, p. 77 – 99. Immenhauser, A., et al., 2004, Barremian – lower Aptian Qishn Formation, Haushi-Huqf area, Oman: A new outcrop analogue for the Kharaib/Shu’aiba reservoirs: GeoArabia, v. 9, no. 1, p. 153 – 194. Koepnick, R. B., 1984, Distribution and vertical permeability of stylolites within a Lower Cretaceous carbonate reservoir, Abu Dhabi, United Arab Emirates — Stylolites and associated phenomena: Relevance to hydrocarbon reservoirs, Abu Dhabi, U.A.E.: Abu Dhabi Reservoir Research Foundation Special Publications, p. 261 – 278. Melville, P., O. Al Jeelani, S. Al Menhali, and J. Gro ¨ tsch, 2004, Three-dimensional seismic analysis in the characterization of a giant carbonate field, onshore Abu Dhabi, United Arab Emirates, in G. P. Eberli, J. L. Masaferro, and J. F. Sarg, eds., Seismic imaging of carbonate reservoir systems: AAPG Memoir 81, p. 123– 148. Mitchum Jr., R. M., 1977, Seismic stratigraphy and global changes in sea level: Part 11 — Glossary of terms used in seismic stratigraphy, in C. E. Payton, ed., Seismic stratigraphy — Applications to hydrocarbon exploration: AAPG Memoir 26, p. 205 – 212. Nelson, R. A., 1984, Geologic analysis of naturally fractured reservoirs: Contributions in petroleum geology and engineering: Oxford, Gulf Publishing Company, 352 p. Park, W. C., and E. H. Schot, 1968, Stylolites: Their nature and origin: Journal of Sedimentary Petrology, v. 38, p. 175 – 191. Pittet, B., F. van Buchem, H. Hillga¨rtner, J. Gro ¨ tsch, and H. Droste, 2002, Ecological succession, paleoenvironmental change, and depositional sequences of Barremian – Aptian shallow-water carbonates in northern Oman: Sedimentology, v. 49, no. 3, p. 555– 581. Rebelle, M., C. J. Strohmenger, A. Ghani, K. Al-Mehsin, and A. Al-Mansouri, 2004, Lower Cretaceous Upper Thamama reservoir high-resolution sequence stratigraphy, United Arab Emirates (abs.): GeoArabia, v. 9, no. 1, p. 136. Sarg, J. F., 1988, Carbonate sequence stratigraphy, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 155 – 181. Sarg, J. F., J. R. Markello, and L. J. Weber, 1999, The secondorder cycle, carbonate-platform growth, and reservoir, source, and trap prediction, in P. M. Harris, A. H. Saller, and J. A. T. Simo, eds., Advances in carbonate sequence stratigraphy: Application to reservoirs, outcrops, and models: SEPM Special Publication 63, p. 11 – 34. Sattler, U., A. Immenhauser, H. Hillga¨rtner, and M. Esteban, 2005, Characterization, lateral variability and lateral

Sequence Stratigraphy and Reservoir Characterization of Upper Thamama Reservoirs / 171 extend of discontinuity surfaces on a carbonate platform (Barremian to lower Aptian, Oman): Sedimentology, v. 52, no. 2, p. 339 – 361. Sharland, P. R., R. Archer, D. M. Casey, R. B. Davies, S. H. Hall, A. P. Heward, A. D. Horbury, and M. D. Simmons, 2001, Arabic plate sequence stratigraphy: GeoArabia Special Publication 2, 371 p. Sharland, P. R., R. Archer, D. M. Casey, R. B. Davies, M. D. Simmons, and O. E. Sutcliffe, 2004, Arabic plate sequence stratigraphy — Revisions to SP2: GeoArabia, v. 9, no. 1, p. 199 – 214. Strasser, A., 1984, Black-pebble occurrence and genesis in Holocene carbonate sediments (Florida Keys, Bahamas, and Tunisia): Journal of Sedimentary Petrology, v. 54, no. 4, p. 1097 – 1109. Strasser, A., and E. Davaud, 1983, Black pebbles of the Purbeckian (Swiss and French Jura): Lithology, geochemistry and origin: Eclogae Geologicae Helveticae, v. 76, p. 551 – 580. Strohmenger, C. J., L. J. Weber, A. Ghani, K. Al-Mehsin, and O. Al-Jeelani, 2004a, Sequence stratigraphy and reservoir characterization of the Lower Cretaceous Kharaib Formation, Abu Dhabi (abs.): GeoArabia, v. 9, no. 1, p. 136. Strohmenger, C. J., L. J. Weber, A. Ghani, M. Rebelle, K. AlMehsin, O. Al-Jeelani, A. Al-Mansoori, and O. Suwaina, 2004b, High-resolution sequence stratigraphy of the Kharaib Formation (Lower Cretaceous, U.A.E.): 11th Abu Dhabi International Petroleum Exhibition & Conference Proceedings, Society of Petroleum Engineers Paper 88729, Abu Dhabi, 10 p. Strohmenger, C. J., L. J. Weber, A. Ghani, A. Al-Mansoori, K. Al-Mehsin, O. Suwaina, O. Al-Jeelani, and L. Vaughan, 2004c, Sequence stratigraphy and reservoir characterization of the Kharaib Formation comparing outcrop and subsurface data (Lower Cretaceous, U.A.E.) (abs.): AAPG International Conference and Exhibition (Abstracts), Cancun, p. A75. Suwaina, O. A., L. J. Weber, C. J. Strohmenger, L. Vaughan, A. Al-Mansoori, S. Khan, and A. Ghani, 2004, Sequence stratigraphy and reservoir characterization of the Thamama reservoirs and outcrop equivalents: A core workshop and field seminar (abs.): GeoArabia, v. 9, no. 1, p. 137 – 138. Vahrenkamp, V. C., 1996, Carbon isotope stratigraphy of the Upper Kharaib and Shuaiba formations: Implications for the Early Cretaceous evolution of the Arabian Gulf region: AAPG Bulletin, v. 80, no. 5, p. 647 – 662. Vail, P. R., 1987, Seismic stratigraphy interpretation using sequence stratigraphy: Part 1. Seismic stratigraphy interpretation procedure, in A. W. Bally, ed., Atlas of seismic stratigraphy, v. 1: AAPG Studies in Geology 27, p. 1 – 10. Vail, P. R., R. M. Mitchum Jr., R. G. Todd, J. M. Widmier, S. Thompson III, J. B. Sangree, J. N. Bubb, and W. G. Hatlelid, 1977, Seismic stratigraphy and global changes

of sea level: Part 1. Overview, in C. E. Payton, ed., Seismic stratigraphy — Applications to hydrocarbon exploration: AAPG Memoir 26, p. 49 – 212. Vail, P. R., F. Audemard, S. A. Bowman, P. N. Eisner, and C. Perez-Cruz, 1991, The stratigraphic signatures of tectonics, eustasy and sedimentology — An overview, in G. Einsele, W. Ricken, and A. Seilacher, eds., Cycles and events in stratigraphy: Berlin, Springer, p. 617 – 659. van Buchem, F. S. P., B. Pittet, H. Hillga¨rtner, J. Gro ¨ tsch, A. I. Al Mansouri, I. M. Billing, H. H. J. Droste, W. H. Oterdoom, and M. van Steenwinkel, 2002, Highresolution sequence stratigraphic architecture of Barremian/Aptian carbonate systems in northern Oman and the United Arab Emirates (Kharaib and Shuaiba formations): GeoArabia, v. 7, no. 3, p. 461 – 500. Van Wagoner, J. C., R. M. Mitchum Jr., H. W. Posamentier, and P. R. Vail, 1987, Seismic stratigraphy interpretation using sequence stratigraphy: Part 2. Key definitions of sequence stratigraphy, in A. W. Bally, ed., Atlas of seismic stratigraphy: AAPG Studies in Geology 27, v. 1, p. 11 – 14. Van Wagoner, J. C., H. W. Posamentier, R. M. Mitchum Jr., P. R. Vail, J. F. Sarg, T. S. Loutit, and J. Hardenbol, 1988, An overview of the fundamentals of sequence stratigraphy and key definitions, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 39 – 46. Vaughan, R. L., S. A. Khan, L. J. Weber, O. Suwaina, A. AlMansoori, A. Ghani, C. J. Strohmenger, M. A. Herrmann, and D. Hulstrand, 2004, Integrated characterization of UAE outcrops: From rocks to fluid flow simulation: 11th Abu Dhabi International Petroleum Exhibition & Conference Proceedings, Society of Petroleum Engineers Paper 88730, Abu Dhabi, 17 p. Yose, L. A., S. Bachtel, L. J. Weber, C. J. Strohmenger, A. Al-Mansoori, and O. Suwaina, 2004a, Integrated approaches to carbonate reservoir characterization and prediction: Examples from the United Arab Emirates fields and outcrops (abs.): GeoArabia, v. 9, no. 1, p. 146. Yose, L. A., et al., 2004b, New frontiers in 3D seismic characterization of carbonate reservoirs: Examples from a supergiant field in Abu Dhabi: 11th Abu Dhabi International Petroleum Exhibition & Conference Proceedings, Society of Petroleum Engineers Paper 88689, Abu Dhabi, 16 p. Yose, L. A., A. S. Ruf, C. J. Strohmenger, J. S. Schuelke, A. Gombos, I. Al-Hosani, S. Al-Maskary, G. Bloch, Y. AlMehairi, and I. G. Johnson, 2006, Three-dimensional characterization of a heterogeneous carbonate reservoir, Lower Cretaceous, Abu Dhabi (United Arab Emirates), in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 173 – 212.

5

Yose, L. A., A. S. Ruf, C. J. Strohmenger, J. S. Schuelke, A. Gombos, I. Al-Hosani, S. Al-Maskary, G. Bloch, Y. Al-Mehairi, and I. G. Johnson, 2006, Threedimensional characterization of a heterogeneous carbonate reservoir, Lower Cretaceous, Abu Dhabi (United Arab Emirates), in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 173 – 212.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Lower Cretaceous, Abu Dhabi (United Arab Emirates) Lyndon A. Yose

Jim S. Schuelke

ExxonMobil Qatar, Inc., Doha, Qatar

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

Amy S. Ruf Andy Gombos

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

Christian J. Strohmenger, Ismail Al-Hosani, Shamsa Al-Maskary, Gerald Bloch, and Yousuf Al-Mehairi

Imelda G. Johnson ExxonMobil Exploration Company, Houston, Texas, U.S.A.

Abu Dhabi Company for Onshore Oil Operations, Abu Dhabi, United Arab Emirates

ABSTRACT

H

igh-effort three-dimensional (3-D) seismic data collected by the Abu Dhabi Company for Onshore Oil Operations (ADCO) are some of the highest quality data ever collected for a carbonate field. The 3-D seismic data were integrated with core and log data to develop a new, volume-based framework for enhanced reservoir characterization. The Lower Cretaceous (Aptian) reservoir is positioned over a platform-to-basin transition and records a diverse range of depositional facies and stratal geometries. Reservoir properties vary predictably based on position along the platform-to-basin profile and position in the sequence-stratigraphic framework. The Aptian reservoir interval (Shuaiba Formation) records a second-order sequence set that is divided into five depositional sequences. Sequences 1 and 2 were deposited during the transgressive phase of the sequence set. These sequences are retrogradational, record the initial formation of a low-relief ramp,

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215877M882562

173

174 / Yose et al.

and are dominated by algal-prone facies. Ramp interior and margin facies of the transgressive phase are characterized by high porosity and low permeability because of mud-dominated textures and development of microporosity. Sequence 3 was deposited during the highstand phase of the sequence set, is mainly aggradational, and records the proliferation or rudists across the platform top. Grain-dominated platform interior and margin facies of the highstand phase are the highest quality reservoir facies in the Shuaiba reservoir. Sequences 4 and 5 were deposited during the late highstand phase of the sequence set. These sequences are progradational and record the progressive downstepping of the platform margin onto a low-angle (1 – 28) slope. Clinoforms of the late highstand phase are characterized by alternations of high and low reservoir quality developed in response to relative sea level changes. Sequence 6 was deposited during the second-order lowstand and forms the base of the next overlying sequence set. Sequence 6 is composed primarily of finegrained siliciclastics and is a nonreservoir. Results from the study have led to an improved understanding of platform evolution and a volume-based framework for reservoir characterization. The integrated data set provides new insights on platform paleogeography, carbonate facies architecture, and the geometry and mechanisms of carbonate platform progradation. In the platform interior area, 3-D seismic data reveal a complex mosaic of tidal channels, high-energy rudist shoals, and intershoal ponds that impact reservoir sweep and conformance. At the basin margin, the seismic data provide high-definition images of platform-margin clinoforms that impact reservoir architecture and well-pair connectivity. Business applications of the volume-based reservoir framework include (1) use of 3-D seismic visualization technology for optimizing well placement, identifying bypassed reservoirs, and evaluating reservoir connectivity; (2) integration of quantitative, volume-based seismic information into reservoir models; (3) maximizing recovery through full integration of all subsurface data; and (4) enhanced communication among geoscientists and engineers, leading to improved reservoir management practices.

INTRODUCTION The stratigraphic and diagenetic complexities inherent in carbonate reservoirs require accurate reservoir descriptions and models to optimize recovery. Three-dimensional (3-D) seismic data provide the only continuous source of information on reservoir properties in the subsurface. With increasingly more 3-D seismic data available over carbonate fields, the challenge is to maximize the value of the seismic data for characterization of carbonate reservoir architecture and rock properties (Eberli et al., 2003; Sarg and Schuelke, 2003; Masafarro et al., 2004; Yose et al., 2004). This study used high-quality seismic data collected over a supergiant onshore field in Abu Dhabi to illustrate the value of 3-D seismic data for carbonate reservoir characterization. The field produces mainly from the Lower Cretaceous (Aptian) Shuaiba reser-

voir and has been under production since the 1960s. As the field matures, a more detailed understanding of the reservoir is required to optimize recovery. To address this need, the Abu Dhabi Company for Onshore Oil Operations (ADCO) acquired a higheffort, 3-D seismic survey over the field. A joint study between ADCO and ExxonMobil was undertaken to fully integrate the 3-D seismic data with core, well, and production data, leading to a new volume-based reservoir description. The integrated data set provides new perspectives on carbonate platform evolution, the response of carbonate systems to sea level change, and the influence of carbonate facies and stratal architecture on reservoir properties and architecture. The volume-based reservoir framework provides a powerful tool for evaluating full-field production and performance issues, leading to improved reservoir management.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 175

FIGURE 1. Regional map showing major hydrocarbon accumulations in the United Arab Emirates and the locations of Field A1 and Field A2. The fields are on separate structures, but both produce from the Shuaiba reservoir. The box around the studied fields in the inset map highlights the area covered in Figure 2.

STUDY AREA Field Location The location of the study area in the United Arab Emirates is shown in Figure 1. Onshore Abu Dhabi includes five supergiant fields, each of which is located over large north-northeast–trending anticlinal structures. The present study is focused on the Lower Cretaceous Aptian reservoir interval (Shuaiba Formation) of one of the onshore fields (Field A1 in Figure 1). As shown in Figure 2, Field A1 occurs over a large, doubly plunging anticline that is about 35  20 km (22  12.5 mi) in dimension. A small field is also present just south of Field A1 (Field A2 in Figure 2) that occurs over a separate and much smaller closure. Fields A1 and A2 were evaluated jointly in this study because the data sets overlap, and both fields are producing mainly from the Shuaiba reservoir. Field A1, discovered in 1962, is estimated to contain more than 20 billion bbl of original oil in place

and is one of the ten largest carbonate fields in the world. The field has been under peripheral water injection since the mid-1970s. Pattern gas and waterinjection schemes are also being implemented in parts of the field. A more detailed understanding of reservoir architecture and connectivity is required to optimize the new injection strategy and future field development to maximize recovery.

Database The study area provides a world-class data set and presents an opportunity to develop and test high-end reservoir characterization technologies. Greater than 500 wells exist in the field, more than 100 of which are cored. Most wells contain a standard suite of openhole logs. More than 40 yr of production and performance data are available to validate geologic interpretations and calibrate reservoir models. Despite the high number of wells in the field, the average well

176 / Yose et al.

FIGURE 2. Time structure map on the top of the reservoir interval at Field A1. The field is located over a large anticlinal structure that plunges to the north and south. Field A2 is a small field on a separate closure developed just south of Field A1. As shown, separate seismic surveys exist over each field.

spacing in the field is 1 km (0.62 mi). A 3-D seismic survey was recently acquired over Field A1, covering an area of nearly 2000 km2 (772 mi2). The 3-D seismic data are 300-fold with 25-m (82-ft) trace spacing and have a peak frequency of 30 –35 Hz at reservoir depth (2440–2745 m; 8000– 9000 ft). A separate, but overlapping, seismic survey was also recently acquired over Field A2 (Figure 1). Seismic data in the southern survey are 150-fold and also provide high-quality imaging of the Shuaiba reservoir. Subsequent figures show extracts from the merged seismic surveys to

highlight stratigraphic and paleogeographic variations across the entire study area. Data evaluated as part of the present study include the 3-D seismic surveys, a strike and dip cross section grid that intersects more than 100 wells, 25 cored wells, and available production and performance data.

Regional Stratigraphy and Paleogeography The stratal hierarchy and nomenclature used in the present study is shown in Figure 3. The Shuaiba reservoir interval is part of a second-order supersequence

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 177

FIGURE 3. Chronostratigraphic chart for the Lower Cretaceous of the United Arab Emirates. The Lower Cretaceous comprises a hierarchy of second- and third-order depositional cycles. The reservoir interval at Field A1 spans most of the Aptian and forms the second-order Aptian sequence set. The Aptian sequence set, in turn, is divided into five thirdorder depositional sequences (1 – 5). A sixth depositional sequence (6) forms a composite lowstand sequence and is part of the next overlying supersequence. OAE1a is a globally recognized oceanic anoxic event and correlates to the maximum flooding interval of the Lower Cretaceous supersequence (coincident with the maximum flooding interval of the Aptian sequence set). The Hardenbol et al. (1998) time sequence terminology is keyed to a nannofossil zonation. Nannofossil dating in the present study confirms the position of the Ap 3, Ap 4, and Ap 5 time sequences (x in the timesequence nomenclature means unconfirmed). MFS = maximum flooding surface. that spans the Valanginian to Aptian stages and can be correlated across the Arabian platform (Sharland et al., 2001). Based on interpretation of facies and their stacking patterns in the present study, and by Strohmenger et al. (2006), the Lower Cretaceous supersequence is subdivided into three second-order sequence sets (Figure 3). The Shuaiba reservoir is within the Aptian sequence set, culminating in the secondorder supersequence boundary at the top. A comparison of the sequence-stratigraphic – based framework with the lithostratigraphic nomenclature is shown in Figure 3. Lithostratigraphic nomenclature in the Shuaiba varies regionally, underscoring the need for a regional, sequence-stratigraphic framework based on chronostratigraphy. The Aptian paleogeography had a pronounced influence on depositional patterns in the study area. As

shown in Figure 4, Field A1 straddles the margin of the Bab basin. Accordingly, the Shuaiba reservoir records a range of platform, margin, slope, and basin environments and is ideally positioned to record the full evolution of the platform-basin system. The Bab basin differentiated in the early Aptian as a shallow intrashelf seaway in the central Arabian platform (Figure 4) (Alsharhan, 1985; Alsharhan and Nairn, 1993). The basin is bounded by an extensive carbonate platform, and restricted, organic-rich facies were deposited within the Bab basin proper. Carbonate platform and margin facies in the Shuaiba form a regional reservoir system, and organic-rich basinal facies provide a potential source rock. The Bab basin is filled by prograding carbonates and siliciclastics that were deposited during the late Aptian sea level lowstand. Shales and argillaceous carbonates of the Nahr

178 / Yose et al.

FIGURE 4. Aptian paleogeographic framework showing the extent of the Bab basin. Field A1 straddles the platform-basin transition. A south-facing margin is also present in Field A2, just south of Field A1. The northern extent of the Bab basin is uncertain. The basin may have been open to the north and connected to the Neo-Tethys open ocean, or isolated in the Arabian platform and surrounded by a shallow, carbonate shelf (Sharland et al., 2001).

Umr Formation were deposited during the subsequent transgression and form a regional top seal over the Shuaiba reservoir.

WORKFLOW Improved recovery from carbonate reservoirs requires a 3-D understanding of reservoir architecture and properties. The 3-D seismic visualization environment provides a powerful platform for data integration and evaluation and a catalyst for multidisciplinary interactions across geoscience and engineering boundaries. A volume-based reservoir characterization workflow was applied to maximize information from the 3-D seismic data (Figure 5). Several key elements of the workflow will be highlighted, including (1) poststack optimization of seismic data; (2) calibration of seismic with core and log data; (3) application of volume interpretation techniques for rapid and accurate definition of reservoir frameworks; (4) direct detection and prediction of reservoir properties from seismic data; and (5) integration of seismic data into reservoir models. As illustrated in Figure 6, a combination of volume types and corendering techniques are used to maximize the information extracted from the 3-D seismic data. The approach used in this study included a combination of poststack filtering of the data to improve

the signal-to-noise ratio (compare Figure 6A, B), selection and generation of optimum seismic attribute volumes to image geologic features (Figure 6B), and corendering of multiple seismic attributes to bring out more detail (Figure 6C). The seismic discontinuity volume shown in Figure 6B highlights the edges of structural and stratigraphic features and is one of the most useful volumes for characterization of the Shuaiba reservoir. The volume mathematically quantifies trace-to-trace variations in the seismic data, with areas of high continuity shown in white and areas of low continuity shown in black. In the current example, facies geometries are resolved with clarity comparable to aerial photographs of modern carbonate systems (Figure 6D). Calibration of the enhanced seismic data with core and well data is required to understand the geologic origin of seismic features and their impact on field performance. In this study, images, such as those shown in Figure 6, were used for optimum selection of wells and cores to sample the full range of seismic variability. Seismic information on reservoir architecture and rock properties is compared with reservoir performance data to validate predictions and assess new opportunities. Seismic interpretations and predictions are also incorporated into the reservoir model to guide reservoir framework development, distribute reservoir features extracted from the seismic

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 179

FIGURE 5. Integrated seismic characterization workflow. Visualization technology allows for improved data integration and evaluation during all stages of the workflow.

data, and to estimate rock properties outside of well control. Many elements of the seismic workflow are applicable to other carbonate reservoirs, although the details will vary on a reservoir-specific basis, depending on such factors as reservoir complexity, data quality and availability, and the business purpose.

STRUCTURAL FRAMEWORK The impact of faults and fractures on reservoir performance is a key uncertainty for many large, lowrelief structures in the Middle East, including onshore fields in Abu Dhabi. Prior to collection of 3-D seismic data over the study area, the distribution and geometry of faults were not well understood (Alsharhan, 1993; Marzouk and Sattar, 1993). As illustrated in Figure 7, the 3-D seismic data provide a new level of detail on the structural framework. Several different seismic attribute volumes were evaluated as part of the structural analysis, including discontinuity, azimuthal amplitude variation with offset, and fault enhancement volumes. The seismic discontinuity volume (Figure 7) reveals a network of dominantly northwest-trending faults that intersect the Lower

Cretaceous reservoirs. In the study area, faults are organized into a system of grabens that are best developed in the central and northern regions of the field. The faults are low offset, most commonly with less than 18 ms (40 m; 130 ft) of vertical displacement, and are commonly arranged en echelon along strike (lateral offsets). These faults are consistent with a regional fault system that has been mapped across onshore and offshore Abu Dhabi (Marzouk and Sattar, 1993; Johnson et al., 2005). The regional faults are interpreted as low-offset wrench faults that developed in response to west-northwest compression during the Late Cretaceous ( Johnson et al., 2005). The impact of the faults and associated fractures on fluid flow and reservoir connectivity in the study area is uncertain at present. The Shuaiba is a matrixdominated flow system, but fracturing could enhance fluid flow in some parts of the field and may become more pronounced with time. As will be discussed further, premature water breakthrough along the northern platform-margin trend may be in part related to enhanced fluid flow along faults. Additional work and data are required to further evaluate the impact of the faults and associated fractures on fluid flow.

180 / Yose et al.

FIGURE 6. Impact of data optimization on imaging of reservoir features in the platform interior area. The images of (A – C) are identical time slices from three different attribute volumes. (A) Time slice of the original, unfiltered amplitude data with reservoir features poorly defined. (B) Enhanced seismic volume that incorporates ExxonMobil proprietary filtering and seismic attribute technologies (see text for explanation). (C) Image of seismic discontinuity corendered with the isochron of the reservoir interval. Isochron thins highlight lower porosity areas (faster velocities), and isochron thicks highlight higher porosity areas (slower velocities). The corendered image brings out even more detail on the platform architecture and qualitative indications of reservoir quality. (D) Aerial photograph of modern carbonates of the Great Barrier Reef illustrating ponds and channels similar to those imaged in seismic data (photo courtesy of National Geographic Society).

The new 3-D fault framework can be used to monitor future production trends and to guide future data collection.

SEQUENCE-STRATIGRAPHIC OVERVIEW Stratigraphic variability is the primary control on Shuaiba reservoir architecture in the study area. An overview of the new sequence model for the Shuaiba is summarized in Figures 3 and 8. Regional perspec-

tives on stratigraphic and paleogeographic variations in the Shuaiba, including summaries of previous work, are provided in Sharland et al. (2001), van Buchem et al. (2002), and Greselle and Pittet (2005).

Sequence Age Dating Age control in the study area is based on unpublished work by I. Billing and E. J. Oswald (1995, personal communication) and new age dating conducted by ExxonMobil. Age dates from I. Billing and E. J.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 181

FIGURE 7. Seismic characterization of structural framework. Image shown includes a 3-D discontinuity probe cut by a vertical seismic amplitude profile. Most of the linear, black discontinuities on the surface of the 3-D cube are faults (fault interpretations shown in inset). Faults are low offset, are best developed in the platform-margin area, and trend parallel to the depositional strike of the margin. Oswald (1995, personal communication) are based on a combination of macrofossil dating (mainly rudists) and carbon isotopes and provide an important constraint on the position of the lower–upper Aptian boundary across the study area. The ExxonMobil age dating is based on nannofossils that established ties to the global Aptian time sequences defined in Hardenbol et al. (1998) (Ap terminology in Figure 3) and provided a higher resolution time-stratigraphic zonation for the lower and upper Aptian. Results from the age dating are summarized in Figure 3. Sequence 1 is of Ap 2 age, sequences 2 and 3 are of Ap 3 age, sequence 4 is of Ap 4 age, and sequence 5 is of probable Ap 5 age (T. C. Huang, 2004, personal communication). The age of sequence 6 is not confirmed, but is interpreted as Ap 6 time. The overlying Nahr Umr Shale is Albian in age based on regional age dates (Sharland et al., 2001). The position of the Aptian stage boundaries in the study area is uncertain (Figure 3). The Barremian–Aptian boundary is interpreted to be within the Kharaib Formation (Strohmenger et al., 2006), and the Aptian–Albian

boundary is interpreted to be within the Bab Member of the Shuaiba. These results are generally consistent with age dates reported in Azer and Toland (1993), Hughes (2000), Sharland et al. (2001), and van Buchem et al. (2002). Based on these results, the time span of the Aptian sequence set is estimated to be on the order of 7 –9 m.y.

Stratal Hierarchy and Reservoir Impact A hierarchy of stratigraphic cyclicity is developed within the Shuaiba that has a pronounced influence on reservoir properties. Stratal geometries developed within the Aptian sequence set record a complete transgressive-regressive cycle that developed in response to a second-order rise and fall in sea level (Figures 3, 8). A similar depositional pattern is recognized regionally within the Aptian (Sharland et al., 2001; van Buchem et al., 2002; Droste, 2004). Secondorder stacking patterns control the overall architecture of the Shuaiba platform. The Aptian sequence set is subdivided into three major phases of deposition, corresponding to the

182 / Yose et al.

FIGURE 8. Sequence- and seismic-stratigraphic overview of the Aptian sequence set. See Figure 13 for explanation facies codes. (A) Sequence-stratigraphic framework and reservoir implications. The reservoir is subdivided into four major phases of deposition, corresponding to phases of a long-term (second-order) sea level cycle, and six depositional sequences (1– 6), corresponding to third-order sea level changes. Wells A – D are type logs for distinct areas of the field and are shown in Figures 17 – 20, respectively. (B) Seismic geometries displayed on 2-D vertical slice (amplitude data, flattened on base reservoir). Reservoir top is highlighted in blue, and reservoir base is highlighted in green. Amplitude anomalies in the platform interior correspond to channel and pond features, as shown in time slice views in Figures 6 and 10.

transgressive, early highstand, and late highstand phases of the second-order sea level cycle (Figure 8). A subsequent lowstand phase forms the base of the overlying second-order sequence set. Stratal geometry, stacking patterns, and facies development vary systematically between each major phase of deposition (Figure 8). The Aptian sequence set is further subdivided into a series of five depositional sequences that are each interpreted to record a third-order cycle of relative sea level rise and fall. Third-order sequences control the distribution of major reservoir barriers and flow compartments in the Shuaiba reservoir (Figure 8A). Third-order sequences are partitioned within the second-order sequence set as follows: sequences 1 and 2 comprise the transgressive phase; sequence 3 comprises the highstand phase; and sequences 4 and 5 comprise the late highstand phase. Sequence 6 forms the lowstand phase of the next overlying sequence set. As shown in Figure 9, third-order depositional sequences are divided into systems tracts that are tied

to different phases of the third-order relative sea level cycles. Most sequences display well-developed transgressive and highstand systems tracts (TST, HST), although the character of systems tracts varies significantly based on position in the second-order cycle. Higher frequency (fourth- or fifth-order) cycles are best developed within the HSTs of third-order sequences and control the distribution and character of flow units. Lowstand systems tracts (LST) (e.g., basinward shifts of the carbonate system) are generally absent in the transgressive and early highstand phases of the Aptian sequence set because of a combination of long-term accommodation increase, and the lack of high-amplitude, shorter term sea level changes during the Cretaceous greenhouse period. Lowstand systems tracts are developed during the late highstand and lowstand phases of the Aptian sequence set, as longterm sea level was falling, and are represented by basinward shifts in carbonate sedimentation. At the second-order scale, sequences 4–6 are all basinally restricted sequences (Figure 8). At the third-order scale,

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 183

FIGURE 9. Time-stratigraphic evolution of the Aptian sequence set, showing the sequential evolution of the carbonate buildup. See Figure 13 for explanation of facies codes. HST = highstand systems tract; LST = lowstand systems tract; TST = transgressive systems tract. lowstand and TSTs that developed during the late highstand are difficult to distinguish because both systems tracts are basinally restricted caused by longterm sea level fall (see sequences 4 and 5, Figure 8).

platform interior (also see Figure 6). The width of the platform interior (along depositional profile) varies throughout time, but is approximately 20 km (12.5 mi) wide in the seismic time slice shown in Figure 10.

PALEOGEOGRAPHIC ELEMENTS

Northern Platform Margin

Paleogeographic variations across the area had a large impact on sequence and reservoir development (Figure 4). Paleogeographic features are well imaged by the seismic data (see cross sectional view in Figure 8B and time slice view in Figure 10). An introduction to the paleogeographic elements and their seismic character is provided below.

Platform Interior The central part of the field area is dominated by stacked, shallow-water platform facies. Seismic horizons are concordant but highly discontinuous because of rapid lateral changes in facies and reservoir properties. The seismic discontinuity time slice shown in Figure 10 highlights a pond-channel complex that gives rise to the complex seismic character of the

The northern platform margin is a broad zone of relatively transparent seismic amplitude character. The margin records the aggradation of high-energy facies, with generally high reservoir quality. The discontinuity time slice shows that the margin is characterized by relatively high continuity, consistent with the high and continuous reservoir quality trends observed in log and core data. The width of the high-energy northern margin averages 8 km (5 mi). Maximum relief on the northern margin is estimated at 75 m (250 ft).

Clinoform Belt The northern part of the field area comprises clinoforms (sequences 4 and 5) that record more than 10 km (6.25 mi) of progradation. The clinoforms are characterized by alternations of low-angle (1–28), sigmoidal seismic reflections and higher angle (3–48), oblique

184 / Yose et al.

FIGURE 10. Seismic discontinuity time slice from flattened volume (flattened on base reservoir). Seismic surveys over Fields A1 and A2 are spliced together to create this volume. Seismic variations highlight the main paleogeographic elements across the field area, each of which has distinct reservoir characteristics. Platform margins are recognized north and south of the central buildup. Green line shows location of cross sections in Figure 8. Red dashes show location of calibration wells that were evaluated in the study.

seismic reflections (Figures 8, 9). Reservoir properties and stratal geometries are highly variable in the clinoforms, resulting in significant reservoir heterogeneity.

the south. Based on platform-margin geometries in seismic data, maximum depositional relief on the southern margin is estimated to be 30 m (100 ft).

Bab Basin

Intrashelf Embayment

The Bab basin is characterized by restricted, lowenergy facies, reflected in a relatively thin, condensed interval. This interval is separated from the overlying clinoform sequences by a seismic downlap surface (Figure 8). Basinal facies are characterized by highly continuous and concordant seismic reflections.

The southern platform margin passes rapidly into low-energy, restricted facies interpreted to have been deposited in a shallow intrashelf embayment that developed during the time of sequence 3. The inferred geometry of the embayment is shown in Figure 4, but is poorly constrained outside the 3-D seismic data area. At a regional scale, it is possible that the Shuaiba platform margin may have evolved into a series of low-relief, isolated banks.

Southern Platform Margin The 3-D seismic data reveal a previously unrecognized platform margin in the southern part of the study area. The southern margin differentiates later in time than the northern platform margin (Figure 8). The southern platform margin is aggradational, with very limited progradation. Clinoforms that developed seaward of the northern margin are absent to

SEQUENCE RECOGNITION Criteria for recognition of depositional sequences and surfaces in the Shuaiba vary depending on position in the Aptian sequence set and position along

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 185

FIGURE 11. Paired seismic section and line drawing illustrating the major phases of platform evolution during the deposition of the Aptian sequence set. The maximum flooding surface (MFS) shown occurs in sequence 2 and is the second-order MFS for the Aptian sequence set.

the depositional profile (platform, margin, slope, and basin). Sequence-stratigraphic correlations are guided by a combination of seismic, log, and core data. The seismic-stratigraphic architecture of the Aptian sequence set is summarized in Figure 11. Preliminary sequence interpretations are weighted largely on seismic geometries and termination patterns (e.g., downlap, toplap, and onlap). Because seismic surfaces correspond closely to time lines, seismic interpretations are used as a template to guide higher resolution correlations derived from core and log observations. In general, seismic surfaces are well expressed in the clinoforms of the northern field area where there are significant amplitude variations and seismic geometries (Figure 11). Seismic surfaces in platform interior and margin areas are difficult to constrain because of less geometric variation and minimal impedance contrast (Figures 8, 11). Further complexity is introduced by the high-amplitude bursts in the platform interior associated with the pond and tidal-channel complex (Figure 8B). Seismic interpretation in platform interior and margin areas is guided by well ties and the use of multiple seismic volumes (e.g., amplitude, band limited impedance, and discontinuity). Some of the sequence-stratigraphic surfaces are well expressed in core (e.g., exposure surfaces, hardgrounds, stylolitic boundaries, shale seams), as illus-

trated in Figure 12. Most of the surfaces are accompanied by rapid facies changes above and below and by changes in cycle stacking patterns. Transects of cored wells were selected to sample the full range of stratigraphic and seismic variability observed across the platform (Figures 8, 10). Surface correlations through the northern platform margin are difficult because of the thick stack of relatively continuous, high-porosity rock in sequences 2 and 3.

Sequence Boundaries Because of the prevailing Cretaceous greenhouse climate pattern during the Aptian, sequence boundaries are generally not pronounced. The best developed sequence boundaries are those developed at the base and top of the Aptian sequence set, especially the upper sequence set boundary (Figures 8, 9). Thirdand fourth-order sequence boundaries are best developed in the highstand and late highstand phases of the sequence set because of a decrease in long-term accommodation (Figures 8, 9). In updip areas of the Shuaiba (platform position), sequence boundaries are commonly expressed as thin, iron-oxide – encrusted surfaces, with pyrite mineralization (Figure 12A, B). Evidence for dissolution is limited to within a few feet or inches of the exposure surfaces. Coarse, blocky calcite cements are developed

186 / Yose et al.

FIGURE 12. Core expression of sequence-stratigraphic surfaces. SB = sequence boundary; MFS = maximum flooding surface; HFS = high-frequency (fourth-order) surface; HST = highstand systems tract. Surfaces in (A, B) are long-term exposure surfaces, forming the sequence boundary of the Lower Cretaceous supersequence. Surfaces shown in (D, E) are candidates for composite exposure and flooding surfaces, as discussed in Immenhauser et al. (2000). Characteristics of the surfaces are discussed in the text. locally below sequence boundaries and can penetrate as deep as 3 m (10 ft) below the exposure surface. Sequence boundaries are also marked by rapid facies changes above and below and are sometimes overlain by coarse skeletal and/or lithoclastic lag deposits. In downdip areas (slope and basin positions), sequence boundaries become conformable and lack evidence for exposure. However, some sequence boundaries record major facies shifts. For example, in the clinoform area, sequence boundaries correspond to a shift from grainy peloidal-skeletal facies (porous zones) in the HST to burrowed, miliolid-rich wackestone (dense zones) in the LST to TST above (Figure 9). In some cases, the downdip sequence boundaries are manifested by clay-rich seams that occur in the same position as the facies shifts (Figure 12C) and generate a clear gamma-ray log response to the surface.

Maximum Flooding Surfaces The maximum flooding surface of the Aptian sequence set is interpreted at the base of the last platformmargin backstep in the platform area, coincident with the maximum flooding surface of sequence 2 (Figures 8, 11). The equivalent surface in the slope and basin environment is interpreted within the condensed interval that separates shallow-water platform facies below from deeper water clinoform facies above (coincident with the seismic downlap surface; Figures 8, 11). Flooding surfaces in the platform area generally occur at or near a transition in facies and stacking patterns that record a change from deepening-upward to shoaling-upward characteristics. In some cases, it is difficult to pick an exact surface and is more appropriate to identify a flooding interval. Flooding surfaces (or intervals) are commonly well expressed in slope

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 187

FIGURE 13. Facies explanation and key to core log symbols used in this paper. Facies are subdivided based on the position in the Aptian sequence set and the position along the platform to basin profile. Core photos and reservoir-quality data for reservoir facies are illustrated in Figures 14 – 16. Core-to-log facies calibrations are shown in Figures 17 – 20. MLP = mud-lean packstone; skel = skeletal; pel = peloidal; bndst. = boundstone; fltst. = floatstone; HST = highstand systems tract; TST = transgressive systems tract.

and proximal basin environments and are generally expressed as condensed intervals that include concentrations of clay or muddy carbonate sediments, burrowed hardgrounds, and/or stylolites (Figure 12D–F).

SEQUENCE DESCRIPTION A general description of depositional sequences in relationship to evolution of the Shuaiba platform is provided below. The discussion is keyed to the four main phases of the Aptian accommodation cycle introduced above. A key to the facies nomenclature and symbols used in this report is provided in Figure 13. Core photos of representative facies types for each major phase of the Aptian sequence set are displayed in Figures 14 – 16 (transgressive, early highstand, and late highstand phases, respectively). Type logs for each of the main areas of the buildup are provided in Figures 17 – 20 (platform interior, platform margin, proximal clinoform sequence, and distal clinoform sequence, respectively).

Transgressive Phase The transgressive phase records the deposition during the rising limb of the long-term sea level cycle and includes sequences 1A–1B and 2 (Figure 9). During this period, the rate of relative sea level rise was greater than the carbonate accumulation rate. As a consequence, depositional relief was established over time, marking the initiation of the Bab basin in the area. The carbonate buildup differentiated as a low-relief ramp that attained as much as 45 m (150 ft) of depositional relief during the transgressive phase. Ramp interior facies are dominated by an algalstromotoporoid assemblage, with some associated corals and rudists (Figure 14). The faunal assemblage is interpreted to indicate relatively restricted circulation during the initial development of the Bab intrashelf basin. Ramp interior and margin facies of the transgressive phase are characterized by high porosity (10–25%) and low permeability (0.1–5 md) because of mud-dominated textures and development of microporosity.

188 / Yose et al.

FIGURE 14. Examples of depositional facies and reservoir properties in the composite transgressive interval (sequences 1 and 2). Colors in the upper right of photos are keyed to Figure 13. Scale bar is in 1-cm (0.4-in.) increments. IP = intraparticle; BP = between particle; md = millidarcys; Perm = permeability; Por = porosity; HST = highstand systems tract; TST = transgressive systems tract. The transgressive phase comprises as many as three depositional sequences (sequences 1A, 1B, and 2) as summarized below.

Sequence 1A Sequence 1A includes the Hawar shale (Figure 9) and was not extensively evaluated in this study. The Hawar comprises argillaceous, Orbitolina-rich packstone and wackestone and contains cyclic alternations of more and less argillaceous deposits. As discussed further in Strohmenger et al. (2006), there are alternative sequence-stratigraphic interpretations of the Hawar shale. Strohmenger et al. (2006) interpreted that the Hawar was deposited in relatively shallow, restricted environments with episodic exposure. Strohmenger et al. (2006) also interpreted that the Hawar is bounded above and below by sequence boundaries (interpretation shown in Figure 3). In

contrast, van Buchem et al. (2002) interpreted the Hawar to record a regional transgression associated with the initial formation of the Bab basin. In this interpretation, the Hawar may be the TST of sequence 1, as opposed to a separate depositional sequence. More regional study is required to test these alternative interpretations.

Reservoir Quality The Hawar is of very low porosity and permeability and forms an effective barrier (dense) between the Kharaib reservoir below and Shuaiba reservoir above.

Sequence 1B Sequence 1B comprises the basal part of the Shuaiba Formation and records a major change in deposition relative to sequence 1A. The argillaceous content in sequence 1B drops off significantly, and

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 189

FIGURE 15. Examples of depositional facies and reservoir properties in the composite highstand interval (sequence 3). Colors in the upper right of photos are keyed to Figure 13. Scale bar is in 1-cm (0.4-in.) increments. IP = intraparticle; BP = between particle; md = millidarcys; Perm = permeability; Por = porosity; TST = transgressive systems tract. there is a shift from Orbitolina-dominated facies to algal-dominated facies. Similar to sequence 1A, sequence 1B records shallow-water deposition across the study area, with no indication of depositional relief. The sequence is divided into three shoaling-upward cycles that can be correlated across the study area based on core and log character (Figures 17, 18). Facies stacking patterns in the high-frequency cycles are variable, but a general progression, from base to top, would include (1) skeletal-peloidal wackestone, (2) algal boundstone-floatstone, and (3) chalky peloidalskeletal packstone (with microporosity).

Reservoir Quality The algal boundstone and floatstone represent a range of facies that have variable reservoir quality (Figures 17, 18). Porosity ranges from 10 to 20% and

permeability from 0.1 to 5 md. The primary controls on reservoir quality variability include (1) sparse versus dense algal boundstone fabrics (intraparticle porosity in algal boundstone is better connected if the algal components are touching [dense]) and (2) grainy versus muddy matrix in the algal floatstone (grainier fabrics have higher porosity and permeability). The chalky, fine-grained facies (skeletal and peloidal wackestones and packstones) are characterized by the development of microporosity with relatively high porosity, but low permeability.

Sequence 2 Sequence 2 records the initial differentiation of the platform margin and inception of the Bab basin in the study area (Figure 9). The sequence is differentiated into distinct TST and HST, each with differing reservoir architecture and quality (Figure 9).

190 / Yose et al.

FIGURE 16. Examples of depositional facies and reservoir properties in the composite late highstand interval (sequences 4 and 5). Colors in the upper right of photos are keyed to Figure 13. Scale bar is in 1-cm (0.4-in.) increments. IP = intraparticle; BP = between particle; md = millidarcys; Perm = permeability; Por = porosity; HST = highstand systems tract; TST = transgressive systems tract. The TST is characterized by a backstepping geometry that records the development of the northern ramp margin. The initial margin develops via two small-scale backsteps that, together, created on the order of 20 m (65 ft) of relief (Figure 9). Algalstromotoporoid boundstone is the dominant facies in the TST (Figure 14A, B). Cycle stacking patterns change from exposure-capped shoaling-upward cycles below (sequence 1A), to amalgamated algalstromotoporoid facies above (sequence 2, TST). The basin is characterized by deposition of organic-rich wackestone and packstone (Figure 14F). These basinal facies immediately overlie shallow-water carbonate facies of sequence 1B and record the initial platform drowning (Figure 12F). The basinal deposits are interpreted as the condensed interval equivalent to the maximum flooding interval of the Aptian sequence set. Nannofossil age dates indicate that the con-

densed interval corresponds with early Aptian global oceanic anoxic event (OAE1a; Figure 3) (Bralower et al., 1994; Erba et al., 1999; Premoli Silva et al., 1999). The HST records aggradation of the northern ramp margin and the inception of the southern ramp margin. A diverse suite of facies is represented including shallow-water, algal-prone facies in the ramp interior and fine-grained mud-lean packstone on the ramp margin (Figure 14B). Faunal diversity increases slightly in the HST, with an increased abundance of corals and rudists (Figure 14C). The increase in faunal diversity may be related to more open-circulation patterns in the Bab intrashelf basin in response to the marine transgression. Evidence for decreasing accommodation during the HST phase includes a shift from retrogradational to aggradational geometry along the northern ramp margin and increased evidence for exposure. An exposure surface is present at the base

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 191

of a prominent biostromal unit in the HST interval (Figure 8). The biostromal unit corresponds to the mid-C coral zone of Russell et al. (2002), ranges between 3 and 8 m (10 and 25 ft) thick, and comprises a complex assemblage of corals, stromotoporoid, and algae, most of which are in growth position (Figure 14C). The biostromal unit is interpreted to have nucleated on the underlying exposure surface. Algal boundstone and floatstone with increasing amounts of rudist debris overlie the biostrome interval and continue up to the sequence boundary at the top of sequence 2. The upper sequence boundary is not a prominent exposure surface, but records a major shift from algaldominated facies below to rudist-dominated facies above. The faunal shift is rapid, commonly occurring in less than 1 ft (0.3 m) of vertical section. The surface recording the faunal shift is of regional extent (Hughes, 2000; van Buchem et al., 2002; Droste, 2004) and appears to indicate a major change in circulation patterns across the Bab basin.

Reservoir Quality Vertical variations in facies and reservoir quality in sequence 2 are shown in Figures 17 and 18. Facies textures in the HST are generally grainier that those in the underlying TST. The HST is dominated by algalboundstone and floatstone with a grain-prone matrix, whereas the underling TST is dominated by algalboundstone and floatstone with a mud-prone matrix. As a result, reservoir quality increases upward in sequence 2 (Figures 17, 18). In the TST, porosity ranges between 15 and 25%, and permeability ranges from 1 to 20 md. In the HST, porosity ranges between 20 and 30%, and permeability ranges from 10 to 100 md. The condensed interval deposited within the basinal area is dense and forms a barrier between the shallowwater platform facies of sequence 1B below and overlying slope facies of sequences 4 and 5 above.

Early Highstand Phase During the early highstand phase, the rate of longterm relative sea level rise was in balance with the carbonate sedimentation, resulting in aggradation of

FIGURE 17. Type log for the platform interior (well A in Figure 8). Vertical scale is in 10-ft (3-m) depth increments. Porosity values below 5% are shaded in gray. The 1-md permeability line is highlighted in red. See Figure 13 for key to facies codes and symbols. HST = highstand systems tract; TST = transgressive systems tract; SB = sequence boundary; MFS = maximum flooding surface.

192 / Yose et al.

the carbonate platform. Maximum depositional relief was attained during the aggradational phase, with more than 75 m (250 ft) of relief established between the platform and basin (Figure 9). Facies and faunal diversity reached a maximum during the early highstand phase, with the proliferation of rudists across the bank top (Figure 15A, B, D) and the development of margins to the north and south of the platform. The development of an open, high-energy platform may have been triggered by changes in circulation patterns in the Bab basin, in combination with relative sea level rise. Platform interior and margin facies deposited during the early highstand phase are the highest reservoir quality facies in the Shuaiba. Sequence 3 comprises the early highstand phase of deposition and is subdivided into distinct TST and HST, as discussed below.

Sequence 3 (TST) The TST of sequence 3 records significant changes in platform and margin deposition. The southern platform margin attained maximum development at this time, with as much as 30 m (100 ft) of relief between the platform and shallow embayment to the south. Depositional patterns in the platform interior changed significantly in response to changes in platform paleogeography. The platform interior differentiated into a complex mosaic of tidal channels, rudist shoals, and intershoal depressions or ponds, marking the establishment of significant cross-bank currents and open-marine conditions (Figures 9, 10). Platform-margin facies differ significantly between the northern and southern margins. The northern platform margin is characterized by a broad (5–7-km; 3.1 –4.4-mi) grainstone belt that passes northward into deeper water slope facies. A narrow band (1 – 2 km; 0.6 – 1.2 mi) of in-situ rudist and stromotoporoid boundstone and bafflestone is located platformward of the grainstone belt and is interpreted to form a discontinuous band of stromotoporoid and rudist patch reefs along the northern platform. The northern margin sand belt is characterized by skeletal-topeloidal packstone and grainstone that are interpreted to have been deposited in a high- to moderate

FIGURE 18. Type log for the northern platform margin (well B in Figures 8, 11). Vertical scale is in 10-ft (3-m) depth increments. Porosity values below 5% are shaded in gray. The 1-md permeability line is highlighted in red. See Figure 13 for key to facies codes and symbols. HST = highstand systems tract; TST = transgressive systems tract; SB = sequence boundary; MFS = maximum flooding surface.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 193

energy carbonate sand flat. Skeletal components are diverse and include local concentrations of Orbitolinarich and miliolid-rich grains. Textures are relatively fine grained and become coarser grained and cleaner (less muddy) upward within the section. Textures become finer and muddier toward the basin (downslope). In contrast, the southern platform margin is characterized by high-energy, rudist-dominated facies that cover a broad region of the outer platform. Depositional facies along the southern margin include a stack of rudist-dominated grainstone and boundstone (Figure 15A). Little vertical variation in this facies exists, and no internal surfaces or bedding features were identified. High-energy facies along the southern margin pass gradationally into the mosaic of rudist shoals, intershoal depressions, and channels characteristic of the platform interior to the north and grade rapidly into Orbitolina-rich packstone and grainstone of the upper- to middle-slope environment to the south.

Sequence 3 (HST)

FIGURE 19. Type log for the clinoform area (well C in Figures 8, 11). Well is in updip position of sequence 4, highlighting the development of the dense TST. Vertical scale is in 10-ft (3-m) depth increments. Porosity values below 5% are shaded in gray. The 1-md permeability line is highlighted in red. See Figure 13 for key to facies codes and symbols. HST = highstand systems tract; TST = transgressive systems tract; SB = sequence boundary; MFS = maximum flooding surface.

The HST records a significant change in platform and margin deposition relative to the underlying TST (Figure 9). Depositional energy levels in the platform interior become much lower. The complex association of channels, ponds, and shoals developed in the TST are absent in the HST. In contrast, the HST records a major decrease in depositional energy across the bank top and an attendant decrease in reservoir quality. Carbonate facies are generally mud dominated, with a notable increase in argillaceous content. In addition, in contrast to the amalgamated rudist shoal facies characteristic of the TST, the HST is characterized by the development of shoaling-upward parasequences with well-developed surfaces. Platform-margin facies and geometries also change significantly. The width of the southern margin narrows to 2 km (1.25 mi) and is characterized by much lower energy facies than in the underlying TST (Figure 9). The northern margin likewise narrows relative to the underlying TST and evolves from a fine-grained sand-flat environment into coarser rudist rudstone and floatstone deposited as shoals. Facies relationships and seismic geometries indicate that the northern margin prograded approximately 3 km (1.9 mi) during the HST as the rate of sea level rise began to decrease (Figure 8). The controlling factor on restriction of the platform interior during the HST may be related to geometry changes at the platform margins. In particular, the northern margin appears to become a more pronounced topographic feature, possibly in response

194 / Yose et al.

to shoaling and progradation of the margin. Progradation and increased topographic expression of the northern platform margin may have restricted circulation in the platform interior and in the intrashelf embayment to the south.

Reservoir Quality Reservoir quality varies both vertically and laterally within sequence 3 (Figures 17–20). Rudist shoal facies deposited in the platform interior of the TST contain the highest reservoir quality of any interval in the Shuaiba. Porosity ranges from 20 to 30%, and permeability ranges from 10 to 500 md. The TST also marks the period of maximum facies diversity caused by the development of the pond-channel network in the platform interior. Ponds and channels are filled with mudprone facies that have variable, but generally low, reservoir quality. Many of these features are nonreservoir. Reservoir quality varies significantly between the northern and southern margins (Figure 15). Rudistdominated facies of the southern margin have porosity of 20 –30% and permeability in the 100–1000-md range. In contrast, fine-grained grainstone and mudlean packstone deposited along the northern margin has similar porosity to the southern margin, but permeability is much lower, in the 10– 100-md range. The HST of sequence 3 is of generally low reservoir quality, with the best reservoir facies corresponding to rudist shoals developed along the northern margin. These facies range from 20 to 30% porosity, and permeability ranges from 10 to 100 md.

Late Highstand Phase

FIGURE 20. Type log for the clinoform area (well D in Figures 8, 11). Well is in downdip position of sequence 4, highlighting the development of the porous HST. Vertical scale is in 10-ft (3-m) depth increments. Porosity values below 5% are shaded in gray. The 1-md permeability line is highlighted in red. See Figure 13 for key to facies codes and symbols. HST = highstand systems tract; TST = transgressive systems tract; SB = sequence boundary; MFS = maximum flooding surface.

The late highstand phase marks a fundamental turning point in the history of the Aptian platform. Relative sea level rise slowed to a near standstill and then started to gradually fall. The carbonate platform had limited space for vertical accumulation and began to move laterally, and downslope, into the Bab basin. In response, the northern platform margin prograded more than 10 km (6.2 mi) into the Bab basin. Seismic geometries in the clinoform belt define two major phases of progradation that correspond to two depositional sequences (sequences 4 and 5; Figure 11). Because of a gradual, long-term fall in sea level, each slope sequence steps lower into the basin. Sequences 4 and 5 are thus detached from the main platform and are slope restricted. The older Aptian platform was exposed during most of the late hightstand phase, and a long-term subaerial exposure surface was developed. The exposure surface becomes progressively younger toward the basin in response to the downstepping geometry (Figure 9). Sequences 4 and 5 have very similar architecture, facies, and reservoir quality. Each sequence includes a muddy, lowstand-to-transgressive phase (low reservoir quality) with low-angle clinoforms, followed by a grainy, highstand phase (moderate to high reservoir

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 195

quality) with higher angle clinoforms (Figures 10, 11). Facies relationships in sequences 4 and 5 are summarized in Figure 16.

Sequences 4 and 5 (TST) The TST is dominated by muddy, nonporous facies that create extensive barriers to fluid flow. The facies model for the TST includes a downdip transition from (1) caprotinid floatstone, bafflestone, and rudstone at the margin (Figure 16D) to (2) mixed-peloidal and skeletal packstone and grainstone in the uppermost slope (Figure 16E) to (3) burrowed wackestone (denses) in the upper- and lower-slope environments (Figure 16F). The TST is dominated by thick, mudprone facies deposited in the upper- and middleslope environments. As will be discussed below, these muds are interpreted to have been generated in situ as mud mounds.

Sequence 4 and 5 (HST) The HST is dominated by grainy, porous clinoforms alternating with higher frequency (fourthorder) dense intervals. Clinoforms of the HST exhibit higher slope angles (2 – 38) than the muddy TST clinoforms (1 – 28). A typical facies progression in the HST includes rudist rudstone at the shelf margin (Figure 16A), skeletal grainstone in the upper slope, peloidal packstone and grainstone in the middle slope (Figure 16B), and foraminifera-packstone in the lower-slope environment (Figure 16C).

Reservoir Quality Reservoir quality varies dramatically between the TST and HST components of the slope sequences and along the depositional profile (Figures 16, 19, 20). The mud-dominated TST of each sequence is dense and forms an effective and laterally continuous barrier. Grainier facies in the updipmost part of the TST exhibit porosity that ranges from 10 to 20% and permeability ranging from 10 to 50 md, thus providing for possible updip connectivity between the slope sequences. The grain-dominated HST of each sequence has much higher reservoir quality, although the internal architecture of the HST clinoforms is complex, resulting in a high degree of heterogeneity. The fourthorder cyclicity generates a high-frequency alternation of flow units and flow barriers. Reservoir quality varies systematically down the profile of each clinoform in response to pore-type changes. Porosity values are fairly consistent (15–25%) down the profile, but permeability varies significantly, ranging from the tens to hundreds of millidarcys in grain-prone updip

facies, to 0.1–1 md in the downdip mud-prone facies of the lower slope. The dominant pore type in lowerslope facies is microporosity. Microporosity development may have been enhanced through meteoric diagenesis, also in association with subaerial exposure.

Lowstand Phase The maximum lowstand of sea level is marked by a significant influx of fine-grained siliciclastics and the demise of the Aptian carbonate platform. Lowstand facies form a basin-restricted wedge (sequence 6) that onlaps the Aptian platform margin (Figures 10, 11) and forms a lateral seal to the reservoir. The lowstand wedge comprises a shoaling- and cleaning-upward sequence that includes an upward succession from finegrained siliciclastics and carbonate mud to skeletal and oolitic carbonate grainstone. Sequence 6 is bounded at the top by an exposure surface that precedes deposition of the overlying Nahr Umr Shale. In the study, only the updip part of sequence 6 is present. Regional relationships indicate that sequence 6, corresponding in large part to the Bab Member of the Shuaiba, continues to prograde into the Bab basin. The Nahr Umr Shale records a rise in sea level that floods back across the previously exposed Aptian platform. However, in contrast to carbonate platform facies below, the marine environment was dominated by fine-grained siliciclastics, with only thin intervals of argillaceous carbonates. The Nahr Umr forms a top seal for the Shuaiba reservoir.

VOLUME-BASED RESERVOIR EVALUATION The sequence-stratigraphic framework and 3-D seismic data are used to guide an integrated evaluation of the Aptian reservoir. Volume interpretation and visualization workflows are applied to further evaluate and quantify geologic variations and their impact on reservoir performance. A corendered image of seismic discontinuity and seismic porosity is shown in Figure 21 and provides a seismic-scale perspective of reservoir architecture and properties. Field A1 is extremely heterogeneous and can be subdivided into three main areas, each with diagnostic geologic, seismic, and production expressions. These areas include the platform interior to the south, the northern platform margin in the central field area, and the clinoform belt to the north. Geologic controls that drive these variations are directly related to the sequence-stratigraphic framework as discussed above and summarized in Figure 8. A peripheral waterflood was implemented during earlier stages of field development (prior to acquisition

196 / Yose et al.

FIGURE 21. Volume-based reservoir evaluation. The volume shown is total porosity corendered with seismic discontinuity. The volume highlights the reservoir heterogeneity resulting from sequence-stratigraphic variations and the impact on field development. Time slice is from a flattened volume datumed on the top of sequence 1 (base of reservoir) and represents a slice through the TST of sequence 3 in the platform interior and margin, cutting through sequences 4, 5, and 6 in the clinoform belt and the Bab basin. The total porosity volume includes the seismic-derived porosity (bandlimited porosity) plus the low-frequency porosity trend. High-frequency porosity data from core and well data are added during the 3-D geologic modeling stage.

of 3-D seismic data) and does not adequately accommodate the reservoir heterogeneity that is now apparent. Several field management challenges have developed, including differential sweep in the platform interior, early water encroachment along the platformmargin grainstone belt, and pressure support in the clinoforms (Figure 21). The ADCO has implemented several field management strategies to further optimize field development. As demonstrated below, the 3-D seismic data will greatly assist in guiding future well placement and field optimization strategy.

This interval contains between 4 and 5 billion bbl of in-place oil reserves. Thus, a small improvement in recovery will have a large economic impact. Understanding the origin and distribution of reservoir quality anomalies in the platform area has presented a long-standing field development challenge, both in terms of avoiding these anomalies with producer and injector wells and the impact of these features on reservoir sweep and flood-front advance (Figure 21).

Seismic Characterization of the Platform Interior

The 3-D seismic data provide a blueprint of the channel and pond network in the platform interior (Figures 6, 21) and a new tool for reservoir evaluation. In our initial reconnaissance of the seismic data, seismic features were identified within the platform interior that were interpreted either as karst features

The platform interior is characterized by a complex spatial distribution of porosity that is related to the channel, pond, and rudist shoal complex developed during the early highstand phase (Figure 21).

Calibration with Well and Core Data

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 197

FIGURE 22. Integrated characterization of the platform interior region. (A) Reservoir isochron corendered with seismic discontinuity (red = isochron thick; blue = isochron thin). Blue well sticks are water injector wells, and red well sticks are selected producers. Ponds and channels correspond to isochron thins, and porous rudist shoals correspond to isochron thicks. Prominent isochron thick to the north is the northern platform margin. Peripheral water injection is diverted by ponds and channels resulting in differential sweep. (B) Example of high-porosity, high-permeability rudist shoal facies developed among the ponds and channels. (C) Example of nonporous submarine pond fill (open-marine skeletal wackestones and packstones with rudist debris). Por = porosity; Perm = permeability; md = milidarcy.

developed in association with the top Shuaiba supersequence boundary or as syndepositional features, such as tidal channels and submarine ponds. As summarized in Figure 22, calibration of the seismic data with well data supports a syndepositional origin for the seismic features. Volume corendering of the seismic discontinuity volume with the reservoir isochron (time thickness) enhanced the detail of stratigraphic features in the platform interior (Figures 6, 22) and guided the selection of calibration wells. Calibration of the seismic data with log data shows that isochron

thins correspond to dense, nonporous carbonate with fast seismic velocities, whereas isochron thicks correspond to porous carbonates with slower seismic velocities. These observations are consistent with the seismic porosity predictions shown in Figure 21, confirming that the pond and channel complex is generally of lower reservoir quality than the surrounding, more continuous seismic facies. Cores from within the pond features comprise largely carbonate mudstone and wackestone that was deposited in open- to restricted-marine settings

198 / Yose et al.

(Figure 22). A range of carbonate facies was observed within the ponds, including open-marine carbonate mudstone; wackestone; peloidal-skeletal packstone; and restricted-marine, organic-rich mudstone and algal laminites. In contrast, cores from wells located outside the ponds comprise rudist-dominated grainstone, floatstone, and rudstone that were deposited in high-energy carbonate shoals and patch reefs. The open-marine facies in the pond fill contain rudists and other sediments that are interpreted to have been transported into the ponds from the adjacent rudist shoals. No evidence was found to indicate that karsting or dissolution was involved in the formation of the pond and channel features. Collapse breccias, fractures, vadose silts, and other diagenetic features typical of karst environments are absent. The ponds and channels appear to be time equivalent to the surrounding rudist shoals and patch reefs. The initial distribution of the rudist shoals and ponds may have been controlled by antecedent relief on the underlying sequence boundary (top of sequence 2), with shoals nucleating on subtle highs. The initial relief was then enhanced through aggradation of the shoals in response to sea level rise during the transgressive phase of sequence 3. This increase in accommodation resulted in the differentiation of the southern platform margin, thus opening the platform top up to strong cross-bank currents and development of submarine channels. As the rate of accommodation began to decrease, the rudist shoals coalesced across the bank top, thus infilling most of the pond and channel topography (Figure 8). The HST of sequence 3 extends across the tidal channels, ponds, and shoals of the underlying TST, with little change in facies (Figures 8, 9). This relationship indicates that the channels and ponds were developed and infilled during the TST of sequence 3 and are not related to the long-term exposure event on the top of the Shuaiba (Lower Cretaceous supersequence boundary).

Reservoir Implications Seismic porosity distribution in the channel-pond network shows that, although porosity is generally lower than the surrounding rudist shoal facies, there is considerable variability, and not all of the volume is nonreservoir (Figure 21). Our calibrations with core and well data indicate that porosity varies with respect to several factors, including the size of the ponds and position of wells in the ponds. Larger ponds are more restricted from marine currents than smaller ponds and are filled with muddier, tighter rock.

In addition, reservoir quality tends to increase progressively from the pond interior to the pond margins. Less data are available to assess facies and reservoir quality variations in the large tidal channels, but seismic porosity indicates that, in general, the channels are filled with higher porosity facies than the ponds. Visualization sessions provide a powerful mechanism for evaluating reservoir sweep and conformance and for identification of bypassed reserves in the platform interior. Visualization of seismic volumes such as those shown in Figures 21 and 22 enabled reservoir engineers to rapidly resolve long-standing production anomalies. For example, many producers with abnormally high or low water cuts can be explained by their position relative to injector wells and the pond-channel complex (Figure 22). Producers that are sheltered from injectors by ponds and channels show less water breakthrough (higher oil cuts) than producers that are not as sheltered. The 3-D seismic data provide new, detailed information on the distribution of reservoir features that is being used to guide reservoir development and placement of future producer and injector wells.

Seismic Characterization of the Platform Margin The northern platform-margin area is characterized by the highest and most continuous porosity in the field area (Figures 18, 21). Production data indicate that premature water breakthrough is occurring along the platform margin in response to peripheral water injection. The seismic data reveal possible stratigraphic and structural controls on fluid flow in the platform-margin area.

Calibration with Well and Core Data Seismic and sequence-stratigraphic relationships of the platform-margin area are summarized in Figure 23. The seismic character of the margin area is generally transparent because of the stack of porous and relatively continuous reservoir facies. Discontinuous basinward-dipping seismic reflectors (clinoforms) are present in the HST of sequence 3. Calibration with core data show that platform-margin facies is comprised of mainly skeletal and nonskeletal grainstone that is remarkably fine grained and well sorted (Figures 15E, F; 16B, C; 23C). The carbonate sand shoals aggraded vertically and then, as the rate of sea level rise began to decrease, prograded basinward. As a result, facies transition upward from a grainy platform margin into rudist-dominated outer platform facies (Figure 23C).

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 199

FIGURE 23. Seismic and depositional relationships in northern platform margin area of the reservoir. Depositional sequence shown in (C) is sequence 3 in Figure 8. The vertical stacking of facies in the platform margin records progradation, with open-platform interior facies prograding over platform-margin facies. Reservoir quality in the margin area is generally high and relatively continuous. (A) Seismic transect through northern platform-margin area (amplitude data). (B) Seismic transect through northern platform-margin area (impedance data; orange = lower porosity, gray = higher porosity). (C) Facies and reservoir-quality relationships in the platform-interior, platform-margin to upper-slope environments. BP = between-particle porosity; IP = intraparticle porosity; micro = microporosity; Por = porosity; Perm = permeability; HST = highstand systems tract; TST = transgressive systems tract; SB = sequence boundary. See Figure 13 for explanation of facies codes.

200 / Yose et al.

FIGURE 24. Three-dimensional perspective on clinoform architecture, showing a time slice of seismic discontinuity with 2-D seismic amplitude cross sections. Systematic variations in clinoform architecture are clearly imaged and are related to the third-order sequence framework as shown. Colored well sticks show the location of wells that were evaluated as part of the current study. Intersections of third-order sequence boundaries with the seismic time slice are denoted by the dotted red lines.

Reservoir Implications The data indicate that the northern platform margin is a broad zone of stacked, moderate- to highreservoir-quality facies. Seismic data show that this trend is continuous across the field area and well connected to the peripheral water injectors (Figure 21), consistent with the observed water breakthrough patterns. However, as previously discussed, faults seen in the seismic data trend parallel to the platform margin and may influence fluid flow and water breakthrough patterns (Figure 7). Sensitivity testing can be conducted via flow simulation to determine if matrix properties can match the performance data or whether additional permeability associated with faults and fractures is required. The 3-D fault framework derived from seismic data (Figure 7) can be leveraged in the sensitivity testing. The seismic and sequence-stratigraphic frameworks also provide an improved understanding of lateral facies relationships and reservoir connectivity in a dip direction across the platform margin (Figure 23).

High-porosity grainstone in the platform-margin transitions into the more variable reservoir quality facies of the platform interior. The platform margin and interior areas are thus in pressure communication, and injected water moving along the strike of the platform margin can also interfinger laterally into higher permeability facies in the platform interior. Water breakthrough has been observed in the proximal platform interior area and may increase over time.

Seismic Characterization of the Clinoforms Slope clinoforms are characterized by complex stratal geometries and significant variations in reservoir properties (sequences 4 and 5; Figure 8). As a consequence, peripheral water injection has not provided adequate pressure support or sweep to this part of the reservoir (Figure 21). In response, ADCO has implemented a pattern gas-injection flood to add pressure support and optimize recovery. As illustrated in Figure 24, seismic data provide new 3-D information

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 201

FIGURE 25. Seismic and depositional anatomy of clinoform sequences. (A) Seismic amplitude section across the entire clinoform belt, highlighting seismic expression of sequences 4 and 5 (TST = transgressive systems tract; HST = highstand systems tract). (B) Close-up of sequence 4 with seismic interpretation added. Well sticks in black are featured in (C). (SB = third-order sequence boundaries; MFS = third-order maximum flooding surface; fourth-order surfaces shown with blue double arrows). (C) Facies schematic for sequence 4 showing impact of clinoforms on flow layering and well pair communication.

on stratal architecture and reservoir quality variations that can be used to assess reservoir connectivity and optimize gas flood management.

by finer scale alternations of porous and dense facies that result from higher order depositional cyclicity (Figure 25).

Calibration with Well and Core Data

Reservoir Implications

The detailed anatomy of a slope-restricted depositional sequence is provided in Figure 25. The seismic character of each sequence includes a low-angle (1 – 28), high-amplitude reflector, followed by a series of shorter, steeper (3– 48) reflectors. Cores were evaluated across several slope sequence transects to calibrate the seismic response (see Figure 24 for location of calibration wells). The low-angle, high-amplitude reflector at the base of the cycle is generated by the impedance contrast between more porous rock below and extremely dense rock above. In the sequence model, this interface is interpreted as a sequence boundary separating more porous highstand deposits below, from muddier, transgressive deposits above. Dense, muddy, lowstand-to-transgressive deposits overlie the sequence boundary and form the base of each depositional sequence. Higher angle reflectors in the highstand part of each depositional sequence are generated

Reservoir connectivity and quality variations in the clinoforms are summarized in Figure 25. Reservoir implications are discussed below in the context of the sequence hierarchy. At the sequence scale, transgressive dense intervals at the base of each depositional sequence form effective barriers to fluid flow and partition the clinoform area into separate pressure compartments (Figure 8). These dense intervals are well imaged by the seismic data, and 3-D geometries of the reservoir barriers can be mapped across the northern field area (Figures 25, 26). Highstand parts of sequences comprise larger scale flow compartments and contain the bulk of the pore volume in the clinoform region. The highstand flow compartments are, in turn, subdivided into a series of alternating flow units and discontinuous flow barriers, corresponding to the development of fourth-order cyclicity. As illustrated

202 / Yose et al.

FIGURE 26. Three-dimensional visualization of clinoforms highlighting reservoir connectivity and continuity. Clinoform geometries have a large impact on connectivity between injectors (yellow) and producers (red). Volume shown is instantaneous phase with opacity filter applied to enhance visualization of seismic-scale flow units and flow barriers. Basal structural surface is draped with seismic porosity and discontinuity from within the reservoir. in Figures 25 and 26, seismic reflection data are able to resolve many of these fourth-order clinoform packages and provide a template to guide interwell correlations. Without seismic data, one may be tempted to correlate on the basis of facies and log similarities, resulting in horizontal, instead of inclined, correlation surfaces (i.e., connecting the colors in Figure 25C). Such correlations crosscut seismic reflectors and geologic time lines and result in incorrect flow layering and rock property distributions in the reservoir model. Integration of the seismic with core and well data provides information on subseismic variations in reservoir properties. Depositional facies change systematically down the profile of each clinoform because of changes in water depth and wave energy. Grain-rich, porous facies deposited in updip positions pass gradually into finer grained, less-porous facies deposited in downdip positions. These facies changes, coupled with diagenetic effects, result in systematic pore-type changes that control porosity and permeability variations in individual flow units (Figure 25C). The geometry and continuity of the flow units, barriers, and baffles will have a pronounced influence on injector-producer well connectivity in the pattern gas flood area. As illustrated in Figure 26, 3-D seismic visualization is used to evaluate the distribution of seismic-scale flow units and flow barriers in

the clinoforms in relation to existing producer and injector wells. Such tools are used to assess 3-D reservoir connectivity, to evaluate producer-injector conformance, to resolve specific well-pair performance issues, and to guide future well placement.

IMPLICATIONS FOR RESERVOIR MODELING Three-dimensional geologic and flow-simulation models provide the ultimate reservoir management tools. The inherent complexity of carbonate reservoirs presents significant challenges to distributing 3-D rock properties in reservoir models. The new volume-based reservoir framework provides a template for distributing flow units and flow barriers in the reservoir and for distributing facies and rock properties in the 3-D framework. Likewise, the 3-D seismic data constrain the structural mapping of the reservoir and have provided a 3-D fault framework that can be incorporated into the reservoir model framework. An integrated static and dynamic reservoir modeling effort that incorporates results from the present study is underway, but was not concluded at the time of this publication. Some examples are provided below to show how a volume-based workflow can lead to more accurate reservoir models and can also reduce the reservoir-modeling cycle time.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 203

FIGURE 27. Seed-detected clinoforms extracted from 3-D seismic volume. Colors correspond to connected clinoform segments. Volume shown is the cosine of instantaneous phase, which enhances continuity of the seismic reflections.

Rapid Interpretation Techniques

Volume-based Data Integration

Development of the reservoir framework is one of the most time-consuming steps in the construction of a 3-D geologic model. The volume-interpretation techniques applied in the present study reduced the time required to construct the structural and stratigraphic frameworks. For example, seismic mapping of individual clinoform surfaces in the northern field area is critical for the reservoir description, but is time consuming because of the stratigraphic complexity. Quantitative seed detection techniques were used to expedite the correlation process. The ExxonMobil proprietary seed detection approach applied in this study uses a combination of amplitude and traceshape attributes to autopick the clinoform surfaces. Results are illustrated in Figure 27. The colored surfaces are autopicked clinoform surfaces, and each color corresponds to a connected clinoform surface. This technique is rapid (interpretations were generated in a few hours) and allows for quantitative assessment of clinoform connectivity. The fast-track interpretations must be further evaluated and optimized by the seismic interpreter. Ultimately, the seismic clinoform surfaces are depth converted and used to guide the correlation of higher order surfaces identified from logs and core data.

Most conventional 3-D modeling workflows focus on the integration of one-dimensional and twodimensional (2-D) data sets, and a full 3-D representation of the reservoir is not developed until the later stages of the workflow, following the construction of the 3-D geologic model. A volume-based workflow provides the basis for 3-D reservoir characterization throughout the workflow. Figure 28 illustrates how the volume-based environment is used to visualize information on structure, stratal architecture, reservoir properties, and static and dynamic well data in a common 3-D environment. Such techniques are used to evaluate relationships between seismic and other geologic and production data and to identify key reservoir issues early in the workflow. Volume-based data integration leads to more accurate reservoir models and promotes a 3-D reservoir perspective throughout the workflow.

Extraction of Reservoir Features from Seismic Data As illustrated in Figure 29, seismic features in the platform interior can be extracted from the seismic volume and incorporated directly into the 3-D geologic model. In this workflow, geologic and seismic attributes are used to guide the extraction process.

204 / Yose et al.

FIGURE 28. Seismic image highlighting variations in reservoir architecture and quality at the full-field scale. Amplitude data are shown in 2-D vertical slice. Discontinuity draped on time structure and corendered with band-limited seismic porosity predictions shown in time slice view. Selected calibration wells are shown as yellow sticks. Major paleographic elements are annotated. The extraction was constrained vertically using seismic horizons tied to the sequence-stratigraphic framework. Integration of the seismic with well and core data indicate that the pond-channel network is confined to a specific sequence-stratigraphic interval (sequence 3, HST; Figure 9), and this knowledge is used to guide the extraction of these features and their placement into the 3-D geologic model. The lateral extent of individual ponds and channels is constrained using a combination of seismic attributes, including discontinuity, isochron, and porosity. Once constrained vertically and laterally, the features can be extracted from any seismic volume, along with the seismic properties. Figure 29B and C show 2-D and 3-D perspectives of the pond and channel features extracted from the seismic porosity volume. The extracted features, populated with seismic porosity estimates, were converted from time to depth and incorporated directly into the 3-D geologic model. Seismic porosity predictions are used in

combination with well log and core information to produce the final 3-D porosity distribution in the geologic model.

Rock Property Modeling Accurate representation of rock properties in reservoir models is challenging because of the different scales of information on rock properties (seismic, well tests, logs, and core) and the impact of differing pore types on the porosity-permeability functions. Rock property modeling requires a combined approach, using 3-D seismic data and the sequence-stratigraphic framework to guide large-scale trends between wells, and high-resolution information from log and core data to characterize pore-scale relationships.

Integration of Seismic Porosity Data The average well spacing in Field A1 is 1 km (0.6 mi), and seismic data provides the only 3-D information on rock properties between wells. Several seismic attributes

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 205

were calibrated to log-derived porosity data using a neural network approach (Hampson et al., 2001) and used to generate a full-field prediction. A more detailed explanation of the seismic porosity prediction workflow conducted in association with this study is provided by Al-Menhali et al. (2005) and Schuelke et al. (2005). Most of the seismic porosity information is contained in the impedance attribute, but adding additional attributes via the neural net approach improved the calibration. Seismic porosity predictions are validated through blind tests, where wells are held out of the calibration process and then compared against the predictions to characterize uncertainty. Visualization techniques are then used to evaluate the seismic porosity predictions relative to the sequence-stratigraphic framework and performance data. Statistical error in the seismic prediction is generally low but increases in areas where rock properties change rapidly over short vertical or lateral distances (e.g., clinoforms of sequences 4 and 5, and tidal channels and ponds in sequence 3). As illustrated in Figures 21 and 28, there is a strong tie between seismic-based porosity predictions and the geologic framework. The seismic porosity volume provides a valuable reservoir evaluation tool and will help to constrain porosity trends in the 3-D geologic model.

Porosity and Permeability Predictions Prediction of porosity-permeability relationships requires understanding pore-type distributions, which, in turn, define reservoir rock types. Reservoir rock types reflect a combination of diagenetic and depositional processes, and depending on the level of diagenesis, rock types may not equate to depositional facies. In such cases, the diagenetic component must be modeled separately. Whereas a considerable diagenetic overprint does exist in the Shuaiba reservoir, most of the diagenesis is facies selective and related to sequence-stratigraphic surfaces (e.g., exposure FIGURE 29. Seismic extraction of platform interior pond and channel features for input into the 3-D geologic model. (A) Boundaries of detected ponds and channels shown on discontinuity volume corendered with reservoir isochron (blue = thin). The combination of these two seismic attributes provides the best lateral constraint on these features. (B) Time slice view of pond and channel features extracted from the seismic porosity volume (blue = low porosity; red = high porosity). (C) Threedimensional perspective of pond and channel complex extracted from seismic porosity volume (blue = low porosity; red = high porosity).

206 / Yose et al.

FIGURE 30. Porosity-permeability crossplot for sequences 4 and 5. Depositional facies provide a close proxy for reservoir rock types in this system. Porosity-permeability relationships vary as a function of sequence, systems tract, and depositional environment and are thus predictable with the sequence framework. Dominant pore types are indicated: BP = betweenparticle porosity; IP = intraparticle porosity; MO = moldic porosity; VUG = vuggy porosity.

surfaces). Thus, depositional facies, in combination with the sequence-stratigraphic framework, provide a good understanding of reservoir rock-type distribution. Figure 30 illustrates the value of using reservoir rock types and the sequence-stratigraphic framework to guide porosity-permeability modeling in the clinoform area. Reservoir rock types and their properties vary predictably based on systems tract (TST and HST) and position along the depositional profile (updip and downdip) (Figure 30). Preliminary 3-D modeling results from the clinoform area are shown in Figure 31. The flow lines in the model correspond to sequence-stratigraphic surfaces and represent geologic time lines. Facies changes that occur along and across these time lines have a significant impact on rock property distribution. Note the strong conformance between the sequencestratigraphic framework of the clinoforms (Figure 8), the seismic porosity response (Figures 21, 28), and the 3-D geologic model (Figure 31). A key uncertainty

in the clinoform area is how connected the clinoforms are in the updip area, below the top Shuaiba supersequence boundary (Figures 8, 31). A clear compartmentalization exists between the TST and HST clinoform intervals downdip, but updip, relationships are not clear. Only a few wells penetrate the narrow updip extension of the TST intervals, so there are limited data. Reconciliation of this issue will require feedback between the geologic and flow-simulation models through integration of production and performance data.

LEARNINGS ON CARBONATE SYSTEMS The volume-based approach used to characterize the Shuaiba reservoir provides details on carbonate system development that are sparsely observed in outcrop or subsurface studies. Some highlights and key learnings are discussed below.

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 207

FIGURE 31. Paired porosity and permeability extracts from the 3-D geologic model in the clinoform area along the northern platform margin. HST = highstand systems tract; TST = transgressive systems tract.

Controls on Platform Architecture and Evolution Integrated characterization of the Shuaiba reservoir reveals a very ordered system, where changes in stratal geometries and facies distribution are tied closely to the long-term accommodation history (Figure 9). However, despite the order and predictability imparted by changes in relative sea level, there is variability in platform development that may be related to other competing factors, such as local tectonics, sediment supply, and cross-bank energy flux.

Platform Asymmetry The Shuaiba platform records a marked asymmetry in depositional facies and stratal geometries from north to south across the study area. Despite a similar accommodation history, the northern and southern margins are very different in terms of geometry and facies (Figure 9). The northern margin is characterized by the development of a broad (10-km; 6.25-mi), grain-dominated sand belt that persisted during most of the time of sequences 2 and 3, followed by strong progradation during deposition of sequences

4 and 5. The southern margin differentiated mainly during the time of sequence 3 and is characterized by high-energy rudist shoals that occupy a broad area of the margin and outer platform (7 –9 km; 4.4 – 5.6 mi) (Figure 9). The southern margin is mainly aggradational, and sequences 4 and 5 appear to be represented as restricted, low-energy mudstone deposited in the intrashelf embayment (Figure 9). Carbonate platform asymmetry is common in many ancient and modern carbonate systems and may develop in response to a variety of controls, including differential subsidence, sediment supply, physical energy flux (e.g., currents, waves), and interference patterns between isolated platforms (Yose and Collins, 2002). Two primary controls on the asymmetry observed within the Shuaiba buildup are interpreted. 1) Physical energy flux: At the time of maximum extent of the Shuaiba platform (sequence 3), we interpret the southern margin as the currentfacing margin and the northern margin as the leeward-facing margin. The current-facing margin was the optimum site for proliferation of the rudists. Facies comprising rudist rudstone are

208 / Yose et al.

common along the southern margin, indicating strong wave and current activity. Large tidal channels imaged by the seismic data all appear to enter the platform from the south and then meander their way into lower energy areas of the platform interior, connecting to many of the submarine pond features (Figures 6, 9). In contrast, the leeward margin is characterized by the deposition of fine-grained skeletal-peloidal grainstone and packstone that grade into mud-dominated slope facies (Figure 8). The sands and muds along the northern margin are interpreted to record offbank transport of fine-grained carbonate sediment along the leeward margin. The tidal-channel complex facilitated cross-bank transport of sediments toward the leeward margin. The interplay of prevailing winds, currents, and paleogeography in controlling the observed patterns in cross-bank energy flux is not clear at present. 2) Position of sea level: The asymmetry developed during deposition of sequences 4 and 5 results from strong progradation along the leeward margin and virtually no progradation to the south (Figures 9, 10). These relationships are consistent with the cross-bank energy flux described above. Another factor, however, is that the long-term fall in sea level may have significantly restricted circulation patterns in the intrashelf embayment to the south (Figure 4). Connections from the larger Bab basin into the embayment may have been restricted as sea level lowered. In contrast, the northern margin faced into the Bab basin, where open circulation was maintained, and rudist fringing shoals developed in downslope positions.

Variations in Clinoform Geometry Variations in the geometry and sedimentology of the slope clinoform sequences can be observed at a variety of scales and illustrate a range of possible controls. At the largest scale, variations in the lateral width of the clinoform belt show a close correspondence to the present-day structure. The clinoform belt is widest (more progradation) along the crest of the structure and narrows offstructure (east and west) (Figure 24). The offstructure thinning of the clinoform belt is accommodated by a combination of increased slope angles to the east and west and a pronounced narrowing of sequence 5 to the east, compensated primarily through thinning of the TST (Figures 9, 24). On the eastern side of the field, it becomes difficult to differentiate sequences 4 and 5 on

the basis of seismic time slices because the shingled HSTs of sequences 4 and 5 are merged (Figure 24). These relationships demonstrate the variability that can occur along the strike of carbonate platform margins. The changes in clinoform geometry could result from along-strike variations in sediment supply or subsidence patterns. The offstructure decrease in the width of the clinoform belt indicates the possibility of an early structural influence on sedimentation. The primary structural movement of the large north-northeast structures on the Arabian plate is interpreted to have been in the Late Cretaceous ( Johnson et al., 2005). However, even a slight increase in subsidence rates offstructure would generate increased accommodation space and could account for the increase in slope angles and the decrease in the distance of progradation observed in the offstructure positions. Conversely, the clinoform geometries could have been produced by variations in sediment supply, with areas of higher progradation corresponding to higher sediment supply. The apparent pinch and swell of individual clinoforms observed in seismic data are interpreted to record lobate geometries resulting from variations in erosion and sediment supply. These variations could also account for geometric variations at the larger scale.

Platform Response to Sea Level Fall The long-term fall in sea level during the Aptian provides an opportunity to evaluate the response of a carbonate system to sea level fall, including perspectives on stratal architecture, sedimentation patterns, and diagenesis.

Downslope Shifts of the Carbonate Factory The Aptian platform demonstrates the potential for carbonate systems to shift downslope in response to sea level fall. Sequences 4 and 5 are interpreted to have accumulated as slope-restricted wedges that are detached from the main platform (Figure 8). Seismic (time) and log (depth) relationships confirm that the clinoform sequences are downstepping into the Bab basin, tracking the second-order sea level fall. In the literature, these relationships are referred to as ‘‘forced regressions,’’ in that sedimentation is being forced basinward (downslope) in response to the sea level fall (Hunt and Tucker, 1992). Younger sequences onlap the margin of older sequences, and the composite unconformity becomes progressively younger toward the basin. Facies relationships from core indicate little to no erosion along the tops of the clinoforms; a normal facies progression is observed up to

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 209

the sequence boundary in each clinoform sequence. Thus, the basinward thinning of the clinoform package is not caused by erosion. The low-angle (2–38) slopes flanking the buildup provided a large area for in-situ carbonate production, and the relative decrease in accommodation during the long-term sea level fall allowed for extensive progradation of the carbonate system into the shallow basin. Forced regressions, as observed within the Aptian sequence set, raise questions on the placement of sequence boundaries and on systems tract nomenclature in the slope sequences. In the Shuaiba example, the Aptian sequence set boundary could be placed below sequences 4 and 5, marking the initiation of long-term exposure of the central Shuaiba platform. In this interpretation, sequences 4 and 5 would be considered as part of the composite lowstand interval. Our preferred interpretation is to place the sequence set boundary at the base of sequence 6. In this interpretation, sequences 4 and 5 become part of the late highstand interval or, alternatively, the falling-stage systems tract of Plint and Nummedal (2000). Sequence 6 is viewed as the lowstand phase of the next sequence set and records a major change in sedimentation patterns, including a large influx of siliciclastics. Issues around nomenclature have no impact on the fundamental sequence interpretations or the reservoir framework. However, for regional correlations, it is important to have a clear and consistent nomenclature.

Mechanisms of Progradation The Shuaiba platform provides new insights on mechanisms of carbonate progradation in response to sea level changes. Each clinoform slope sequence (sequences 4 and 5) includes a low-angle, muddominated base (TST), followed by deposition of higher angle grainy clinoforms (HST) (Figures 9, 10). Each sequence records on the order of 5 km (3.1 mi) of progradation, resulting in more than 10 km (6.2 mi) of total progradation (Figures 24, 25). In the central part of the clinoform belt, partitioning of progradation between the TST and HST intervals is roughly equal, but the mechanisms of progradation are different. A two-phase style of progradation is proposed. The TST phase includes aggradation of carbonate mud banks that nucleate downslope in response to sea level fall and then aggrade vertically during the subsequent transgression. Aggradation of the mud banks creates new depositional relief and effectively steps the margin 2 – 3 km (1.2 – 1.9 mi) outboard relative to the previous highstand. During the HST

phase, the rate of accommodation begins to slow, and higher energy shoals initiate along the newly developed platform margin and then prograde basinward (Figure 25A, B). The source of the mud in the transgressive interval is enigmatic because the adjacent bank top is interpreted to be exposed during the deposition of the slope sequences. The carbonate factory was thus restricted to the slope. The transgressive muds thicken upslope, forming a basin-thinning wedge. Based on evaluation of core, we interpret that the upper- to middle-slope environment was an area of in-situ carbonate sediment production. As the carbonate system was catching up with rising sea level after the sea level fall, the main type of sediment produced was carbonate mud. The source of the muds may have been sea grass and algal meadows developed in the upper- to middle-slope environments. The dramatic upslope thickening of denses supports the concept that the carbonate muds were produced locally (Figure 9). The presence and abundance of miliolid foraminifera, which lived attached to sea grasses, further supports the interpretation that sea grass may be an important factor in both the baffling and production of carbonate mud. Hydrodynamic ponding of muds along the upper-slope environment may also be a factor.

Impact of Long-term Subaerial Exposure The supersequence boundary that developed on the top of the Aptian sequence set resulted in longterm exposure of the carbonate platform, providing an opportunity to assess the impact of long-term exposure on the carbonate system. The exposure surface is time transgressive and becomes progressively younger toward the basin because of the progressive downslope shift of the carbonate system (Figure 9). The time span of the exposure surface is estimated to range from 3 – 4 m.y. in the platform interior to 1 – 2 m.y. in sequence 5. Based on observations from core and seismic data, there is no evidence for karsting of the platform in response to the long-term sea level fall. Exposure surfaces are marked by thin (2–6-cm; 0.8–2.4-in.) crusts of iron mineralization, with only minor evidence for dissolution (Figure 12A, B). Although evidence for karsting is absent, long-term exposure of the carbonate platform is interpreted to have had a significant impact on porosity development near sequence boundaries and in some sequences. Blocky calcite cements are common below surfaces and can extend down for several feet, creating tight intervals immediately below

210 / Yose et al.

sequence boundaries. Subaerial exposure of grainprone facies results in the development of abundant vuggy and moldic porosity through dissolution of skeletal and nonskeletal grains. Selective dissolution is observed to penetrate tens of feet below sequence boundaries into the underlying sequence. Dissolution enhances porosity in facies that have abundant primary interparticle porosity, leading to high porosity and permeability (Figure 16A). Abundant microporosity is observed in algal platform facies of sequence 1B (Figure 14A, B, E) and in the middle- to lower-slope facies of sequences 3–5 (Figures 15F, 16C). Development of microporosity may have been enhanced by meteoric fluids. Mircroporosity results in relatively high porosity but low permeability.

CONCLUSIONS High-quality 3-D seismic data acquired over an onshore field in Abu Dhabi demonstrate the value of seismic data for integrated reservoir characterization and field optimization. The seismic data quality, abundance of core, well and production data, and the reservoir heterogeneity all combine to produce an ideal data set to test the limits of high-end seismic technologies in carbonates and to demonstrate the impact on reservoir characterization. Seismic data, in combination with sequence-stratigraphic concepts, provide a valuable framework for reservoir evaluation and for incorporating flow layering and rock property variations into reservoir models. Key elements of and insights gained from the workflows applied in the present study are summarized below. 1) Optimizing the seismic data quality: The impact of high-end seismic technologies increases as the data quality improves. Efforts on seismic data optimization, including poststack filtering, should occur early in the seismic workflow. The present study demonstrates that poststack filtering can improve the signal-to-noise ratio and the interpretability of 3-D seismic surveys, even those with high-effort seismic data acquisition and prestack processing. 2) Calibration of seismic data to core, well, and production data: Seismic data must be calibrated with other subsurface data to groundtruth the seismic response to variations in geology and reservoir properties. A fundamental understanding of the underlying geology is critical for carbonate reservoir characterization. Volume-based visualiza-

3)

4)

5)

6)

tion is a useful tool to facilitate integration and evaluation of a wide range of subsurface data and their relationships. Use of multiple seismic attributes for structural and stratigraphic interpretations: A range of seismic attribute volumes should be evaluated to facilitate interpretation of the seismic data. Different attribute volumes provide different information and perspectives on the structural and stratigraphic frameworks and distribution of rock properties. Attribute volumes that were most useful in this study included amplitude, discontinuity, impedance, isochron, porosity, dip, instantaneous phase, and quadrature. Attribute volumes can also be corendered in different combinations to provide additional detail and information. Evaluation of seismic data within an integrated geologic framework: Application of high-end seismic technologies must be conducted within integrated structural and stratigraphic frameworks. This is a continuous process because seismic data is used to develop geologic frameworks, and the frameworks, in turn, are used to guide more detailed seismic attribute analyses, such as porosity prediction and fault and fracture analysis. Multidisciplinary approach: Accurate reservoir characterization and evaluation requires a multidisciplinary approach. Three-dimensional visualization of seismic and other subsurface data facilitates interactions among geoscientists and engineers, leading to a collective understanding of the reservoir and improved reservoir management. Volume-based reservoir optimization: As demonstrated in the present study, calibrated 3-D seismic data can provide a powerful foundation for reservoir evaluation and optimization. Visualization of seismic data, along with well and production data, can help to resolve longstanding reservoir performance issues, identify new opportunities, and guide future field development strategies.

ACKNOWLEDGMENTS The authors thank the Abu Dhabi National Oil Company, the Abu Dhabi Company for Onshore Oil Operations, the ExxonMobil Exploration Company, and the ExxonMobil Upstream Research Company for supporting this collaborative study and for permission to publish these results. We also thank our colleagues for their contributions and valuable

Three-dimensional Characterization of a Heterogeneous Carbonate Reservoir, Abu Dhabi / 211

discussions, including Jim Anderson, Jim Markello, Peter Holterhoff, Imelda Johnson, Po Tai, Nat Stevens, Steve Bachtel, Brian Coffey, and Linda Corwin. The manuscript was improved by reviews from Jim Weber, Scott Tinker, Tony Simo, Sherry Becker, and Mike Kozar.

REFERENCES CITED Al-Menhali, S. S., W. L. S. Abu, and J. S. Schuelke, 2005, Rock property prediction using multiple seismic and geologic attributes provides insight to field development for a large U.A.E. field: International Petroleum Technology Conference (Qatar) Paper 10595, CD-ROM. Alsharhan, A. S., 1985, Depositional environment, reservoir units evolution, and hydrocarbon habitat of Shuaiba Formation, Lower Cretaceous, Abu Dhabi, United Arab Emirates: AAPG Bulletin, v. 69, p. 899 – 912. Alsharhan, A. S., 1993, Bu Hasa field — United Arab Emirates: Rub’ al Khali basin, Abu Dhabi, in N. H. Foster and E. A. Beaumont, compilers, Structural traps VIII: AAPG Treatise of Petroleum Geology, Atlas of Oil and Gas Fields, v. A-26, p. 99 – 127. Alsharhan, A. S., and A. E. M., Nairn, 1993, Carbonate platform models of Arabian Cretaceous reservoirs, in J. A. Simo, R. W. Scott, and J. Masse, eds., Cretaceous carbonate platforms: AAPG Memoir 56, p. 173 – 184. Azer, S. R., and C. Toland, 1993, Sea level changes in the Aptian and Barremian (upper Thamama) of offshore Abu Dhabi, U.A.E.: Society of Petroleum Engineers Middle East Oil Technical Conference and Exhibition, Bahrain, SPE Paper 25610, p. 141 – 154. Bralower, T. J., M. A. Arthur, R. M. Leckie, W. V. Sliter, D. J. Allard, and S. O. Schlanger, 1994, Timing and paleoceanography of oceanic dysoxia/anoxia in the late Barremian to early Aptian: Palaios, v. 9, p. 335 – 369. Droste, H. J., 2004, Regional controls on reservoir properties in the Shuaiba Formation of North Oman (abs.): 6th Middle East Geoscience Conference and Exhibition, Bahrain, CD-ROM. Eberli, G. P., G. T. Baechle, F. S. Anselmetti, M. L. Incze, 2003, Factors controlling elastic properties in carbonate sediments and rocks: The Leading Edge, v. 22, no. 7, p. 654 – 660. Erba, E., J. E. T. Channell, M. Claps, C. Jones, R. Larson, B. Opdyke, I. Premoli-Silva, A. Riva, G. Salvini and S. Torricelli, 1999, Integrated stratigraphy of the Cismon APTICORE (southern Alps, Italy): A ‘‘reference section’’ for the Barremian – Aptian interval at low latitudes: Journal of Foraminiferal Research, v. 29, p. 371 – 391. Greselle, B., and B. Pittet, 2005, Fringing carbonate platforms at the Arabian plate margin in northern Oman during the late Aptian – middle Albian: Evidence for high-amplitude sea-level changes: Sedimentary Geology, v. 175, p. 367 – 390.

Hampson, D., J. S. Schuelke, and J. Quirein, 2001, Use of multi-attribute transforms to predict log properties from seismic data: Geophysics, v. 66, no. 1, p. 220 – 236. Hardenbol, J., J. Thierry, M. B. Farley, T. Jacquin, P. De Graciansky, and P. R. Vail, 1998, Mesozoic and Cenozoic sequence chronostratigraphic framework of European basins, in P. De Graciansky, J. J. Hardenbol, T. Jacquin, and P. R. Vail, eds., Mesozoic and Cenozoic sequence stratigraphy of European basins: SEPM Special Publication 60, p. 3 – 14. Hughes, G. W., 2000, Bioecostratigraphy of the Shuaiba Formation, Shaybah field, Saudi Arabia: GeoArabia, v. 5, p. 545 – 578. Hunt, D., and M. E. Tucker, 1992, Stranded parasequences and the forced regressive wedge systems tract: Deposition during base-level fall: Sedimentary Geology, v. 81, p. 1 – 9. Immenhauser, A., A. Cresusen, M. Esteban, and H. B. Vonhof, 2000, Recognition and interpretation of polygenic discontinuity surfaces in the middle Cretaceous Shuaiba, Nahr Umr, and Natih formations of northern Oman: GeoArabia, v. 5, no. 2, p. 299 – 322. Johnson, C. A., T. Hauge, S. Al-Mehhali, S. B Sumaidaa, B. Sabin, and B. West, 2005, Structural styles and tectonic evolution of onshore and offshore Abu Dhabi, U.A.E.: International Petroleum Technology Conference (Qatar) Paper 10646, CD-ROM. Marzouk, I. M., and M. A. Sattar, 1993, Implications of wrench tectonics on hydrocarbon reservoirs, Abu Dhabi, U.A.E.: Proceedings of 8th Middle East Oil Show and Conference, Bahrain, Society of Petroleum Engineers Paper 25608, p. 119 – 130. Masafarro, J. L., R. Bourne, and J. C. Jauffred, 2004, Threedimensional seismic volume visualization of carbonate reservoirs and structures, in G. P. Eberli, J. L. Masaferro, and J. F. Sarg, eds., Seismic imaging of carbonate reservoirs and systems: AAPG Memoir 81, p. 11 – 41. Plint, A. G., and D. Nummedal, 2000, The falling stage systems tract: Recognition and importance in sequence stratigraphic analysis, in D. Hunt and R. L. Gawthorpe, eds., Sedimentary responses to forced regressions: Geological Society (London) Special Publication 172, p. 1 – 18. Premoli Silva, I., E. Erba, G. Salvini, C. Locatelli, and D. Verga, 1999, Biotic changes in Cretaceous oceanic anoxic events of the Tethys: Journal of Foraminiferal Research, v. 29, p. 352– 370. Russell, S. D., M. Akbar, B. Vissapragada, and G. M. Walkden, 2002, Rock types and permeability prediction from dipmeter and image logs: Shuaiba reservoir (Aptian), Abu Dhabi: AAPG Bulletin, v. 86, p. 1709 – 1732. Sarg, J. F., and J. S. Schuelke, 2003, Integrated seismic analysis of carbonate reservoirs: From the framework to the volume attributes: The Leading Edge, v. 22, no. 7, p. 640 – 645. Schuelke, J. S., L. A. Yose, S. Al-Menhali, and W. Soroka,

212 / Yose et al. 2005, Seismic rock property predictions provide insight to field development (abs.): AAPG Annual Meeting Program, v. 14, p. A125. Sharland, P. R., R. Archer, D. M. Casey, R. B. Davies, S. H. Hall, A. P. Heward, A. D Horbury, and M. D. Simmons, 2001, Arabian plate sequence stratigraphy: GeoArabia Special Publication 2, 371 p. Sharland, P. R., D. M. Casey, R. G. Davies, M. D. Simmons, and O. E. Sutcliffe, 2004, Arabian plate sequence stratigraphy — Revisions to SP2: GeoArabia, v. 9, no. 1, p. 199 – 214. Strohmenger, C. J., L. J. Weber, A. Ghani, K. Al-Mehsin, O. Al-Jeelani, A. Al-Mansoori, T. Al-Dayyani, L. Vaughan, S. A. Khan, and J. C. Mitchell, 2006, High-resolution sequence stratigraphy and reservoir characterization of upper Thamama (Lower Cretaceous) reservoirs of a giant Abu Dhabi oil field, United Arab Emirates, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reser-

voirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 139– 171. van Buchem, F. S. P., B. Pittet, H. Hillgartner, J. Grotsch, A. I. Al Mansouri, I. M. Billig, H. H. J. Droste, W. H. Oterdoom, and M. van Steenwinkel, 2002, Highresolution sequence stratigraphic architecture of Barremian/Aptian carbonate systems in northern Oman and the United Arab Emirates (Kharaib and Shuaiba formations): GeoArabia, v. 7, p. 461 – 500. Yose, L. A., and J. F. Collins, 2002, Windward-leeward models for carbonate platforms revisited (abs.); AAPG Annual Meeting Program, v. 11, p. A196. Yose, L. A., et al., 2004, New frontiers in 3-D seismic characterization of carbonate reservoirs: Example from a supergiant field in Abu Dhabi: 11th Abu Dhabi International Petroleum Exhibition and Conference, Society of Petroleum Engineers, SPE Paper 88689, 16 p.

Strohmenger, C. J., P. E. Patterson, G. Al-Sahlan, J. C. Mitchell, H. R. Feldman, T. M. Demko, R. W. Wellner, P. J. Lehmann, G. G. McCrimmon, R. W. Broomhall, and N. Al-Ajmi, 2006, Sequence stratigraphy and reservoir architecture of the Burgan and Mauddud formations (Lower Cretaceous), Kuwait, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 213 – 245.

6

Sequence Stratigraphy and Reservoir Architecture of the Burgan and Mauddud Formations (Lower Cretaceous), Kuwait Christian J. Strohmenger,1 John C. Mitchell, Howard R. Feldman, Patrick J. Lehmann, and Robert W. Broomhall ExxonMobil Exploration Company, Houston, Texas, U.S.A.

Timothy M. Demko University of Minnesota Duluth, Duluth, Minnesota, U.S.A.

Robert W. Wellner ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

Penny E. Patterson

G. Glen McCrimmon

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

Hibernia Management and Development Company, St. John’s, Newfoundland and Labrador, Canada

Ghaida Al-Sahlan Kuwait Oil Company, Ahmadi, Kuwait

Neama Al-Ajmi Kuwait Oil Company, Ahmadi, Kuwait

ABSTRACT

A

new sequence-stratigraphic framework is proposed for the Burgan and Mauddud formations (Albian) of Kuwait. This framework is based on the integration of core, well-log, and biostratigraphic data, as well as seismic interpretation from giant oil fields of Kuwait. The Lower Cretaceous Burgan and Mauddud formations form two thirdorder composite sequences, the older of which constitutes the lowstand, transgressive, and highstand sequence sets of the Burgan Formation. This composite sequence is subdivided into 14 high-frequency, depositional sequences that are characterized by tidal-influenced, marginal-marine deposits in northeast Kuwait that grade into fluvial-dominated, continental deposits to the southwest.

1

Present address: Abu Dhabi Company for Onshore Oil Operations, Abu Dhabi, United Arab Emirates.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215878M883271

213

214 / Strohmenger et al.

The younger composite sequence consists of the lowstand sequence set of the uppermost Burgan Formation and transgressive and highstand sequence sets of the overlying Mauddud Formation. This composite sequence is sand prone and mud prone in southern and southwestern Kuwait and is carbonate prone in northern and northeastern Kuwait. The lowstand sequence set deposits of the Burgan Formation are subdivided into five high-frequency depositional sequences, which are composed of tidal-influenced, marginal-marine deposits in northeastern Kuwait that change facies to fluvial-dominated deposits in southwestern Kuwait. The transgressive and highstand sequence sets of the Mauddud Formation are subdivided into eight high-frequency, depositional sequences. The Mauddud transgressive sequence set displays a lateral change in lithology from limestone in northern Kuwait to siliciclastic deposits in southern and southwestern Kuwait. The traditional lithostratigraphic Burgan –Mauddud contact is time transgressive. The Mauddud highstand sequence set is carbonate prone and thins south- and southwestward because of depositional thinning. Significant postdepositional erosion occurs at the contact with the overlying Cenomanian Wara Shale. The proposed sequence-stratigraphic framework and the incorporation of a depositional facies scheme tied to the sequence-stratigraphic architecture allow for an improved prediction of reservoir and seal distribution, as well as reservoir quality away from well control.

INTRODUCTION A regional sequence-stratigraphic analysis of the Lower Cretaceous Burgan and Mauddud formations was undertaken through a joint study conducted by ExxonMobil Exploration Company and Kuwait Oil Company. The study focused on the stratigraphic architecture of selected major oil fields throughout Kuwait, including the supergiant Greater Burgan field in southeastern Kuwait, and Raudhatain and Sabiriyah fields in northern Kuwait (Figure 1). Three culminations constitute the supergiant Greater Burgan field: Burgan, Magwa, and Ahmadi. These three culminations are located near the crest of the north–south-trending Kuwait arch (Fox, 1961; Adasani, 1965; Brennan, 1990b; Carman, 1996) (Figure 1). The first well on these structures was drilled in 1938, followed by wells in 1951 and 1952. Production of 28 – 368 API oil comes from the Burgan (the major oil-producing reservoir), the Mauddud (a minor oilproducing reservoir), and the Wara formations (Kaufman et al., 1997). The recoverable oil reserves are estimated to be in the tens of billions of barrels (Christian, 1997). Raudhatain field was discovered in 1955. It is a faulted anticlinal dome with production of 28 – 408 API oil from the Ratawi, Zubair, Burgan, and Mauddud formations (Milton and Davies, 1965; Adasani, 1967; Al-Rawi, 1981; Brennan, 1990a; Carman, 1996;

Al-Eidan et al., 2001) (Figures 1, 2). The recoverable oil reserves are estimated to be in the billions of barrels (Christian, 1997). Sabiriyah field was discovered in 1956. It is an elongated faulted anticline with production of 28 – 328 API oil from the Burgan and Mauddud formations (Milton and Davies, 1965; Adasani, 1967; AlRawi, 1981; Brennan, 1990a; Carman, 1996; Kaufman et al., 1997; Al-Eidan et al., 2001) (Figures 1, 2). The recoverable oil reserves are estimated to be in the billions of barrels (Christian, 1997). The Burgan and Mauddud formations are part of the Wasia Group that overlies the Lower Cretaceous Thamama Group of the Arabian plate (Alsharhan and Nairn, 1993, 1997). The lower to middle Albian Burgan Formation is the major oil-bearing sandstone reservoir throughout the Greater Burgan field, as well as at Raudhatain and Sabiriyah fields in northern Kuwait. The thickness ranges from approximately 1250 ft (380 m) at the Greater Burgan field area to approximately 900 ft (275 m) at Raudhatain and Sabriya fields area (Bou-Rabee, 1996). The overlying upper Albian Mauddud Formation is a major oil-bearing carbonate reservoir in northern Kuwait (Al-Anzi, 1995). Thickness of the Mauddud Formation ranges from only a few feet at the Greater Burgan field and Minagish field areas (south and southwest Kuwait) to approximately 450 ft (140 m) at the Abdali, Raudhatain, and Sabiriyah fields in northern Kuwait (Bou-Rabee, 1996).

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 215

FIGURE 1. Location map showing the major oil fields of Kuwait (green) and the wells (black dots) studied.

Methodology The emphasis of this study involved the interpretation of the sequence-stratigraphic architecture based on correlation of regional stratal surfaces (sequence boundaries [SB], transgressive surfaces [TS], maximum flooding surfaces [MFS], and flooding surfaces [FS]) throughout Kuwait, as well as on the sequence-stratigraphykeyed facies analyses of the Burgan and Mauddud formations. Approximately 9600 ft (2930 m) of conventional core from 30 wells penetrating the Burgan Formation and approximately 5000 ft (1520 m) of conventional core from 35 wells penetrating the Mauddud Formation were described sedimentologically. The depositional environments interpreted from core were correlated to well-log signatures and used to develop the regional sequence-keyed sequence-stratigraphic framework. More than 100 wells were correlated within this sequence-stratigraphic context. In addition, the results from seismic-stratigraphic and biostrati-

graphic interpretations were integrated into this study. The identified chronostratigraphic surfaces (SB, TS, MFS, and FS) were assigned to high-frequency sequences (HFS) and numbered sequentially from the top down; the underlying sequence boundary giving name to the overlying high-frequency sequence (Figure 3). The HFSs were subsequently grouped into the sequence sets of two third-order composite sequences (Mitchum, 1977; Mitchum et al., 1977; Vail et al., 1977, 1991; Haq et al., 1987, 1988; Vail, 1987; Van Wagoner et al., 1987, 1988; Sarg, 1988; Haq, 1991; Mitchum and Van Wagoner, 1991; Sarg et al., 1999) (Figure 3).

High-frequency Sequences High-frequency sequences in the siliciclastic Burgan Formation (Figure 3) are defined by the stacking patterns and facies distributions. These sequences may contain lowstand, transgressive, and highstand

216 / Strohmenger et al.

FIGURE 2. Seismic cross section oriented northwest – southeast showing Raudhatain and Sabiriyah structures and the interpreted main stratigraphic horizons. Seismic line runs through the center of Raudhatain and Sabiriyah fields shown in Figure 1.

systems tracts (LST, TST, and HST). Each systems tract exhibits distinct facies trends, thickness distributions, and reservoir quality (Strohmenger et al., 2002; Demko et al., 2003). The LSTs consist of incised-valley fills (IVF). These valleys become thinner, as well as more laterally discontinuous and tidal influenced downdip to the northeast. Transgressive systems tracts correspond to retrogradational successions of coarsening-upward, marginal-marine mudstones and sandstones that grade into marine carbonates downdip and mudstone-prone coastal-alluvial plain deposits updip. Highstand systems tracts correspond to progadational successions of coarsening-upward, marginal-marine mudstones and sandstones that grade updip into mudstoneprone coastal-alluvial plain deposits (Strohmenger et al., 2002; Demko et al., 2003). High-frequency sequences in the carbonatedominated Mauddud Formation (Figure 3) are defined by parasequence (PS) stacking patterns, facies distributions, and microkarst or exposure surfaces. High-frequency sequences of the Mauddud Formation may contain TST and HST (Strohmenger et al., 2002; Demko et al., 2003). Transgressive systems tracts are generally more mud dominated with intercalated sandstones and glauco-

nitic sandstones in the northern Kuwait Raudhatain and Sabiriyah fields area and grade into siliciclastics toward the south at Greater Burgan field and toward the southwest at Minagish field areas. Highstand systems tracts typically show an upward increase in grain richness (graining upward) as well as porosity (Strohmenger et al., 2002; Demko et al., 2003).

Composite Sequences High-frequency sequences are grouped into sequence sets of two third-order composite sequences, based on the stacking patterns and facies distributions (Figure 3). Each composite sequence consists of a lowstand sequence set (LSS), transgressive sequence set (TSS), and a highstand sequence set (HSS) (Strohmenger et al., 2002; Demko et al., 2003). In the siliciclastic Burgan Formation of Kuwait, LSSs are characterized by an aggradational stacking of HFSs dominated by braided fluvial deposits. The TSS exhibits an overall retrogradational stacking pattern, dominated by nonmarine facies at the base, whereas the uppermost HFSs contain increasing marginal-marine components. The HSS forms an overall progradational succession dominated by marginalmarine facies, especially in northern Kuwait (Strohmenger et al., 2002; Demko et al., 2003).

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 217

FIGURE 3. Mauddud – Burgan sequence-stratigraphic framework showing the lower and upper third-order composite sequences, as well as the interpreted high-frequency depositional sequences (Raudhatain field type well RA-G). GR = gamma-ray log; MD = measured depth (feet); RES = resistivity log; NEU = neutron porosity log; DENS = density log. The Mauddud TSS shows a lateral change in lithology from limestone in northern Kuwait to siliciclastics in southern and southwestern Kuwait. An overall shoaling-upward or progradational signature characterizes the HSS. Most of the upper HSS is removed by erosion throughout much of southern and southwestern Kuwait (Strohmenger et al., 2002; Demko et al., 2003).

BURGAN FORMATION A regional sequence-stratigraphic analysis of the Burgan and Mauddud formations reveals that the traditional lithostratigraphic Burgan-Mauddud contact is a time-transgressive facies boundary (Stroh-

menger et al., 2002; Demko et al., 2003). To define coeval facies successions in both the Burgan and Mauddud formations, a chronostratigraphically significant regional flooding surface (B100_TS) was defined as the Burgan-Mauddud contact (Figure 3). Within this chronostratigaphic framework, the uppermost Burgan and overlying Mauddud formations form a composite sequence (Figure 3) that becomes more siliciclastic prone to the southwest and carbonate prone to the northeast. A second composite sequence encompasses the rest of the underlying Burgan Formation (Figure 3). This composite sequence is dominated by marginal-marine deposits to the northeast and nonmarine deposits to the southwest. The Burgan Formation, as defined in this study, comprises 19 HFSs (Figure 3). Each of these HFSs

218 / Strohmenger et al.

FIGURE 4. Sequence-keyed depositional facies models for the Burgan Formation. In these models, the lowstand systems tract consists of incised-valley deposits, whereas the trangressive and highstand systems tracts are composed of wavedominated shoreface depositional systems. GR = gamma-ray log.

contains LSTs, TSTs, and HSTs, each with distinct facies trends, thickness distributions, and reservoir quality. The LSTs of these sequences consist of IVF. These valleys become thinner, as well as more laterally discontinuous and tidal influenced downdip to the northeast. The TSTs of the Burgan sequences display a systematic downdip to updip change from marine carbonates to marginal-marine mudstones and sandstones to mudstone-prone alluvial- and coastal-plain deposits. The HSTs are dominated by marginal-marine sandstones and mudstones downdip and alluvialand coastal-plain mudstones and sandstones updip. The shorelines in these highstands trend northwest – southeast and are best developed in the northern Raudhatain and Sabiriyah fields area.

The primary reservoirs in the Burgan Formation are fluvial and tidal deposits that formed within the incised valleys (Figure 4). Shoreline sandstones in the TSTs and HSTs are also potential reservoirs (Figure 4). However, these marginal-marine sandstones have lower porosity and permeabilities because of their finer grained and increased clay matrix due to bioturbation. In summary, the sequence-stratigraphic analysis of the Burgan Formation provides an improved understanding of the spatial and temporal distribution of reservoirs that can be used to address explorationscale to production-scale issues. In general, the Burgan Formation is a classic regressive-transgressive-regressive package. It is dominated by sandstone-prone fluvial deposits at its base

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 219

and top and marginal-marine mudstones, sandstones, and limestones in the middle parts to the northeast. The Burgan Formation is bounded at its top by a regional flooding surface, referred to as the Burgan transgressive surface B100_TS (Figure 3). In general, this boundary is marked by a change from blocky, high net/gross, fluvial-dominated sandstones of the Burgan Formation below to low net/gross marine mudstones, sandstones, and limestones of the Mauddud Formation above. Within this context, fine-grained siliciclastics, traditionally included at the top of the Burgan Formation to the south, are assigned to the clastic member of the Mauddud Formation. This Mauddud clastic member occurs above the Burgan transgressive surface B100_TS, a regional flooding surface, defined as the top of the Burgan Formation and beneath the carbonate strata traditionally included within the Mauddud Formation (Mauddud carbonate member).

Lithofacies and Depositional Environments Within the Burgan Formation, 22 lithofacies were identified and define 6 distinct facies. The physical criteria used to delineate the individual lithofacies in this study include grain size, composition, sorting, grading, physical and biogenic sedimentary structures, stratal boundaries, presence or absence of clay drapes, organic-rich drapes, organic debris, and diagenetic features. For ease of description, each lithofacies was classified into six categories based on mud content and apparent reservoir quality. The six categories range from well-sorted, very clean sand (lithofacies 1: excellent reservoir quality) to mudstone (lithofacies 5: very poor reservoir quality) and, ultimately, to coal (lithofacies 6). In addition, lithofacies were assigned an alphabetic qualifier to account for subtle differences in grain size, mud content, dominant sedimentary structures, and type and extent of bioturbation. Trace fossil assemblages were used to infer environments of deposition, which are based on models proposed by Pemberton et al. (1992a, b). In addition, our regional database includes two lithofacies that were not observed in the Burgan strata. They are lithofacies 3A, coarsening-upward, bioturbated mudstone to sandstone; and lithofacies 4C, bioturbated to planar laminated mudstone to sandstone.

Lithofacies 1A: Trough Cross-bedded Sandstone Lithofacies 1A (Figure 5A) consists of poorly to moderately sorted, fine-grained to coarse-grained sandstone. The dominant sedimentary structures are trough

cross-bedding and current-ripple cross-lamination. Large plant fossils, including compressed branches and leaves, are common. Mudstone drapes and bioturbation are uncommon. Grain size trends are typically fining upward. Small-scale fining-upward beds are 0.2–2 ft (0.06–0.6 m) thick and typically associated with individual trough cross-bed sets. Largerscale, fining-upward trends range from 20 to 50 ft (6 to 15 m) of thickness. The cross-bedding and grain-size trends all indicate deposition in a low-sinuosity fluvial setting. Some fining-upward trends suggest sedimentation in sandy point bars. This lithofacies occurs primarily in incised valleys and in updip braid-plain deposits.

Lithofacies 1B: Trough Cross-bedded Sandstone with Minor Clay Drapes Lithofacies 1B (Figure 5B) consists of poorly to moderately sorted, fine-grained to coarse-grained sandstones that possess sparse to common clay drapes and thick-thin couplets. Sedimentary structures are dominated by trough- and current-ripple cross-bedding. Small mudstone and siderite clasts may be locally abundant. Small-scale, grain-size trends are typically fining upward. Large-scale, grain-size trends are either coarsening upward or fining upward. Large plant fossils, such as compressed sticks and leaves, may be common. Bioturbation is sparse to moderate. Common burrow types are horizontal sand-filled tubes with a circular cross section, approximately 1 cm (0.4 in.) in diameter (Planolites). The clay drapes and thick-thin couplets indicate that some tidal influence occurred during deposition. This lithofacies occurs at the fluvial to tidal transition in IVF and at the top of coarsening-upward tidal bars.

Lithofacies 1C: Current-rippled, Cross-laminated Sandstone This lithofacies (Figure 5C) consists of well-sorted, very fine-grained to fine-grained sandstone. Sedimentary structures are characterized by current-ripple cross-lamination with minor small-scale trough crossbedding. Clay drapes are sparse to absent. This facies occurs in beds typically a few feet thick with no grainsize trends. Bioturbation is uncommon. Small plant fossils, such as small leaves and fine plant fragments, are common. Lithofacies 1C occurs in a range of environments. Where it is associated with coastal-plain deposits, it

220 / Strohmenger et al.

FIGURE 5. Siliciclastic lithofacies 1, slabbed core photographs. (A) Lithofacies 1A: trough cross-bedded sandstone. (B) Lithofacies 1B: trough cross-bedded sandstone with minor clay drapes. (C) Lithofacies 1C: current-rippled, crosslaminated sandstone, which is interbedded with small-scale, trough cross-beds.

represents splays and channel levees. Within distal incised valleys, this lithofacies represents sedimentation in tidal channels.

Lithofacies 2A: Clay-draped, Current-rippled to Laminated Sandstone Lithofacies 2A (Figure 6A) consists of well-sorted, very fine- to fine-grained sandstone. Sedimentary structures include current-ripple cross-laminations and minor horizontal laminations, with abundant thin (<1-mm; <0.04-in.) clay and organic drapes, commonly on ripple foresets. Some laminations exhibit centimeter-scale cyclicity in mud-sand couplets. Siderite cement is common. Bedsets typically fine upward. Bioturbation is uncommon to moderate and consists

of a low diversity of burrows, including Planolites, Skolithos, and Arenicolites. Fine plant debris and amber flakes are common. This lithofacies represents deposition in tidal creeks or tidal flats.

Lithofacies 2B: Clay-draped, Current-rippled to Trough Cross-bedded Sandstone Lithofacies 2B (Figure 6B) consists of very fineto fine-grained sandstone. The dominant sedimentary structures are current-ripple cross-laminations to small-scale, trough– cross-bedding with abundant thin (1-mm) clay and organic drapes, commonly on ripple and trough foresets. Bedsets may distinctly coarsen upward. Bioturbation is sparse to moderate

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 221

FIGURE 6. Siliciclastic lithofacies 2, slabbed core photographs. (A) Lithofacies 2A: clay-draped, current-rippled to laminated sandstone. (B) Lithofacies 2B: clay-draped, current-rippled to trough cross-bedded sandstone. (C) Lithofacies 2C: clay-draped, horizontally laminated sandstone. (D) Lithofacies 2D: glauconitic sandstone that has been extensively bioturbated by Teichichnus (Te) and Planolites (Pl) burrows. and is characterized by the low diversity of burrows (mostly Planolites). Fine plant debris and amber flakes are locally abundant. This lithofacies is inferred to have been deposited as subtidal bars in estuaries.

Lithofacies 2C: Clay-draped, Horizontally Laminated Sandstone Lithofacies 2C (Figure 6C) is dominated by horizontal laminations, minor current ripples with abundant thin (1-mm) clay and organic drapes that define horizontal laminations. Most laminations show some centimeter-scale cyclicity of mud-sand couplets. Lamina and bedsets typically fine upward. Bioturbation is sparse to moderate and is characterized by Planolites, Palaeophycus, Arenicolites, Skolithos, and Cylindrichnus. Fine plant debris and amber flakes and pebbles may be locally abundant. This lithofacies is interpreted to have been deposited on tidal flats and in associated low-energy, tidal point bars.

than those of the detrital quartz). Bioturbation is moderate to extensive and is characterized by Thalassinoides, Asterosoma, Planolites, Teichichnus, Palaeophycus, and Scolicia. Siderite concretions are common, particularly associated with Thalassinoides burrows. This facies is almost always cemented by calcite and grades into overlying limestone beds. Mollusk shells and shell fragments are common. Lithofacies 2D represents deposition in the distal parts of low-energy, marine shorelines.

Lithofacies 3B: Bioturbated, Muddy Sandstone This lithofacies (Figure 7A) consists of poorly to moderately well-sorted, upper fine-grained to lower medium-grained muddy sandstone. Bioturbation is typically extensive and is characterized by multiple tiers of Thalassinoides, Teichichnus, Asterosoma, Planolites, Palaeophycus, Skolithos, and Cylindrichnus. This lithofacies represents deposition along brackish, estuarine, and marine low-energy shorelines.

Lithofacies 2D: Glauconitic Sandstone

Lithofacies 3C: Hummocky Cross-bedded Sandstone

This lithofacies (Figure 6D) consists of fine-grained to medium-grained sandstone with abundant grains of glauconite (which are invariably coarser grained

Lithofacies 3C (Figure 7B), which is not often observed in the Burgan Formation, consists of moderately well- to very well-sorted, very fine-grained to

222 / Strohmenger et al.

FIGURE 7. Siliciclastic lithofacies 3, slabbed core photographs. (A) Lithofacies 3B: bioturbated, muddy sandstone containing shell fragments. (B) Lithofacies 3C: hummocky cross-bedded sandstone. (C) Lithofacies 3D: interlaminated sandstone, silt, and shale. (D) Lithofacies 3E: current-rippled, interbedded sandstone and mudstone.

lower fine-grained sandstone. The dominant sedimentary structures are hummocky, cross-stratified beds with wave-rippled and bioturbated upper surfaces. Mudstone beds as much as several centimeters in thickness are only rarely preserved between hummocky bedsets. Bioturbation is typically absent to sparse, but may increase upward with moderate burrowing at the top of some bed sets. Trace fossils recognized include Planolites, Palaeophycus, and uncommon Ophiomorpha. This lithofacies was most likely deposited in marine, proximal lower shoreface settings.

Lithofacies 3D: Interlaminated Sandstone, Silt, and Shale Lithofacies 3D (Figure 7C) consists of interlaminated, very fine-grained sandstone, silt, and shale. Sedimentary structures include horizontal bedding, starved current-ripple cross-lamination, and uncommon syneresis cracks. Depositional strata range from mostly clay to mostly sand, but are characterized by sand and mud couplets, thick and thin couplets, and centimeter-scale cycles of thickening and thinning sand laminae. Bioturbation is sparse to moderate and include trace fossils of Planolites, Palaeophycus, Sko-

lithos, and Cylindrichnus. Plant debris, amber, and siderite-cemented bands are common. This lithofacies represents deposition on proximal (sandy) to distal (muddy) tidal flats, and the centimeterscale cycles are interpreted as neap-spring tidal cycles.

Lithofacies 3E: Current-rippled, Interbedded Sandstone and Mudstone Lithofacies 3E (Figure 7D) consists of interbedded, lower medium-grained sandstone and mudstone. Sedimentary structures are dominated by horizontal lamination and current-ripple cross-lamination, with uncommon syneresis cracks. Bioturbation is sparse to moderate and is characterized mostly by Planolites, Palaeophycus, Skolithos, and Cylindrichnus. Plant debris and amber are locally abundant as well as sideritecemented bands. This lithofacies is interpreted to have been deposited in tidal flats and distal tidal bars.

Lithofacies 3F: Calcareous, Bioturbated Sandstone This lithofacies (Figure 8A) consists of fine-grained to lower medium-grained sandstone. Bioturbation is typically extensive (churned). Mollusk shells, shell fragments, and foraminifera are locally abundant.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 223

FIGURE 8. Siliciclastic lithofacies 3, slabbed core photographs. (A) Lithofacies 3F: calcareous, bioturbated sandstone containing shell fragments. (B) Lithofacies 3G: wave-rippled sandstone. (C) Lithofacies 3H: carbonaceous, muddy sandstone.

This facies is tightly cemented by calcite and grades into limestone beds above or below. This lithofacies represents deposition in shallowmarine shoreline or shoreface settings adjacent to subtidal carbonate facies.

common to slight and comprises Planolites burrows and rootworking. Large plant fossils and amber are very abundant. This lithofacies represents deposition in floodplain, crevasse splay, or levee settings.

Lithofacies 3G: Wave-rippled Sandstone

Lithofacies 4A: Heterolithic Siltstone to Mudstone

Lithofacies 3G (Figure 8B) consists of well-sorted, very fine-grained to fine-grained sandstone. Sedimentary structures consist of wave-ripple cross-laminations. Clay drapes are common between ripple laminae and laminae sets. Bioturbation is sparse to moderate and includes Astersoma, Teichichnus, and Planolites burrows. This lithofacies is interpreted to have formed in distal, lower shoreface settings.

This lithofacies (Figure 9A) consists mostly of mudstone with siltstone to very fine-grained sandstone stringers (>80% mudstone). Sedimentary structures include parallel-lamination, silt stringers, uncommon syneresis cracks, and starved current-ripple crosslamination. Bioturbation is uncommon to slight and is characterized by Planolites burrows. Plant fossils and amber are common to abundant. Lithofacies 4A is interpreted to represent deposition in flood-plain and abandoned channel-fill settings.

Lithofacies 3H: Carbonaceous, Muddy Sandstone Lithofacies 3H (Figure 8C) consists of very fine- to fine-grained sandstone with abundant clay and organic drapes ranging from 5 to 10 mm (0.2 to 0.4 in.) in thickness. Organic layers range from 1 mm (0.04 in.) to a few centimeters in thickness. Sedimentary structures are dominated by horizontal lamination and current-ripple cross-lamination. Bioturbation is un-

Lithofacies 4B: Bioturbated to Wave-rippled Sandstone to Mudstone This lithofacies (Figure 9B) is composed of lower very fine-grained sandstone to siltstone to clay. Sedimentary structures include isolated to amalgamated

224 / Strohmenger et al.

FIGURE 9. Siliciclastic lithofacies 4, slabbed core photographs. (A) Lithofacies 4A: heterolithic siltstone to mudstone. (B) Lithofacies 4B: bioturbated to wave-rippled sandstone to mudstone. This core interval has been extensively bioturbated by Teichichnus (Te), Asterosoma (As), and Planolites (Pl) burrows. (C) Lithofacies 4D: bioturbated mudstone. The upper interval of this core has been extensively bioturbated by Teichichnus (Te) burrows. (D) Lithofacies 4E: laminated siltstone.

wave-ripple cross-laminations. Bioturbation is slight to extensive (churned) and is characterized by Asterosoma, Teichichnus, Thalassinoides, Planolites, Palaeophycus, Scolicia, and Skolithos. Fine plant fragments and mollusk shells are uncommon. This lithofacies is interpreted as deposition in marine, distal lower shoreface settings.

Lithofacies 4D: Bioturbated Mudstone Lithofacies 4D (Figure 9C) is mostly mudstone, but may grade vertically into siltstone. Bioturbation is moderate to extensive (churned) and is characterized by mostly indistinct burrows with some discernible Planolites, Teichichnus, and Thalassinoides burrows. This lithofacies is inferred to have been deposited in offshore to lower shoreface settings.

Lithofacies 4E: Laminated Siltstone This lithofacies (Figure 9D) is composed of horizontally laminated to wavy laminated siltstone. Bioturbation is uncommon to slight and is characterized by small (<5-mm; <0.2-in.) indistinct horizontal burrows. Lithofacies 4E represents deposition in lacustrine, flood-plain, pond, and abandoned-channel settings.

Lithofacies 5A: Laminated Gray Shale Lithofacies 5A (Figure 10A) is composed of darkgray, horizontally laminated shale. Bioturbation is slight to moderate and is characterized by Planolites, Teichichnus, sand-filled Chondrites, and sparse Zoophycos burrows. Bivalve and gastropod shells range from sparse to common, and small plant fragments are uncommon. This lithofacies is interpreted to have been deposited in offshore marine settings, below storm-wave base.

Lithofacies 5B: Carbonaceous Mudstone Lithofacies 5B (Figure 10B) is composed of carbonaceous, horizontally laminated mudstone, which commonly displays postdepositional, compactional slickensides. Bioturbation is slight to moderate and is characterized mostly by rootworking (Figure 11A). Large leaves and amber (Figure 11B, C) are very abundant. This lithofacies represents deposition in clastic swamp and abandoned-channel settings.

Lithofacies 5C: Laminated Dark Gray Shale Lithofacies 5C (Figure 10C) is composed of very dark-gray, horizontally laminated shale. Bioturbation is uncommon to slight and is characterized by small

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 225

FIGURE 10. Siliciclastic lithofacies 5 and 6, slabbed core photographs. (A) Lithofacies 5A: laminated gray shale. (B) Lithofacies 5B: carbonaceous mudstone. (C) Lithofacies 5C: laminated dark-gray shale. (D) Lithofacies 6: coal.

Planolites burrows. Large leaves, sticks, and amber are very abundant. This lithofacies is interpreted to represent deposition in lacustrine, abandoned channel, and floodplain settings.

Lithofacies 6: Coal This lithofacies (Figure 10D) is composed exclusively of coal. Some coals are rooted and represent peat swamps. Other coals are composed of allochthonous plant detritus and formed in abandoned channels to estuarine settings with low clastic input.

Facies, Facies Associations, Depositional Environments, and Reservoir Quality

FIGURE 11. Slabbed core photographs of (A) rooted horizon, (B) amber, and (C) fragment of amber.

The 22 lithofacies are grouped into 6 distinct facies. These groupings are either spatially reoccurring groups of different lithofacies or thick occurrences of the same lithofacies. In

226 / Strohmenger et al.

turn, the six facies are grouped into two unique facies associations. The facies associated with each are mutually exclusive (Figure 4).

valley system (sensu Van Wagoner et al., 1990). The typical idealized Burgan updip to downdip facies succession and associated lithofacies are

Facies Association I



Facies association I (Figure 4) consists of four unique facies, which are interpreted as the normal, shoaling-upward facies succession of a low-energy, wave-dominated, shoreline (sensu Walker and Plint, 1992). This facies association is the basic building block of highstand and transgressive deposits in the Burgan Formation. The typical idealized vertical facies succession in this facies association from top to base is   



carbonaceous facies (lithofacies 1C, 3H, 5B, and 6): coastal-plain and backshore bioturbated to stratified sandstone facies (lithofacies 3B and 3C): proximal lower shoreface bioturbated to interstratified mudstones and sandstones (lithofacies 2D, 3F, 3G, 4B, and 4D): distal lower shoreface bioturbated mudstone facies (lithofacies 5A): offshore

Lithofacies association I is interpreted to represent the deposits of a low-energy shoreline environment. Wave-dominated shoreface sandstones are oriented along a northwest-southeast belt and primarily occur in northeastern Kuwait. Shoreface deposits change facies to marginal-marine sandstones and mudstones and offshore mudstones in a northeast transect. They change facies updip to coastal-plain deposits, which dominate the southwestern region of Kuwait. The shoreface deposits are interpreted to represent a lowenergy environment based on the paucity of highenergy stratification and the presence of strata that have been moderately to intensely bioturbated by trace fossil assemblages indicative of open-marine conditions. In general, reservoir quality ranges from moderate in deposits of the proximal lower shoreface to poor in the distal lower shoreface and offshore strata. Although sandstone reservoirs may exist within small (<0.5-km [<0.31-mi]-wide) fluvial channels on the coastal-plain, for the most part, the coastal-plain part of this succession is mud prone and nonprospective.

Facies Association II Facies association II (Figure 4) consists of two unique facies, which are interpreted as the normal updip to downdip facies variation in an incised-



cross-stratified sandstone facies (lithofacies 1A and 1C): fluvial-dominated valley fill heterolithic tidal facies (lithofacies 1B, 2A, 2B, 2C, 3D, 3E, and 4A): tidal-dominated valley fill

The incised valleys, identified in this study, typically trend southwest to northeast across Kuwait and are perpendicular to somewhat oblique to the shoreline trend of the underlying highstand and overlying transgressive deposits. Deposits of the incised valley are characterized by fluvial-dominated facies in updip regions to the southwest and tidal-dominated facies in the distal regions to the northeast. The fluvial facies are interpreted as low-sinuosity, braidedstream deposits based on the prevalence of finingupward, fine-grained to coarse-grained, trough–crossbeds and bedsets. The fluvial strata change facies downdip to tidal-influenced deposits that possess more heterolithic lithofacies, indicative of fluctuation-energy conditions. Reservoir quality is excellent in the braided fluvial strata and in the updip parts of the mixed fluvial-tidal systems. However, reservoir quality rapidly diminishes in the seaward direction as the valley systems thin and become more mudstone prone.

High-frequency Sequences Nineteen high-frequency depositional sequences were interpreted within the Burgan Formation, as defined in this study (Figure 3). From bottom to top, these are B900, B850, B800, B750, B725, B700, B650, B600, B550, B500, B450, B400, B350, B300, B250, B200, B150, B125, and B100 (LST). In the nonmarine to marine intervals of the Burgan Formation, high-frequency sequence boundaries are interpreted as abrupt vertical changes in stratal stacking pattern or abrupt basinward shift in environments of deposition. In the downdip position, high-frequency sequence boundaries are evident as abrupt changes from coarsening-upward marginalmarine parasequences to blocky fluvial-tidal deposits. Updip, these sequence boundaries can be traced into an abrupt change from interpreted low net/gross coastal-alluvial plain (transgressive or highstand) deposits (below) into blocky fluvial (lowstand) deposits (above; Figure 12). Downdip and laterally, the sequence boundaries and transgressive surface become coincident. However, in the most distal sequences in

FIGURE 12. Regional cross section showing facies distribution and sequence-stratigraphic framework of the Burgan Formation. Note that the fluvial-dominated sandstones of the incised-valley fills, which constitute the lowstand systems tracts, thicken toward the south in the vicinity of the Greater Burgan field area. The overlying Mauddud clastic member and Mauddud carbonate member also display thickness variations along the depositional transect. The detailed sequence-stratigraphic framework for the Mauddud Formation is shown in Figures 22 and 23. GR = gamma-ray log; MD = measured depth (feet); RES = resistivity log; DT = sonic log; NEU = neutron porosity log; DENS = density log.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 227

stratigraphic architecture of the B100 high-frequency, depositional sequence, which is the uppermost lowstand systems tract in the lowstand sequence set of the upper, third-order composite sequence. (A) The lowstand deposits fill incised valleys and are composed of tidal-influenced strata in the downdip positions and fluvialdominated strata in the updip regions. (B) Incised valleys become thinner and more discontinuous downdip. IVF = incised-valley fills; HST = highstand systems tract.

FIGURE 13. Depositional environment and isopach maps of a lowstand systems tract in the Burgan Formation. These maps are interpreted from the sequence-

228 / Strohmenger et al.

the Burgan Formation, an onlapping succession of carbonates appears to mark a depositional shelf break and downdip limit of clastic progradation in each sequence. In more marine-dominated parts of the Burgan Formation, the TSs are interpreted at the vertical change from prograding to retrograding marine parasequences. In more medial settings, these surfaces coincide with the vertical facies change, from blocky, fluvial-tidal deposits below to retrograding marine parasequences above. In more proximal (updip) settings, the TSs are placed at the vertical change from blocky fluvial-tidal below to low net/gross coastal- and alluvialplain deposits above (Figure 12). The MFSs of these sequences are interpreted at major marine incursions marked by a change from retrograding to prograding stacking of marine parasequences. In the downdip parts of the most distal sequences, marine limestones occur beneath interpreted MFSs (Figure 12). Lowstand systems tracts primarily consist of incised-valley systems that are filled by tidalinfluenced sandstones and mudstones in the downdip regions and braided fluvial sandstones in updip regions (Figure 13A). In the low-relief, HFSs identified within the Burgan Formation, onlapping lowstand clastic wedges, as well as basin-floor lowstand fans are absent. Clastic lowstand deposition appears limited to IVF. The valleys become thinner and more discontinuous downdip (Figure 13B) and do not appear to extend to the maximum downdip progradational limit of underlying sandstone-prone highstands. These relationships strongly suggest that transgressive erosion has modified the

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 229

FIGURE 14. Paleogeographic map showing the depositional environments of a highstand systems tract in the Burgan Formation. This paleogeographic map is interpreted from the sequence-stratigraphic architecture of the B600 highfrequency, depositional sequence, which is a highstand systems tract in the transgressive sequence set of the lower, third-order composite sequence.

primary distribution and thickness patterns in the most distal (downdip) parts of the Burgan incisedvalley systems. The TSTs correspond to retrogradational successions of coarsening-upward strata and display a systematic downdip to updip change from marine carbonates, to marginal-marine mudstones and sandstones, to mudstone-prone alluvial- and coastal-plain deposits. Highstand systems tracts correspond to progadational successions of coarsening-upward, dominantly marginal-marine sandstones and mudstones downdip and alluvial- and coastal-plain mudstones and

sandstones updip (Figure 14). The shorelines in these highstands trend northwest– southeast and are best developed in northern Kuwait.

Composite Sequences The 19 HFSs identified within the Burgan Formation define four distinct sequence sets that form parts of two composite sequences. The basal four HFSs (B900, B850, B800, and B750) stack in an overall aggradational succession and form the LSS of the lower Burgan composite sequence (Figure 3). This aggradational sequence set is dominated by braided fluvial deposits that contain few mudstone breaks

230 / Strohmenger et al.

(Figure 12). The basal aggradational sequence set is overlain by six HFSs (B725, B700, B650, B600, B550, and B500) that stack in an overall retrogradational pattern, which constitutes the TSS of the lower Burgan composite sequence (Figure 3). In the Raudhatain and Sabiriyah field area in northern Kuwait, the basal HFSs in this retrogradational succession are dominated by nonmarine facies, whereas the uppermost sequences contain increasing marginal-marine components (Figure 12). The next four HFSs (B450, B400, B350, and B300) stack in an overall progradational succession and comprise the HSS of the lower Burgan composite sequence (Figure 3). In the northern fields area, these sequences are dominated by marginal-marine highstand deposits (Figure 12). The uppermost five sequences (B250, B200, B150, B125, and the LST of B100) stack in an aggradational pattern and form the LSS of the upper Burgan and Mauddud upper composite sequence (Figure 3). In the northern fields area, these HFSs consist of alternating nonmarine LSTs and marginal-marine TSTs and HSTs (Figure 12). Based on this delineation of sequence sets, a composite sequence boundary B900_SB is placed at the base of the Burgan Formation (Figure 3). In general, the overlying aggradational, retrogradational, and progradational sequence sets correspond to the LSS, TSS, and HSS of the lower composite sequence. It should be noted, however, that the high-frequency transgressive surface B725_TS and the high-frequency maximum flooding surface B500_MFS are used as composite TS and MFS, respectively, for the lower composite sequence (Figures 3, 12). The uppermost aggradational sequence set is interpreted as a second LSS. The composite transgressive surface B100_TS is interpreted as the TS of this composite sequence, with strata in the overlying Mauddud Formation, forming the TSS and the HSS of the younger composite sequence. In general, the various sequence sets are thicker and more sandstone prone in the southwest area of Kuwait and thinner and more mudstone prone in the northeast (Figures 12 – 14).

Reservoir Quality Reservoir quality of the Burgan Formation is closely related to the interpreted depositional environments. Fluvial-dominated sandstones, which include lithofacies 1A and 1C, possess the best reservoirquality attributes. These lithofacies types have average porosity and permeability values of 25% and 1600 md, respectively. Tidal-dominated sandstones,

which encompass lithofacies 1B, 2A, 2B, 2C, and 2F, exhibit moderately good reservoir-quality properties. They have average porosity and permeability values of 23% and 270 md, respectively. The marginal-marine deposits, which include lithofacies 2A, 2E, and 3B, generally display poorer reservoir quality as a result of variable extents of bioturbation. Average porosity and permeability values for these lithofacies are 19% and 10 md, respectively.

Reservoir Quality Distribution Stratal distributions, thickness variations, and regional facies architecture of the Burgan systems tracts can be interpreted from isopach and paleogeographic maps of these intervals (Figures 13, 14). Highfrequency lowstands of the Burgan Formation consist of fluvial and tidal deposits that filled incised valleys. In general, the incised valleys become thinner, narrower, and more tidal influenced downdip to the northeast (Figure 13). Conversely, the valleys become thicker, more widespread, and more fluvially influenced updip to the south and southwest (Figure 13). High-frequency highstands of the Burgan consist of shoreline deposits, which, in general, trend northwest–southeast, with more marine facies to the northeast and more nonmarine facies toward the southwest (Figure 14). The sequence-stratigraphic framework presented in this paper provides an improved understanding of the distribution of sandstone-prone lowstands and marginal-marine highstands in each of the 19 HFSs identified within the Burgan Formation (Figure 12). Furthermore, some additional new play concepts were identified. Within the context of the new Mauddud– Burgan boundary, distinct isolated incised-valley systems in the clastic member of the Mauddud Formation can be defined. Within the Burgan proper, the possibility exists that marginal-marine shorelines in the TSSs may also contain hydrocarbons off the flank of structures. This combined structural-stratigraphic trap would depend on mudstone-prone coastal-plain deposits in each sequence acting as the updip lateral seal and marine shales in the overlying sequence acting as the top seal.

MAUDDUD FORMATION A sequence-stratigraphic framework for the Mauddud Formation has been established using all available well and core data from Kuwait. The Mauddud Formation can be described by the TSS and the HSS

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 231

FIGURE 15. Schematic facies model of the Mauddud transgressive sequence set (upper third-order composite sequence).

as a third-order composite sequence that is composed of eight high-frequency, depositional sequences. The top (MAU100_TS) and base (B100_TS) of the Mauddud Formation are marked by TSS (Strohmenger et al., 2002; Demko et al., 2003) (Figure 3). The lower Mauddud (Mauddud TSS) shows a lateral change in lithology from limestone in northern Kuwait at the Raudhatain and Sabiriyah fields area to siliciclastics in southern (Greater Burgan field area; Figure 15) and southeastern Kuwait (Minagish field area). The upper Mauddud (Mauddud HSS) is mostly eroded in southern and southwestern Kuwait (Strohmenger et al., 2002; Demko et al., 2003) (Figure 16). The Mauddud Formation in Kuwait comprises eight carbonate lithofacies and three siliciclastic lithofacies. Carbonate lithofacies were deposited in inner to lower ramp, normal to slightly restricted environments. Clastic lithofacies were deposited in innerramp (deeper lagoon), nearshore marine, and offshore marine environments (Strohmenger et al., 2002; Demko et al., 2003) (Figures 15, 16), similar to those described for the Burgan Formation. The Cenomanian Wara Formation overlies the Mauddud Formation and provides a regional top seal (Figure 3). It characteristically is a slightly calcareous to noncalcareous marine shale in northern Kuwait

(Raudhatain and Sabiriyah fields area). In southern (Greater Burgan field area) and southwestern Kuwait (Minagish field area), glauconite-rich, deltaic sandstones are more common.

Lithofacies Types and Depositional Model Sediments of the Mauddud Formation were deposited along a gently northward- and northeastward-dipping homoclinal ramp (Strohmenger et al., 2002; Demko et al., 2003). Lithofacies and lithology of the Mauddud Formation of Kuwait vary by geographic area and time. In the Mauddud transgressive sequence set, siliciclastic deposits of southern (Greater Burgan field area) and southwestern Kuwait (Minagish field area) grade into carbonates toward the north and northeast (Abdali, Raudhatain, and Sabiriyah fields area; Figure 15). The Mauddud Formation can be described by means of 11 lithofacies types (F1 – F11). The facies scheme follows, in great parts, the one established by I. Goodall, N. Cross, and D. Payne (1996, personal communication), C. Hollis, N. Cross, T. Needham, and B. Jones (1998, personal communication), and N. Cross, R. Heath, and G. Paintal (1999, personal communication), with some modifications. Lithofacies types range from moderate- to high-energy upper-ramp

232 / Strohmenger et al.

FIGURE 16. Schematic facies model of the Mauddud highstand sequence set (upper third-order composite sequence).

deposits (F1, F2, and F3) through medium-energy, upper- to middle-ramp (F4), and lagoonal deposits (F4, F5 and F6) to low-energy, lower-ramp (F7, F8, and F11), and deeper lagoonal deposits (F7, F8, F9, F10, and F11; Figures 15, 16). The thickness of the Mauddud Formation increases toward the north and northeast and decreases toward the south and southwest. The thinning is the result of reduced accommodation during the lower Mauddud TSS (facies change from carbonates to siliciclastics; Figure 15) and pronounced erosion of the upper Mauddud HSS in southern (Greater Burgan field area) and southwestern Kuwait (Minagish field area) (Figure 16).

Lithofacies 1 (F1): Rudist Floatstone to Rudstone Lithofacies type F1 (Figure 17A) is rich in rudists or rudist fragments. Miliolids, conical orbitolinids, other benthic foraminifera, skeletal fragments, and echinoderms are common. Discoidal orbitolinids are uncommon. Nonskeletal grains are peloids. Bioturbation is moderate. This lithofacies type commonly grades into the high-energy skeletal-peloidal grainstone (lithofacies F2), as well as into the restricted lagoon nodular, miliolid-bearing packstone-wackestone (lithofacies

F5). The grain composition (occurrence of miliolids), as well as the facies-stacking pattern, suggests the rudist floatstone-rudstone represents a ramp-crest to inner-ramp, moderate to high-water-energy sediment, deposited laterally or mostly lagoonward of lithofacies F2 (skeletal-peloidal grainstone). Relatively thick accumulations (as much as 60 ft [18 m]) of rudist floatstone-rudstone occur within Mauddud sequences MAU450 and MAU200. Especially within Mauddud sequence MAU450, rudist floatstone-rudstone forms very porous intervals that are easily identified by low gamma-ray- and highresistivity-log responses. Cementation is minor. Dominant cement type is blocky calcite cement, partly to completely filling the molds of dissolved rudist shells. Dominant porosity types are moldic and vuggy porosity. The average porosity is 18%, and the typical permeability is 5 md.

Lithofacies 2 (L2): Skeletal and Peloidal Grainstone Lithofacies type F2 (Figure 17B) is rich in conical orbitolinids, skeletal fragments, and echinoderms. Other benthic foraminifera, red algae, and green algae are common. Discoidal orbitolinids and gastropods

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 233

FIGURE 17. Carbonate lithofacies types, slabbed core photographs. (A) Lithofacies F1: rudist floatstone-rudstone. (B) Lithofacies F2: skeletal-peloidal grainstone showing Glossifungites (Gl) burrows as well as fenestral structures (keystone vugs). (C) Lithofacies F3: skeletal-peloidal mud-lean packstone showing bioturbation. (D) Lithofacies F4: bioturbated skeletal-peloidal packstone.

are sparse. Nonskeletal grains are coated grains and peloids. Bioturbation is low. This lithofacies type frequently shows Glossifungites burrows and/or, less common, desiccation cracks (Figures 17B, 18A, B), as well as, also more uncommonly, karstification (Figure 18C). Sedimentary structures indicating intertidal conditions, such as trough – cross-bedding, fenestral structures (keystone vugs and sheet cracks; Figure 17B), geopedal fillings (vadose silt), and circumgranular cracks, are generally limited to lithofacies type F2. This lithofacies generally occurs at the top of the Mauddud parasequences and HFSs. The lack of ma-

trix and the overall facies-stacking pattern suggest the skeletal-peloidal grainstone to represent a highenergy, shoal (ramp-crest) to upper-ramp deposits. Cementation varies between minor and extensive. Typical cement types are neomorphosed isopachous rim cement, blocky calcite cement, and syntaxial calcite cement. Glossifungites burrows as well as karst fillings are commonly dolomitized. Dominant porosity type is microporosity. Interparticle, intraparticle, and moldic porosity are minor porosity types. The average porosity is 15%, and the typical permeability is 2 md.

234 / Strohmenger et al.

FIGURE 18. Sedimentary structures, slabbed core photographs. (A) Erosional surface with Glossifungites burrows. (B) Glossifungites burrows and/or desiccation cracks with younger sediment infill. (C) Microkarst with sediment infill.

Cementation generally is minor. Cement types are blocky calcite and syntaxial calcite cements. Dominant porosity type is microporosity. Intraparticle and moldic porosity are minor porosity types. The average porosity is 15%, and the typical permeability is 2 md.

Lithofacies 4 (F4): Bioturbated Skeletal and Peloidal Packstone

Lithofacies 3 (F3): Skeletal and Peloidal Mud-lean Packstone Lithofacies type F3 (Figure 17C) is rich in conical orbitolinids. Skeletal fragments and echinoderms are common. Discoidal orbitolinids, other benthic foraminifera, and gastropods are sparse. Nonskeletal grains are peloids. Bioturbation varies from moderate to high. This lithofacies type commonly alternates with lithofacies type F2 (skeletal-peloidal grainstone). The relatively low mud content and the overall faciesstacking pattern suggests the skeletal-peloidal mudlean packstone to represent a moderate-energy, upperramp deposit. Like lithofacies type F2 (skeletal-peloidal grainstone), the skeletal-peloidal mud-lean packstone locally is karstified.

Lithofacies type F4 (Figure 17D) is rich in conical and discoidal orbitolinids and echinoderms. Skeletal fragments are common. Other benthic foraminifera, rudist fragments, and thin-shelled bivalves are sparse. Nonskeletal grains are peloids. Bioturbation is high. This lithofacies typically underlies lithofacies type F3 (skeletalpeloidal mud-lean packstone) and, less commonly, lithofacies type F5 (nodular, miliolid-bearing packstone-wackestone). It commonly overlies lithofacies type F2 (skeletal-peloidal grainstone) and lithofacies type F3 (skeletal-peloidal mudlean packstone). Grain composition, texture, and the overall facies-stacking pattern suggest that the bioturbated skeletal-peloidal packstone represents a lowto moderate-energy, upper- to middle-ramp deposit. It may also be present behind the high-energy bar deposits (lithofacies F1, F2, and F3) in an inner-ramp, protected lagoonal environment. Cementation is minor to extensive. Dominant cement type is blocky calcite cement. Dominant porosity type is microporosity. Intraparticle and moldic porosity are minor porosity types. The average porosity is 12%, and the typical permeability is 0.5 md.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 235

FIGURE 19. Carbonate lithofacies types, slabbed core photographs. (A) Lithofacies F5: nodular, miliolid-bearing packstonewackestone. (B) Lithofacies F6: skeletal wackestone-mudstone (C) Lithofacies F7: clay-rich, bioturbated skeletal wackestone showing Thalassinoides (Th) and Planolites (Pl) burrows. (D) Lithofacies F8: bioturbated glauconitic packstone.

Lithofacies 5 (F5): Nodular, Miliolid-bearing Packstone to Wackestone Lithofacies type F5 (Figure 19A) commonly contains miliolids, conical orbitolinids, other benthic foraminifera, rudist fragments, skeletal fragments, and thin-shelled bivalves. Sponge spicules are common and only occur within this lithofacies type. Discoidal orbitolinids and echinoderms are uncommon. Nonskeletal grains are peloids. Bioturbation and burrowing is high, most probably causing its nodular appearance. This lithofacies type is restricted to the upper Mauddud (Mauddud HSS; Figure 16) and commonly over- and underlies the rudist floatstone-rudstone (lithofacies F1), predominantly within Mauddud sequence MAU200. The grain composition (occurrence

of miliolids), as well as the facies-stacking pattern, suggests the nodular, miliolid-bearing packstonewackestone to represent a low- to moderate-energy, inner-ramp, restricted lagoonal deposit. Cementation is minor, and the dominant cement type is blocky calcite cement. The dominant porosity type is microporosity. Intraparticle and moldic porosity are minor porosity types. The average porosity is 19%, and the typical permeability is 0.5 md.

Lithofacies 6 (F6): Skeletal Wackestone to Mudstone Lithofacies type F6 (Figure 19B) is mud rich, light gray, with sparse skeletal fragments and peloids. Bioturbation is very low. Pyrite framboids are quite frequent.

236 / Strohmenger et al.

This lithofacies type is present only in the upper Mauddud HSS (Mauddud sequence MAU300; Figure 16). We interpret this lithofacies type to represent a lowenergy, inner-ramp, protected lagoonal deposit. Cementation (fine-crystalline, blocky calcite filling the pore space of dissolved skeletal fragments) is minor. The dominant porosity type is microporosity, and intercrystalline porosity is present. Average porosity and typical permeability (only four samples analyzed) is within the range of lithofacies F5 (nodular, miliolid-bearing packstone-wackestone).

Lithofacies 7 (F7): Clay-rich, Bioturbated Skeletal Wackestone Lithofacies type F7 (Figure 19C) is rich in discoidal orbitolinids and echinoderms, as well as in trace fossils, including Thalassinoides and Planolites. Skeletal fragments, thin-shelled bivalves, and pelagic foraminifera are common. Nonskeletal grains are peloids. Bioturbation is very high (solution-seam rich). Pyrite is common. This lithofacies typically underlies lithofacies type F4 (bioturbated skeletal-peloidal packstone) or overlies lithofacies type F2 (skeletal-peloidal grainstone) and lithofacies type F3 (skeletal-peloidal mud-lean packstone). Grain composition, texture, and the overall facies-stacking pattern suggest the clay-rich, bioturbated skeletal wackestone to represent a low-energy, middle- to lower-ramp deposit. It may, however, also be present behind the high-energy bar deposits (lithofacies F1, F2, and F3) in the inner-ramp, protected lagoon environment. Cementation is minor. The dominant cement type is fine-crystalline, blocky calcite cement. Dolomitization occurs along solution seams. The dominant porosity type is microporosity, and intraparticle and intercrystalline porosity are minor porosity types. The average porosity is 4%, and the typical permeability is 0.01 md.

Lithofacies 8 (F8): Bioturbated Glauconitic Packstone Lithofacies type F8 (Figure 19D) commonly contains discoidal orbitolinids, echinoderms, and skeletal fragments. Conical orbitolinids, thin-shelled bivalves, and rudist fragments are uncommon. Nonskeletal grains are glauconite (glauconitized peloids), peloids, quartz, and pyrite. Bioturbation is high. This lithofacies type typically overlies sequence boundaries on top of lithofacies types F2 (skeletalpeloidal grainstone) and F3 (skeletal-peloidal mud-

lean packstone). We interpret this lithofacies type as representing periods of low sedimentation rates during rapid sea-level rise. It dominantly occurs within the Mauddud TSS. The depositional environment ranges from inner ramp, lagoonal (Mauddud TSS; Figure 15) to middle to lower ramp (Mauddud HSS; Figure 16). Cementation is minor to extensive. Dominant cement types are blocky calcite, ferroan dolomite, and ferroan calcite cement. Ferroan cements are most common in southern Kuwait (Greater Burgan field area). Dominant porosity type is microporosity. Intraparticle and moldic porosity are minor porosity types. The average porosity is 7%, and the typical permeability is 0.05 md.

Lithofacies 9 (F9): Bioturbated Glauconitic Sandstone Lithofacies type F9 (Figure 20A) is quartz rich and contains sparse discoidal orbitolinids and skeletal fragments. It is rich in trace fossils, including Teichichnus, Asterosoma and Terebellina (uncommon). Grain types are detrital quartz, glauconite, pyrite, and siderite. Bioturbation is very high. This lithofacies is similar to lithofacies 2D (glauconitic sandstone) of the Burgan Formation but has a higher glauconite content. Like lithofacies type F8 (bioturbated glauconitic packstone), this lithofacies typically overlies sequence boundaries on top of lithofacies types F2 (skeletalpeloidal grainstone), F3 (skeletal-peloidal mud-lean packstone), and F10 (bioturbated mud-rich sandstone). It dominantly occurs within the Mauddud TSS (Figure 15). The bioturbated glauconitic sandstone is interpreted to represent an inner-ramp, nearshoremarine, lower-shoreface deposit. Cementation is minor to extensive. Dominant cement types are blocky calcite, ferroan dolomite, and ferroan calcite cement. Predominant porosity type is microporosity, and intraparticle and intergranular porosity types are present. The average porosity is 10%, and the typical permeability is 0.3 md.

Lithofacies 10 (F10): Bioturbated Mud-rich Sandstone Lithofacies type F10 (Figure 20B) is intensely bioturbated and rich in trace fossils, including Teichichnus, Asterosoma, Thalassinoides, Planolites, and Chondrites. Lithofacies type F10 occurs within the Mauddud TSS (Figure 15). Relatively thick intercalations (>60 ft; >18 m) occur within the Mauddud sequence MAU500. This lithofacies is similar to lithofacies 3F (calcareous, bioturbated sandstone) of the Burgan Formation.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 237

FIGURE 20. Siliciclastic lithofacies types, slabbed core photographs. (A) Lithofacies F9: bioturbated glauconitic sandstone showing Teichichnus (Te) burrows. (B) Lithofacies F10: bioturbated mud-rich sandstone showing Teichichnus (Te) and Thalassinoides (Th) burrows. (C) Lithofacies F11: darkgray shale and mudstone.

The bioturbated mud-rich sandstone is interpreted to represent an inner-ramp, nearshore-marine, lowershoreface deposit. Cementation is minor; dominant porosity type is intergranular. The average porosity is 12%, and typical permeability is 11 md.

Lithofacies 11 (F11): Dark-gray Shale and Mudstone Lithofacies F11 (Figure 20C) shows only sparse skeletal fragments, thin-shelled bivalves, planktonic foraminifera, and nannofossils. This lithofacies is similar to lithofacies 5A (laminated gray shale) of the Burgan Formation. Millimeter-scale laminae indicate low bioturbation caused by stressed (lagoon) or deeper marine, offshore (lower-ramp) environment of deposition. Dominant lithofacies type of the Mauddud transgressive sequence set (Figure 15); deposited largely in an inner-ramp, deeper lagoonal setting, juxtaposed to lithofacies types F7 (clay-rich, bioturbated wackestone), F8 (bioturbated glauconitic packstone), and F9

(bioturbated glauconitic sandstone; Figure 15). The fact that it is not highly bioturbated as it would be expected for offshore shales underlying lower shoreface deposits (lithofacies F9 and F10) indicates a high-stress environment. The dark-gray shale and mudstone are interpreted to represent a low-energy, deeper marine, lower-ramp to basinal deposit during the Mauddud HSS (Figure 16) and the Wara Formation. Predominant porosity type is microporosity. The average porosity is 5%, and typical permeability is 0.03 md.

High-frequency Sequences Eight high-frequency, depositional sequences were interpreted within the Mauddud Formation. From bottom to top, these are B100 (TST and HST), MAU600, MAU500, MAU450, MAU400, MAU350, MAU300, and MAU200 (Figure 3). Eight SBs, two TSs, three MFSs, and two FSs were identified. All surfaces, except for one flooding surface (MAU600_FS), can be correlated throughout all of Kuwait. Sequence boundaries typically are characterized by an abrupt change in texture and lithofacies. Glossifungites burrows and/or desiccation cracks are common in the upper part of each high-frequency sequence (Figures 18A, B; 21). Karstification is seen locally (Figure 18C). Evidence of at least periodic exposure, such as fenestral structures (keystone vugs and sheet cracks; Figure 17B) and circumgranular cracks, is also present in the upper part of each high-frequency sequence. Sandstone or packstone, rich in glauconite,

238 / Strohmenger et al.

FIGURE 21. Mauddud Formation idealized carbonate parasequence (upper part of a high-frequency sequence) showing shallowing-upward trend of facies from base to top: bioturbated skeletal-peloidal packstone (F4, blue color), skeletal-peloidal mud-lean packstone (F3, red color), and skeletal-peloidal grainstone (F2, orange color). Flooding surface/sequence boundary (FS/SB) is interpreted by Glossifugites burrows or, uncommonly, karstification (microkarst) and is overlain by bioturbated glauconitic packstone (F8) or bioturbated glauconitic sandstone (F9, green color).

typically immediately overlies each high-frequency sequence boundary (Figure 21), recording periods of rapid flooding and resulting in low sedimentation rate. The identified high-frequency sequence boundaries correspond to combined flooding surface and sequence boundaries (FS/SB, Figure 21). Maximum flooding surfaces are interpreted at major marine incursions marked by a change from retrograding to prograding stacking of marine parasequences. Flooding surfaces are interpreted on top of shallowingupward parasequences, commonly corresponding to Glossifungites surfaces (Figure 21).

Mauddud Sequence B100 (B100_SB to MAU600_SB) In northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22), this sequence is shale and mud rich at the base of the TST, grading upward, to grain rich at the top of the HST. Dominantly tidalflat deposits occur in southern Kuwait (Greater

Burgan field area; Figure 23), and dominantly coastalplain deposits occur in southwestern Kuwait (Minagish field area; Figure 23). The sequence is partly eroded by Mauddud sequence boundary MAU600_SB in southern Kuwait (Figure 23). The sequence consists of lithofacies F11 (dark-gray shale and mudstone) and lithofacies F9 (bioturbated glauconitic sandstone) grading upward into lithofacies F7 (clay-rich, bioturbated skeletal wackestone), lithofacies F8 (bioturbated glauconitic packstone), and lithofacies F4 (bioturbated skeletal-peloidal packstone) in northern Kuwait (Raudhatain and Sabiriyah fields area; Figures 22, 23).

Mauddud Sequence MAU600 (MAU600_SB to MAU500_SB) This sequence is composed of mud-dominated carbonates in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22), dominantly coastal-plain and tidal-influenced deposits in southern Kuwait

FIGURE 22. Field-scale cross section (northwest – southeast) showing facies distribution and sequence-stratigraphic framework of the Mauddud Formation throughout the northern Kuwait Raudhatain and Sabiriyah fields. Mauddud transgressive sequence set: B100_TS to MAU400_MFS. Mauddud highstand sequence set: MAU400_MFS to MAU100_SB. Mauddud lowstand systems tract: MAU100_SB to MAU100_TS. GR = gamma-ray log; FAC = Mauddud carbonate and siliciclastic facies and Burgan deposits (based on core); MD = measured depth (feet); DT = sonic log; NEU = neutron porosity log; DENS = density log.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 239

FIGURE 23. Regional cross section oriented north – south and east – west showing facies distribution and sequence-stratigraphic framework of the Mauddud Formation. Note that chronostratigraphic boundaries (time lines) crosscut the lithostratigraphic boundary between the Mauddud carbonate member (Mauddud Formation) and the Mauddud clastic member (Burgan Formation, time equivalent to Mauddud Formation). The proposed sequence-stratigraphic correlation is supported by age dating and palynofacies analyses (T. D. Davies and T. C. Huang, 2000, personal communication). Stratigraphy-diagnostic palynofacies assemblages are shown as colored dots. Blue dots: only found above Mauddud transgressive surface MAU100_TS (in Wara Formation). Yellow dot: only found between Mauddud sequence boundary MAU100_SB and Mauddud trangressive surface MAU100_TS (Mauddud lowstand systems tract, onlapping on Mauddud sequence boundary MAU100_SB in southern and southwestern Kuwait). Green dots: only found between Burgan transgressive surface B100_TS and Mauddud maximum flooding surface MAU400_MFS (Mauddud transgressive sequence set: lower part of Mauddud carbonate member in northern Kuwait and Mauddud clastic member in southern and southwestern Kuwait). Red dots: only found below Burgan transgressive surface B100_TS (Burgan lowstand sequence set). No stratigraphy-diagnostic palynofacies assemblage was found in grain-dominated, shallow-water carbonates between Mauddud maximum flooding surface MAU400_MFS and Mauddud sequence boundary MAU100_SB (Mauddud highstand sequence set). Color codes for Mauddud carbonate and siliciclastic facies as well as for Burgan siliciclastic deposits are shown in Figure 22. GR = gamma-ray log; FAC = Mauddud carbonate and siliciclastic facies and Burgan deposits (based on core); MD = measured depth (feet); DT = sonic log; NEU = neutron porosity log; DENS = density log.

240 / Strohmenger et al.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 241

(Greater Burgan field area; Figure 23), and dominantly coastal-plain and tidal-flat deposits in southwestern Kuwait (Minagish field area; Figure 23). Dominantly lithofacies F11 (dark gray shale and mudstone) and minor lithofacies F8 (bioturbated glauconitic sandstone) and lithofacies F7 (clay-rich, bioturbated skeletal wackestone) occur in the northern fields area. The succession may show upward increase of porosity and permeability and shallowingupward lithofacies trend, as well as coarsening-upward texture in the northern fields area (Figures 21, 22).

Mauddud Sequence MAU500 (MAU500_SB to MAU450_SB) Marginal-marine (distal lower shoreface) siliciclastics occur in northern Kuwait (Raudhatain and Sabiriyah fields area) and were not eroded by sequence boundary MAU450_SB (Figure 22). Dominantly coastalplain deposits occur in southern Kuwait (Greater Burgan field area; Figure 23) and dominantly tidal-influenced to fluvial deposits occur in southwestern Kuwait (Minagish field area; Figure 23).

Mauddud Sequence MAU450 (MAU450_SB to MAU400_SB) This sequence is composed of grain-dominated carbonates in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22) and dominated by marginal-marine (distal lower shoreface) deposits in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23). The sequence is composed predominantly of lithofacies F2 (skeletal-peloidal grainstone), with minor lithofacies F3 (skeletal-peloidal mud-lean packstone) and lithofacies F4 (bioturbated skeletal-peloidal packstone). Relatively thick accumulations (as much as 60 ft [18 m]) of highly porous lithofacies F1 (rudist floatstone-rudstone) occur as aerially restricted intercalations throughout the northern fields area. This sequence may show an upward increase of porosity and permeability and a shallowing-upward lithofacies trend, as well as coarsening-upward texture (northern fields area; Figures 21, 22).

Mauddud Sequence MAU400 (MAU400_SB to MAU350_SB) This sequence is composed of grain-dominated carbonates in northern Kuwait (Raudhatain and Sabiriyah fields area), with lithofacies 8 (bioturbated glauconitic packstone) commonly overlying Mauddud sequence boundary MAU400_SB (Figure 22). Lithofacies F8 (bioturbated glauconitic packstone) overlies

Mauddud sequence boundary MAU400_SB in southern Kuwait (Greater Burgan field area; Figure 23). Lithofacies F9 (bioturbated glauconitic sandstone) overlies Mauddud sequence boundary MAU400_SB in southwestern Kuwait (Minagish field area; Figure 23). This sequence is dominated by lithofacies F4 (bioturbated skeletal-peloidal packstone) in southern Kuwait (Greater Burgan field area; Figure 23) and dominated by marginal-marine siliciclastics with thin, karstified lithofacies F3 (skeletal-peloidal mudlean packstone) at the top in southwestern Kuwait (Minagish field area; Figure 23). It is partly eroded in southwestern Kuwait (Minagish field area; Figure 23). The sequence is composed predominantly of lithofacies F2 (skeletal-peloidal grainstone) and lithofacies F3 (skeletal-peloidal mud-lean packstone), as well as minor lithofacies F4 (bioturbated skeletalpeloidal packstone) and lithofacies F7 (clay-rich, bioturbated skeletal wackestone) in the northern fields area (Figures 22, 23).

Mauddud Sequence MAU350 (MAU350_SB to MAU300_SB) This sequence is composed predominantly of lithofacies F2 (skeletal-peloidal grainstone) and lithofacies F3 (skeletal-peloidal mud-lean packstone), as well as minor lithofacies F4 (bioturbated skeletalpeloidal packstone) and intercalations of lithofacies F1 (rudist floatstone-rudstone) in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22). It is partly eroded in the Greater Burgan field area (lithofacies F4: skeletal-peloidal packstone; Figure 23) and eroded (not present) in southwestern Kuwait (Minagish field area; Figure 23). This sequence may show an upward increase of porosity and permeability and shallowing-upward lithofacies trend as well as coarsening-upward texture in the northern fields area (Figures 21, 22).

Mauddud Sequence MAU300 (MAU300_SB to MAU200_SB) This sequence is composed predominantly of lithofacies F2 (skeletal-peloidal grainstone) and lithofacies F3 (skeletal-peloidal mud-lean packstone), as well as minor lithofacies F4 (bioturbated skeletalpeloidal packstone) and lithofacies F1 (rudist floatstone-rudstone) in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22). The sequence is eroded (not present) in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23).

242 / Strohmenger et al.

Mauddud Sequence MAU200 (MAU200_SB to MAU100_SB) This sequence is composed predominantly of lithofacies F5 (nodular, miliolid-bearing packstonewackestone) and lithofacies F3 (skeletal-peloidal mud-lean packstone), as well as minor lithofacies F4 (bioturbated skeletal-peloidal packstone) and lithofacies F1 (rudist floatstone-rudstone) in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22). The sequence is eroded (not present) in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23).

Mauddud Interval MAU100_SB to MAU100_TS This interval is composed predominantly of lithofacies F11 (gray shale and mudstones) and lithofacies F7 (clay-rich, bioturbated skeletal wackestone). The interval was deposited during the third-order relative sea-level rise above Mauddud sequence boundary MAU100_SB (Figures 22, 23) and is not present in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23).

Composite Sequence The TSS of the Mauddud composite sequence starts at the base of Mauddud transgressive surface B100_TS and is bounded on the top by Mauddud maximum flooding surface MAU400_MFS of the Mauddud highfrequency sequence MAU400. The overlying HSS is bounded on top by Mauddud sequence boundary MAU100_SB. The top of the Mauddud Formation is marked by the transgressive surface MAU100_TS (Figures 22, 23) that merges with the Mauddud sequence boundary MAU100_SB toward southern and southwestern Kuwait (Figure 23). The Mauddud TSS shows a lateral change in lithology from limestone in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22) to siliciclastics in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23). An overall shoaling-upward or progradational signature characterizes the Mauddud HSS. The HSS is interpreted to be eroded down to and below sequence boundary MAU350_SB in southern (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23). Our sequence-stratigraphic interpretation of the Mauddud Formation contrasts with the previously used lithostratigraphic correlation. Lithostratigraphic correlation places the boundary between the Mauddud Formation and the underlying Burgan Formation at the base of the deepest limestone bed. In con-

trast, the sequence-stratigraphic boundary between the Mauddud Formation and the underlying Burgan Formation occurs approximately 100 ft (30 m) deeper in the clastic section of the Mauddud Formation. The lithostratigraphic boundary between the Mauddud carbonate member and the Mauddud clastic member is highly diachronous (Figure 23). The thickness relationships of the two Mauddud Members are reciprocal. The Mauddud carbonate member becomes very thin in locations where the Mauddud clastic member is thick. Toward the north (Raudhatain and Sabiriyah fields area), the Mauddud clastic member becomes very thin to absent, whereas the Mauddud carbonate member is thinning toward southern Kuwait (Greater Burgan field area; Figure 23) and southwestern Kuwait (Minagish field area; Figure 23). The sequence-stratigraphic correlation of the Upper Burgan and the Mauddud presented here is supported by biostratigraphic and palynofacies data (T. D. Davies and T.-C. Huang, 2000, personal communication) (Figure 23).

Reservoir Quality Distribution In general, most Mauddud lithofacies have moderate porosity and overall low permeability. Lithofacies types F2, F3, F4, F5, and F6 have essentially the same porosity-permeability distributions. Only lithofacies type F1 (rudist floatstone-rudstone) and lithofacies F10 (bioturbated, mud-rich sandstone) show higher porosity and/or permeability values. Intervals that show enhanced reservoir quality are related to fracturing and faulting (enhanced permeability), the occurrence of microkarst (Figure 18C), the occurrence of rudist floatstone-rudstone (lithofacies F1: Mauddud high-frequency sequence MAU450; Figure 17A), or the occurrence of the bioturbated, mud-rich sandstone (lithofacies F10: Mauddud high-frequency sequence MAU500; Figure 20B).

CONCLUSIONS Sequence-stratigraphic and biostratigraphic analyses indicate that the traditional lithostratigraphic Burgan-Mauddud contact is actually a time-transgressive boundary (Figure 23). The uppermost Burgan and overlying Mauddud formations form a third-order composite sequence that becomes more siliciclastic prone toward the south (in the Greater Burgan field area) and toward the southwest (in the Minagish field area; Figure 23) and carbonate prone to the north and northeast of Kuwait near the Raudhatain and Sabiriyah fields (Figures 22, 23). A lower, third-order

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 243

composite sequence encompasses the remainder of the underlying Burgan Formation (Figure 12). It comprises marginal-marine deposits to the northeast and nonmarine deposits to the southwest of Kuwait (Figure 12). Within the Burgan Formation, 19 high-frequency, depositional sequences are recognized and stack to form one third-order composite sequence and the LSS of a younger third-order composite sequence (Figures 3, 12). Lowstand systems tracts consist primarily of fluvial, incised-valley-fill deposits in southern and western Kuwait that become thinner, laterally more discontinuous, and more tidal-influenced in a downdip direction to the northeast (Figure 13). The TSTs display a systematic updip to downdip change from mud-prone alluvial- and coastal-plain strata to marginal-marine mudstones and sandstones to marine carbonates. The HSTs are dominated by updip alluvial- and coastal-plain mudstones and sandstones and marginal-marine sandstones and mudstones downdip (Figure 14). Facies analysis of the Burgan Formation reveals the presence of 22 lithofacies, which are delineated based on unique physical criteria, including grain size, composition, sorting, grading, and physical and biogenic sedimentary structures. These lithofacies describe six distinct facies, which form the basic building blocks of two facies associations (Figure 4). The primary reservoirs in the Burgan Formation are fluvial and tidal deposits of the incised-valley-fill facies association (Figure 4). Minor potential reservoirs are shoreline sandstones of the TSTs and HSTs (Figure 4). In addition, the possibility exists that marginal-marine shorelines in the TSSs may contain hydrocarbons off the flanks of present-day structures. This combination of structural and stratigraphic trap would be dependent on mud-prone coastal-plain sediments in each sequence acting as the updip lateral seal and marine shales in the overlying sequence serving as a (very high-risk) top seal. The Mauddud Formation is sequence stratigraphically subdivided into eight high-frequency depositional sequences (Figures 3, 22, 23). The identified high-frequency SBs, TSs, MFSs, and FS can be correlated regionally throughout Kuwait. The Mauddud transgressive sequence set displays a lateral change in lithology from limestone in northern Kuwait (Raudhatain and Sabiriyah fields area; Figure 22) to siliciclastics in southern (Greater Burgan field area) and southwestern Kuwait (Minagish field area; Figures 15, 23), interfingering with what has traditionally been recognized as the Burgan Formation (Figures 3, 23).

The Mauddud highstand sequence set is carbonate prone and is mostly eroded in southern and southwestern Kuwait, resulting in a significant southward and southwestward thinning (Figures 16, 23). A total of eight carbonate lithofacies and three siliciclastic lithofacies types were identified within the Mauddud Formation. Carbonate lithofacies were deposited in inner- to lower-ramp, normal- to slightly restricted-marine environments. Siliclastic lithofacies were deposited in a fairly wide range of environments, including deeper lagoon, nearshore-marine, and offshore-marine (Mauddud carbonate member), as well as coastal-plain, fluvial, and tidal, (Mauddud clastic member) environments. In general, most Mauddud lithofacies have moderate porosity and low permeability, with microporosity as the dominant pore type. Intervals with enhanced reservoir quality can be related to fracturing and faulting, the occurrence of microkarst (Figure 18C), the presence of rudist floatstone-rudstone (Figure 17A), and the distribution of intercalated mud-rich sandstone (Figure 18A). Within the context of the new Mauddud – Burgan chronostratigraphic boundary, distinct isolated incised-valley systems can be defined within the Mauddud clastic member. These IVF are expected to have reservoir characteristics similar to those of the underlying Burgan Formation. The proposed sequence-stratigraphic framework and the sequence-stratigraphy-keyed facies scheme result in a predictable distribution of reservoir and seal facies and allow for a better prediction of the vertical and lateral distribution of reservoir quality and reservoir continuity at both field scale and regional scale.

ACKNOWLEDGMENTS The authors gratefully acknowledge the managements of ExxonMobil Exploration Company, ExxonMobil Upstream Research Company, and Kuwait Oil Company (KOC) for their permission to publish this paper. For valuable discussions, we thank Linda W. Corwin (ExxonMobil), Daniel H. Cassiani (ExxonMobil), Kathleen M. McManus (ExxonMobil), Menahi Al-Anzi (KOC), Ahmed Al-Eidan (KOC), and Mohammed Al-Ajmi (KOC). We extend our thanks to David Awwiller (ExxonMobil) for his detailed reservoirquality analyses and to Tom D. Davies and TingChang Huang for performing the biostratigraphic and palynofacies analyses. Dolores A. Claxton (ExxonMobil) is thanked for drafting the figures. We extend special thanks to Arthur D. Donovan (BP; formerly

244 / Strohmenger et al.

with ExxonMobil) for his valuable contributions to this study. The authors greatly appreciate the thorough and thoughtful reviews of J. R. (Rick) Sarg and James McGovney.

REFERENCES CITED Adasani, M., 1965, The Greater Burgan field: Fifth Arab Petroleum Congress, Cairo, p. 7 – 27. Adasani, M., 1967, The northern Kuwait oil fields: Sixth Arab Petroleum Congress, Baghdad, p. 1 – 39. Al-Anzi, M., 1995, Stratigraphy and structure of the Bahra field Kuwait, in M. I. Al-Husseini, ed., The Middle East petroleum geosciences (Geo94): Selected Middle East papers from the Middle East Geoscience Conference: Bahrain, Gulf PetroLink, v. 1, p. 53 – 64. Al-Eidan, A. J., W. B. Wethington, and R. B. Davies, 2001, Upper Burgan reservoir distribution, northern Kuwait: Impact on reservoir development: GeoArabia, v. 6, no. 2, p. 179 – 208. Al-Rawi, M. M., 1981, Geological interpretation of the oil entrapment in the Zubair Formation, Raudhatain field: Society of Petroleum Engineers, Middle East Oil Technical Conference, Bahrain, SPE Paper 9591, p. 149–159. Alsharhan, A. S., and A. E. M. Nairn, 1993, Carbonate platform models of Arabian Cretaceous reservoirs, in J. A. T. Simo, R. W. Scott, and J.-P. Masse, eds., Cretaceous carbonate platforms: AAPG Memoir 56, p. 173 – 184. Alsharhan, A. S., and A. E. M. Nairn, 1997, Sedimentary basins and petroleum geology of the Middle East: Amsterdam, Elsevier, 843 p. Bou-Rabee, F., 1996, Geologic and tectonic history of Kuwait as inferred from seismic data: Journal of Petroleum Science and Engineering, v. 16, p. 151 – 167. Brennan, P., 1990a, Raudhatain field, Kuwait, Arabian basin, in E. A. Beaumont and N. H. Foster, eds., AAPG Treatise of Petroleum Geology, Atlas of Oil and Gas Fields: Structural Traps 1, p. 187 – 210. Brennan, P., 1990b, Greater Burgan field, Kuwait, Arabian basin, in E. A. Beaumont and N. H. Foster, eds., AAPG Treatise of Petroleum Geology, Atlas of Oil and Gas Fields: Structural Traps 1, p. 103 – 128. Carman, G. J., 1996, Structural elements of onshore Kuwait: GeoArabia, v. 1, no. 2, p. 239 – 266. Christian, L., 1997, Cretaceous subsurface geology of the Middle East region: GeoArabia, v. 2, no. 3, p. 239 – 256. Demko, T. M., P. E. Patterson, H. R. Feldman, C. J. Strohmenger, J. C. Mitchell, P. J. Lehmann, G. Alsahlan, H. Al-Enezi, and M. Al-Anezi, 2003, Sequence stratigraphy and reservoir architecture of the Burgan and Mauddud formations (Lower Cretaceous), Kuwait (abs.): AAPG Annual Meeting Program, v. 5, p. A39. Fox, A. F., 1961, The development of the southeastern Kuwait oil fields: Institute of Petroleum Review, v. 15, no. 180, p. 373 – 379. Haq, B. U., 1991, Sequence stratigraphy, sea-level change, and significance for the deep sea, in D. I. M. Mac-

donald, ed., Sedimentation, tectonics and eustasy: Sealevel changes at active margins: International Association of Sedimentologists Special Publication 12, p. 3 – 39. Haq, B. U., J. Hardenbol, and P. R. Vail, 1987, Chronology of fluctuating sea levels since the Triassic: Science, v. 235, p. 1156 – 1167. Haq, B. U., J. Hardenbol, and P. R. Vail, 1988, Mesozoic and Cenozoic chronostratigraphy and cycles of sealevel change, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 71 – 108. Kaufman, R. L., H. Dashti, C. S. Kabir, J. M. Pederson, M. S. Moon, R. Quttainah, and H. Al-Wael, 1997, Characterizing the Greater Burgan field: Use of geochemistry and oil fingerprinting: Society of Petroleum Engineers 10th SPE Middle East Oil Show and Conference, Bahrain, SPE Paper 37803, p. 385 – 394. Milton, D. I., and C. C. S. Davies, 1965, Exploration and development of the Raudhatain field: Journal of the Institute of Petroleum, v. 51, no. 493, p. 17 – 28. Mitchum Jr., R. M., 1977, Seismic stratigraphy and global changes in sea level: Part 11 — Glossary of terms used in seismic stratigraphy, in C. E. Payton, ed., Seismic stratigraphy — Applications to hydrocarbon exploration: AAPG Memoir 26, p. 205 – 212. Mitchum Jr., R. M., and J. C. Van Wagoner, 1991, Highfrequency sequences and their stacking patterns: Sequence-stratigraphic evidence of high-frequency eustatic cycles, in K. T. Biddle and W. Schlager, eds., The record of sea-level fluctuations: Sedimentary Geology, v. 70, p. 131 – 160. Mitchum Jr., R. M., P. R. Vail, and S. Thompson III, 1977, Seismic stratigraphy and global changes of sea level: Part 2 — The depositional sequence as a basic unit for stratigraphic analysis, in C. E. Payton, ed., Seismic stratigraphy — Applications to hydrocarbon exploration: AAPG Memoir 26, p. 53 – 62. Pemberton, G. S., J. A. MacEachern, and R. W. Frey, 1992a, Trace fossil models; environmental and allostratigraphic significance, in R. G. Walker and J. P. Noel, eds., Facies models; response to sea level change: Geological Association of Canada, p. 47 – 72. Pemberton, G. S., J. C. Van Wagoner, and G. D. Wach, 1992b, Ichnofacies of a wave-dominated shoreline: SEPM Core Workshop 17, p. 339 – 382. Sarg, J. F., 1988, Carbonate sequence stratigraphy, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 155 – 181. Sarg, J. F., J. R. Markello, and L. J. Weber, 1999, The secondorder cycle, carbonate-platform growth, and reservoir, source, and trap prediction, in P. M. Harris, A. H. Saller, and J. A. T. Simo, eds., Advances in carbonate sequence stratigraphy: Application to reservoirs, outcrops and models: SEPM Special Publication 63, p. 11 – 34.

Sequence Stratigraphy and Reservoir Architecture of Burgan and Mauddud Formations / 245 Strohmenger, C. J., T. M. Demko, J. C. Mitchell, P. J. Lehmann, H. R. Feldman, A. Douban, A. J. Al-Eidan, G. Alsahlan, and H. Al-Enezi, 2002, Regional sequence stratigraphic framework for the Burgan and Mauddud formations (Lower Cretaceous, Kuwait): Implications for reservoir distribution and quality (abs.): GeoArabia, v. 7 , no. 2, p. 304. Vail, P. R., 1987, Seismic stratigraphy interpretation using sequence stratigraphy: Part 1 — Seismic stratigraphy interpretation procedure, in A. W. Bally, ed., Atlas of seismic stratigraphy, v. 1: AAPG Studies in Geology 27, p. 1 – 10. Vail, P. R., R. M. Mitchum Jr., R. G. Todd, J. M. Widmier, S. Thompson III, J. B. Sangree, J. N. Bubb, and W. G. Hatlelid, 1977, Seismic stratigraphy and global changes of sea level: Part 1— Overview, in C. E. Payton, ed., Seismic stratigraphy— Applications to hydrocarbon exploration: AAPG Memoir 26, p. 49– 212. Vail, P. R., F. Audemard, S. A. Bowman, P. N. Eisner, and C. Perez-Cruz, 1991, The stratigraphic signatures of tectonics, eustasy and sedimentology — An overview, in G. Einsele, W. Ricken, and A. Seilacher, eds., Cycles and events in stratigraphy: Berlin, Springer, p. 617 – 659.

Van Wagoner, J. C., R. M. Mitchum Jr., H. W. Posamentier, and P. R. Vail, 1987, Seismic stratigraphy interpretation using sequence stratigraphy: Part 2 — Key definitions of sequence stratigraphy, in A. W. Bally, ed., Atlas of seismic stratigraphy, v. 1: AAPG Studies in Geology 27, p. 11–14. Van Wagoner, J. C., H. W. Posamentier, R. M. Mitchum Jr., P. R. Vail, J. F. Sarg, T. S. Loutit, and J. Hardenbol, 1988, An overview of the fundamentals of sequence stratigraphy and key definitions, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea-level changes: An integrated approach: SEPM Special Publication 42, p. 39 – 46. Van Wagoner, J. C., R. M. Mitchum, K. M. Campion, and V. D. Rahmanian, 1990, Siliciclastic sequence stratigraphy in well logs, cores, and outcrop: AAPG Methods in Exploration Series 7, 55 p. Walker, R. G., and A. G. Plint, 1992, Wave- and stormdominated shallow marine systems, in R. G. Walker and N. P. James, eds., Facies models — Response to sea-level change: Geological Association of Canada, p. 219 – 238.

7

Dull, D. W., R. A. Garber, and W. S. Meddaugh, 2006, The sequence stratigraphy of the Maastrichtian (Upper Cretaceous) reservoir at Wafra field, Partitioned Neutral Zone, Saudi Arabia and Kuwait: Key to reservoir modeling and assessment, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 247 –279.

The Sequence Stratigraphy of the Maastrichtian (Upper Cretaceous) Reservoir at Wafra Field, Partitioned Neutral Zone, Saudi Arabia and Kuwait: Key to Reservoir Modeling and Assessment Dennis W. Dull Chevron Energy Technology Company, Houston, Texas, U.S.A.

Raymond A. Garber Chevron Energy Technology Company, Houston, Texas, U.S.A.

W. Scott Meddaugh Chevron Energy Technology Company, Houston, Texas, U.S.A.

ABSTRACT

T

he Maastrichtian (Upper Cretaceous) reservoir is one of five prolific oil reservoirs in the giant Wafra oil field. Although discovered and first produced in 1959, the reservoir is currently in early development because of its low but variable oil gravity, high sulfur content, relatively high water cut, and apparent compartmentalization. This made it a much less attractive resource than other productive intervals at Wafra field. Less than 1% of the original oil in place in the Maastrichtian has been produced. The Maastrichtian oil production is largely from subtidal dolomites at an average depth of 760 m (2500 ft). Carbonate deposition occurred on a very gently dipping, shallow, arid, and restricted ramp setting that transitioned between normal-marine conditions to restricted lagoonal environments. The average porosity of the reservoir interval is about 15%, although productive zones have porosity values as much as 30–45%. The average permeability of the reservoir interval is about 30 md; individual core plugs have measured permeability as much Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215879M883030

247

248 / Dull et al.

as 1200 md. This study was undertaken to (1) determine reservoir volumetrics, (2) understand the areal and stratigraphic distribution of intervals likely to yield higher volumes of better quality oil, and (3) provide a reservoir property model for use in fluid-flow simulation. The key to modeling the reservoir was the construction of an appropriately detailed sequence-stratigraphic framework for use in building a geostatistical reservoir model with high-quality descriptions from five cored wells in the reservoir. The geostatistical model of the Maastrichtian reservoir demonstrates the layered and compartmentalized nature of the reservoir and clearly shows that the location of the reservoir facies in the Maastrichtian is controlled by the original depositional fabric and subsequent dolomitization, both of which have been influenced by the paleotopography. Such understanding is critical to efficiently develop the 1.5 billion bbl of Maastrichtian oil at Wafra field.

INTRODUCTION The Wafra field is located in the Partitioned Neutral Zone (PNZ) between Kuwait and Saudi Arabia. Figure 1 shows the location of the major oil fields in the PNZ. Figure 2 provides a generalized stratigraphic column for the PNZ showing the five producing horizons at Wafra field. Oil was discovered in the Upper Cretaceous Maastrichtian in the PNZ at the South Fuwaris field lo-

cated south and west of Wafra field in March 1959. The first Maastrichtian production at Wafra field occurred in November 1959. The field has produced about 20 million bbl of oil primarily from two wells. Figure 3 provides a structure map for the Maastrichtian showing the location of the Maastrichtian producing wells, five wells with core data used to build the sequencestratigraphic interpretation, and the 123 wells used to generate the geostatistical reservoir model.

FIGURE 1. Location of the Maastrichtian reservoir, Wafra field, Partitioned Neutral Zone (PNZ). S = south. Also shown are the other producing fields, S. Umm Gudair, S. Fuwaris, and Humma in the PNZ.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 249 





the integration of the cores with the well logs for extrapolating the highfrequency sequence (HFS) correlations beyond the cores in constructing the reservoir model the impact of the data integration in a threedimensional (3-D) Earth model to provide insight into possible varying oilwater contacts and reservoir compartmentalization the importance of sequence stratigraphy and 3-D modeling in identifying the reservoir potential of the Wafra field Maastrichtian reservoir

REGIONAL STRATIGRAPHY The Maastrichtian interval in the Wafra field consists of the Upper Cretaceous Tayarat Formation, which is the uppermost part of the Aruma Group and was deposited on a stable platform (Figure 4). The geological setting for the Maastrichtian carbonates of the Arabian shield includes (1) the AraboNubian shield, situated in Saudi Arabia, covering several hundreds of kilometers in width; (2) a zone of clastic sediments that rim the eastern margin of the shield conFIGURE 2. Stratigraphic column for the PNZ showing the five producing reservoirs of taining both continental and Wafra field. shallow-marine siliciclastics, sourced from the shield; and (3) a carbonate shelf lying seaward of the clastic belt, The objectives of this chapter are to demonstrate which was the site of shallow shelf carbonate sedi mentation (Harris et al., 1984). This shelf contains the importance of core in defining the sequenceintrashelf basins, which were filled with shales and stratigraphic framework and depositional deep-water limestones. environment  The Upper Cretaceous sequence was deposited durthe methodology used to define the sequenceing a period of mild tectonic activity and is variable in stratigraphic framework

250 / Dull et al.

FIGURE 3. Structure map on top of high-frequency sequence (HFS) M00 (first Maastrichtian shale) constructed from 3-D seismic data. The contour interval is 50 ft (15 m). The map shows production from 35 wells at an average depth of 2700 ft (822 m). Map shows well locations used in the modeling, Maastrichtian producing wells, and cored well locations.

regional thickness and facies. Most of the tectonism affected areas to the east of the PNZ, particularly in the northern Oman Mountains (Harris et al., 1984). A regional unconformity is observed at the top of the Cretaceous (upper Maastrichtian), and moldic porosity is commonly developed at this contact with the overlying Tertiary. In some places, the upper Maastrichtian is characterized by large collapse breccias formed by

evaporite solution (Harris et al., 1984). A lithostratigraphic correlation chart is found in Figure 5. In outcrop, the Aruma is treated as a formation and is divided into three members from base to top: Khanasir Limestone Member, Hajajah Limestone Member, and Lina Shale Member (El-Nakhal and ElNaggar, 1994; Philip et al., 2002). The lower two members are believed to be of Maastrichtian age, whereas

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 251

FIGURE 4. Map of the Arabian Peninsula showing location of the Partitioned Neutral Zone (PNZ) in relation to major structural features.

the uppermost member is believed to be of Paleocene age based on biostratigraphic analysis. In the subsurface, the Aruma is given group rank, and in Kuwait, it is divided into three formations (Gudair, Bahrah, and Tayarat) (El-Nakhal and El-Naggar, 1994). The Aruma increases in thickness from the outcrop areas in the west into the subsurface at the edges of the Persian

Gulf (El-Nakhal and El-Naggar, 1994) (Figure 6). In wells in the Wafra field, the Tayarat Formation is about 400 m (1300 ft) thick, whereas in outcrop, the Tayarat equivalent is about 160 m (524 ft) thick. The Aruma Formation unconformably overlies sandstones of the Wasia Formation and is overlain by dense argillaceous limestone and thin pyritic shales of the Umm er Radhuma Formation. As the Aruma succession thickens from outcrop to the subsurface to the east around the Persian Gulf, the type locality is not complete (Alsharhan and Nairn, 1990). In addition, different formation names have been used, and it has been subdivided differently in different places. Consequently, the Aruma is typically referred to as a group and may include anywhere from two to six formations (Alsharhan and Nairn, 1990). As noted earlier, the Maastrichtian interval in the Wafra

FIGURE 5. Lithostratigraphic correlation of the Upper Cretaceous in the Arabian Peninsula (Alsharhan and Nairn, 1990).

252 / Dull et al.

FIGURE 6. Schematic cross section showing Aruma surface to subsurface relationships (El-Nakhal and El-Naggar, 1994). field in the PNZ is found within the Tayarat Formation, a part of the Aruma Group. In Abu Dhabi, the Shah field produces minor oil and gas-condensate from the Upper Cretaceous Aruma Group, which consists of Coniacian to Maastrichtian lithologies (Alsharhan and Kendall, 1995). There, the Aruma Group consists of basal Laffan Formation (Coniacian) containing mainly slightly calcareous shales, which is overlain by the Halul Formation (Santonian), which consists of an argillaceous limestone. This is overlain by the Fiqa Formation (Campanian), which consists of alternating calcareous shales, marls, and argillaceous limestone. These are capped by the Maastrichtian Simsima Formation containing fossiliferous and shaly limestone and dolomitic limestone rimming the Omani foredeep (Alsharhan and Kendall, 1995). The Simsima Formation, equivalent to the Aruma Formation in Oman, is interpreted to have been deposited on a shallow carbonate platform. In the Shah field, it consists of two large-scale depositional cycles. The lower cycle consists of a coarseningupward wackestone-mudstone and rudistid packstonegrainstone, with finely crystalline dolomite and minor anhydrite deposited in intertidal to supratidal environments. The upper unit is composed of a vertical succession of fining-upward sequences containing packstone, wackestone, shale, and very finely crystalline dolomite deposited in a subtidal to intertidal setting. It becomes more restricted at the top. The subsurface Simsima commonly contains interparticle, moldic, and vuggy porosity with locally good

intercrystalline porosity in the dolomites (Alsharhan and Kendall, 1995). During the Maastrichtian, the Arabian Peninsula was located about 128 north of the equator on a relatively stable continental margin on the African plate as it moved north toward the Asian plate (Golonka et al., 1994; Morris et al., 2002). The Neotethys foredeep narrowed during the Late Cretaceous, and the paleogeographic reconstruction by Golonka et al. (1994) shows the development of a land mass in the northeast in part of the present-day Arabian Peninsula caused by the continued northward drift of the Arabian plate (Sharland et al., 2001; Morris et al., 2002). The work of Sayed and Mersal (1998) in Jebel Rawdah in the northern Oman Mountains (a westnorthwest- to east-southeast-trending, postobduction fold) shows that the Oman Mountains were emergent in the early Maastrichtian. Smith et al. (1995) placed the ophiolite in the Oman Mountains as latest Campanian to Maastrichtian, whereas later work by Schreurs and Immenhauser (1999) showed that the ophiolite was deposited in late Maastrichtian to Paleocene. The proximity of a landmass to the northeast of the PNZ provides a likely source of the shale in the second Maastrichtian interval. Alsharhan and Nasir (1996), in their research in the western Oman Mountains, showed that the Late Cretaceous Qalah formation, which uncomformably onlaps the Simsima Formation, indicates that the obducted Semail ophiolite was extensively weathered in a tropical environment. Skelton et al. (1990), in their study of the Qalah and Simsima Formations, suggest the Semail

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 253

ophiolite was emergent, and that a more open shelf environment exists. The absence of evaporites in the second Maastrichtian interval is consistent with deposition on a shallow, gently dipping ramp under humid conditions. The obducted complex was strongly emergent at the start of the Maastrichtian, the Aruma Sea transgressed onto its western flank, and the ophiolite subsided through that stage, forming a relatively open shelf in the north, becoming broader and more restricted farther south (Skelton et al., 1990). The continued narrowing of the Neotethys could also possibly explain the restriction in the marine circulation that resulted in the lagoonal sedimentation during sea level lowstands in the upper Maastrichtian interval. Overall, the abundant evidence of subaerial exposure, thin beds, dolomitization, and anhydrite suggests that deposition during the first Maastrichtian occurred on a very gently dipping, shallow, arid, and restricted ramp setting that transitioned between normal-marine conditions to restricted lagoonal environments.

RESERVOIR DESCRIPTION The Maastrichtian reservoir is composed mainly of dolomite and limestone with thin beds of organicrich shale. Individual shale layers ranges from less than 2 cm to 1 m (0.78 in. to 3.3 ft) or more. The second Maastrichtian shale separates the Maastrichtian into an upper and lower portion with significantly different depositional characteristics. The type log given in Figure 7 shows the vertical extent of the first and second Maastrichtian intervals. The upper part of the Maastrichtian is also known as the first Maastrichtian reservoir. The lower portion is commonly referred to as the second Maastrichtian reservoir. This chapter will use upper and lower Maastrichtian, instead of first and second Maastrichtian because second Maastrichtian shale does not extend across the entire modeled area. The lower Maastrichtian is composed largely of dolomite with local limestone and shale, whereas the upper Maastrichtian is composed largely of limestone with locally abundant dolomite and shale. The shale in the upper Maastrichtian is organic rich, with total organic content as much as nearly 50% of the rock volume and it is largely nonmarine. The shale in the lower Maastrichtian contains very little organic matter, less than 1% of the rock volume, and is largely marine in origin, which is supported by the palyno-

logical interpretation as summarized in Table 1. Anhydrite occurs throughout the reservoir interval as nodules (Figure 8), void and fracture filling (Figures 9, 10), and, occasionally, as thick intervals of coalesced nodules (Figure 11). Oil production from the upper Maastrichtian reservoir production occurs in the dolomitized, muddominated peloidal packstones, skeletal-peloidal grain-dominated packstones, and rudist rudstones. The oil production from the lower Maastrichtian interval occurs in dolomitized peloidal grainstones and packstones. The field has produced less than 1% of the original oil in place (OOIP) based on an estimate of nearly 1.5 billion bbl obtained from the reservoir model completed as part of this study. Production is highly variable; initial rates range between several hundred barrels of oil per day (BOPD) to more than 3000 BOPD. Some wells are exceptional producers; one well has produced more than 5 million bbl of oil and some water out very quickly. Although representing a very significant oil resource, the Maastrichtian reservoir provides several challenges that this study addresses, including (1) the presence of multiple barriers and baffles that compartmentalize the reservoir areally and vertically; (2) the areal variation of oil gravity (13 –218 API); (3) the absence of a well-defined oilwater contact; (4) a water zone in the uppermost part of the reservoir interval; and (5) the distribution of possible fractures and faults that may provide an explanation for why some wells water out very quickly whereas other wells produce at very low water cuts for many years. Prior published reports on the Maastrichtian reservoir at Wafra field are very limited. Brief mention is made by Nelson (1968) and Alsharhan and Nairn (1997). Danielli (1988) provides some information about the shallow reservoirs at Wafra field, including the Maastrichtian.

SEQUENCE-STRATIGRAPHIC FRAMEWORK OF THE MAASTRICHTIAN The Maastrichtian reservoir represents a shallowing-upward, lower order sequence culminating in a regionally correlative subaerial exposure surface that marks the top of the Maastrichtian and is interpreted to be equivalent to the top of tectonostratigraphic megasequence (TMS) AP9 of Sharland et al. (2001). Abdel-Kireem et al. (1994) described an unconformity at base of the Paleocene and the top of the late

254 / Dull et al.

FIGURE 7. Type log cored well 3 for the Maastrichtian showing the reservoir interval, high-frequency sequencestratigraphic picks, and lithologic core description. The well-log curves are track 1 (CGR = corrected gamma ray; SGR = spectral gamma ray; VCLGR = clay volume from gamma ray); track 2 (core porosity; PHIE = effective porosity); track 3 (PHIE; BVO = bulk volume oil; BVW = bulk volume water); and track 4 (facies). Depth in feet.

Maastrichtian in northeastern Iraq. The TMS AP9 marks the end of the Cretaceous ophiolite obduction on the northern Arabian plate, resulting in widespread regression (Beydoun, 1991, 1993). This exposure sur-

face is overlain by a marine incursion that is the base of the Paleocene-age Umm Er Radhuma Formation. A stratigraphic cross section through the five cored wells showing the identified evidence for subaerial

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 255

Table 1. Interpreted environments of deposition from the shale palynology study of the Maastrichtian. Well

Depth (ft)

Interpretation

HFS

5 5 4 3 5 4 3 4 3 5 4

2567 2619 2743 2784 2780 2796 2854 2851 2932 2851 2885

Marine No palynomorphs Possible marine Marsh and embayment No palynomorphs Marine Restricted lagoon and fresh water Possible marine Distal lagoon, marine influence Marine Probable marine

M00 M00 M10 M20 M20 M20 M30 M40 M50 M50 M50

exposure as hardgrounds and karst, interpreted environments of deposition from the palynology, and interpreted HFSs that were used to define the sequencestratigraphic framework of the Maastrichtian is given in Figure 12. The Maastrichtian is an overall regressive sequence composed of a transgressive systems tract (TST) and a highstand systems tract (HST). Figures 8–11 and 13–31 show the core slab photos for the HFS M10 to M80 interval for cored well 3. The location of the cored wells and cross section location are shown on the map of the field in Figure 3. The establishment of an appropriately detailed sequence-stratigraphic framework provides the foundation for the reservoir models as the stratal relationships, and the genetically related sequences largely control the spatial distribution of the facies and petrophysical properties of the reservoir (Kerans and Tinker, 1997; Deutsch, 2002). The basic framework of the Maastrichtian is the HFS. The HFS, as defined by Kerans and Tinker (1997), is bounded by a baselevel-fall to base-level-rise turnaround and is composed of genetically related cycles and cycle sets. A cycle is defined as the smallest set of genetically related lithofacies representing a single base-level rise and fall. This is comparable to the parasequence of Van Wagoner et al. (1988, 1990). The cycle set is a bundle of cycles that show a consistent trend of aggradation or progradation (Kerans and Tinker, 1997). The HFS was chosen as the basic building block for correlation in the Maastrichtian because it could be correlated with a great deal of confidence. Individual cycle correlation in the Maastrichtian is hampered by large distances between wells, abundance of exposure surfaces, and thin beds related to low accommo-

dation space, making it uncertain if a facies change was caused by eustatic sea level changes or topography. The sequence-stratigraphic modeling for the Maastrichtian was based on 413 m (1356 ft) of core from five wells and well logs from 123 wells. The core description and the correlation of the five Maastrichtian cores have revealed what is interpreted as nine HFSs of an overall regressive sequence. Figure 7 is the type log for the Maastrichtian showing nine of the HFSs and the lithofacies from the core description. With only five cored wells, it was essential to use the well logs from the other wells to construct the 3-D stratigraphic framework for the full Maastrichtian reservoir interval. As documented below, a synthetic lithofacies curve was generated from the petrophysical data to assist in the fieldwide stratigraphic interpretation and correlation. The HFS for the Maastrichtian was identified using established bounding surface criteria that indicate a base-level-fall to base-level-rise turnaround. The bounding surfaces of the HFS were identified most commonly by subaerial exposure surfaces or unconformities expressed as hardgrounds that show burrow and root casts filled with anhydrite commonly capped by organic terrestrial lagoonal shale (Figure 32). Less commonly, the bounding surfaces were identified by karst breccia with solution-enhanced open fractures and void-filling anhydrite occasionally associated with subaerial hardgrounds (Figure 33). Uncommonly, the HFS were identified by (1) tidal flat caps of dolomudstone and dolowackestone, with pseudomorphs of anhydrite after gypsum (Figure 34), (2) lithofacies tract offset from high-energy peloidal dolograinstones to subtidal-marine dolowackestones and mud-rich dolopackstones (Figure 35), or (3) rudist and skeletal dolorudstone with tidal-flat caps (Figure 36).

Fieldwide Correlation The fieldwide sequence-stratigraphic correlation of the 10 HFSs of the Maastrichtian identified in the cored wells was extended to all the wells based on a synthetic facies curve (FACIESG) and clay volume from gamma-ray curve (VCLGR). Both curves were used to assist in the fieldwide correlation. The correlation between well-log response and reservoir quality was essential to predicting facies on noncored wells with well logs. A review of the core material, measured porosity, and measured permeability indicated seven lithofacies or rock types. In the course of core review, it became necessary to distinguish the mud-lean dolopackstones from the mudrich dolopackstones and the dolorudstone from the

256 / Dull et al.

FIGURE 8. Core slab photos 2905 – 2920 ft (855 – 890 m) for cored well 3.

FIGURE 9. Core slab photos 2887 – 2905 ft (879 – 855 m) for cored well 3.

FIGURE 10. Core slab photos 2938 – 2956 ft (895 – 900 m) for cored well 3.

FIGURE 11. Core slab photos 2869 – 2887 ft (874 – 879 m) for cored well 3.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 257

FIGURE 12. Stratigraphic cross section through the five cored wells showing the identified evidence for subaerial exposure, interpreted environments of deposition from the palynology, and interpreted HFSs that were used to define the sequence-stratigraphic framework of the Maastrichtian. The Maastrichtian is an overall regressive sequence composed of a transgressive system tract and a highstand system tract. Depth in feet.

dolofloatstones. In addition, the cemented hardground or dolostone and dolomudstone were combined because they were indistinguishable on the well logs and poor reservoir rocks. Figure 37 is a graph of porosity and permeability of the dolopackstones showing the difference between the mud-rich and mud-lean dolopackstones. The dolorudstones and dolofloatstones shown in Figure 38 also show a significant difference in porosity and permeability. Figure 39 is a graph of core porosity and permeability for the seven lithofacies indicating the grain-rich facies; dolograinstone, mud-lean dolopackstone, dolorudstone, and mud-rich dolopackstone are the primary reservoir facies (lithofacies 4–7). Lithofacies 1–3, dolomudstone, dolowackestone, and dolorudstone are of low reservoir quality and potential barriers or baffles to flow. A method of predicting lithofacies in noncored wells was developed based on the multiple cross-product, nonlinear method as documented by Lawrence et al.

(1997). The FACIESG curve was calculated using the core lithofacies data from cored wells 2– 5. Seven distinct carbonate lithofacies and shale were identified from the core description that could be distinguished using well-log curves. The pore system is all secondary in nature and is primarily intercrystalline because of the fabric-destructive dolomitization. The original depositional textures appear to have influenced the development of the intercrystalline pores. The mud-dominated lithofacies (lime mudstone, dolomudstone, dolowackestone, and dolofloatstone) (Figure 40a–c) are of poor reservoir quality because of the predominance of original fine lime micrite matrix or dolomitized muddy matrix resulting in connected pore system-dominated very fine intercrystalline micropores. Figure 40d–g are thin-section photomicrographs of the more grain-dominated lithofacies (mud-rich dolopacktstone, dolorudstone, mudlean dolopackstone, and dolograinstones). These rock

258 / Dull et al.

FIGURE 13. Core slab photos 2722 – 2740 ft (829 – 835 m) for cored well 3.

FIGURE 14. Core slab photos 2740 – 2753 ft (835 – 839 m) for cored well 3.

FIGURE 15. Core slab photos 2753 – 2771 ft (839 – 844 m) for cored well 3.

FIGURE 16. Core slab photos 2771 – 2784 ft (844 – 848 m) for cored well 3.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 259

FIGURE 17. Core slab photos 2784 – 2802 ft (848 – 854 m) for cored well 3.

FIGURE 19. Core slab photos 2815 – 2833 ft (858 – 863 m) for cored well 3.

FIGURE 18. Core slab photos 2802 – 2815 ft (854 – 858 m) for cored well 3.

FIGURE 20. Core slab photos 2833 – 2851 ft (863 – 868 m) for cored well 3.

260 / Dull et al.

FIGURE 21. Core slab photos 2851 – 2869 ft (868 – 874 m) for cored well 3.

FIGURE 23. Core slab photos 2956 – 2974 ft (900 – 906 m) for cored well 3.

FIGURE 22. Core slab photos 2920 – 2938 ft (890 – 895 m) for cored well 3.

FIGURE 24. Core slab photos 2974 – 2989 ft (906 – 911 m) for cored well 3.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 261

FIGURE 25. Core slab photos 2989 – 3007 ft (911 – 916 m) for cored well 3.

FIGURE 27. Core slab photos 3025 – 3040 ft (922 – 926 m) for cored well 3.

FIGURE 26. Core slab photos 3007 – 3025 ft (916 – 922 m) for cored well 3.

FIGURE 28. Core slab photos 3040 – 3058 ft (926 – 932 m) for cored well 3.

262 / Dull et al.

FIGURE 29. Core slab photos 3058 – 3076 ft (932 – 937 m) for cored well 3.

FIGURE 30. Core slab photos 3076 – 3094 ft (937 – 943 m) for cored well 3.

FIGURE 31. Core slab photos 3094 – 3202 ft (943 – 975 m) for cored well 3.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 263

FIGURE 32. Core slab photo from cored well 3 (2857 ft; 870 m) showing uneven and eroded hardground with burrows and root casts filled with anhydrite overlain by organic-rich lagoonal shale.

fabrics are of better reservoir quality because of the predominance of more coarsely crystalline dolomite, producing the larger, better connected poor system and lack of microporosity associated with a muddy matrix. In addition, the permeability of the grain-dominated lithofacies is higher because of common fine interparticle pores and associated late-stage etching. Table 2 shows the mean porosity and permeability of the carbonate lithofacies ranked by reservoir quality. Lithofacies data from well 1 were not used because of difficulties ensuring a reasonable depth match given the very limited amount of core data. The welllog curves PHIE (effective porosity), Sw (water saturation), VCLGR (clay volume from corrected gammaray curve), CGR (corrected gamma ray for uranium), and SGR (spectral gamma ray) were used to calculate

FIGURE 33. Subaerial hardground overlying karst breccia with solution-enhanced open fractures and void-filling anhydrite. Cored well 5 at depth of 2765 ft (842 m).

264 / Dull et al.

The FACIESG curve was critical to the fieldwide correlation for the Maastrichtian because it provided lithofacies curves to assist in identifying the cycle hierarchy. The cycle hierarchy and the HFSs identified in the core description, coupled with the FACIESG curves on the noncored wells, were used for fieldwide correlation and to construct the 3-D sequencestratigraphic framework. The VCLGR curve generated from the corrected CGR was used in the sequence-stratigraphic correlation, particularly in the upper Maastrichtian HFS M40 to M00. The core description revealed that many of the hardground surfaces were capped with organic-rich shale (Figure 41). The hardgrounds are consistently identified by their low porosity and permeability and appear as mudstones on the FACIESG curves. The shale can easily be distinguished with the VCLGR curve as shown on the type log given previously in Figure 7.

FIGURE 34. Dolomudstone with anhydrite after gypsum crystals (cored well 4 at 2854 ft [869 m]).

the FACIESG curve. The correlation coefficient varied between 0.41 and 0.79. A relatively low correlation between actual facies and predicted facies (FACIESG) was found for wells 2 and 4 and was attributed to having less core data as compared to the other two wells and are thus likely to be less precisely depth matched than the other cored wells.

FIGURE 35. Core slab photographs from cored well 3 showing cycle top of HFS M70 of high-energy peloidal dolograinstones capped with subtidal-marine low-energy mud-rich peloidal dolopackstones and dolowackestones (3007 – 3025 ft; 916 – 922 m).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 265

FIGURE 36. Core slab photographs (2878 – 2905 ft; 877 – 885 m) from cored well 5 showing dolofloatstone capped with intertidal dolomudstone, karst, and hardground defining the top of HFS M40.

Lower Maastrichtian HFS M90 to M50: Transgressive Systems Tract The deposition in the lower Maastrichtian reservoir from the HFS M90 to HFS M50 interval represents a marine transgression with a maximum flood

back in the middle Maastrichtian. The maximum flooding surface (MFS) in HFS M50 at Wafra field is interpreted to be equivalent to the MFS K180 of Sharland et al. (2001). The maximum flooding event resulted in the deposition of the second Maastrichtian shale across most but not all of the Wafra field area. The TST is composed of five HFS. The HFS M90 to M60 represents overall shallowing-upward sequences composed of cycles from subtidal, peloidal, and bioclastic dolowackestone to dolopackstone or dolograinstone and, less commonly, subtidal mudstone to peloidal dolograinstone cycles. The uppermost HFS records the maximum transgression and is composed of marine shale, argillaceous mudstone, and peloidal wackestone. The existence of the peloidal grainstones in the lower Maastrichtian in the M70 interval in cored well 3 seems to indicate a lessrestricted and higher energy environment of deposition, with cycles that are generally thicker, indicating a greater accommodation space. The deepening HFS M50 includes an MFS that is marked by the deposition of the second Maastrichtian shale. The second Maastrichtian was probably deposited in more open-marine conditions as evidenced by the lack of organic matter as compared to the upper Maastrichtian and as suggested from the palynological interpretation from HFS M50 in wells 3, 4, and 5 as given in Table 1.

FIGURE 37. Graph of porosity and permeability for lithofacies showing the difference on reservoir quality between mud-rich (4) and mud-lean dolopackstone (6).

266 / Dull et al.

FIGURE 38. Graph of porosity and permeability for lithofacies showing the difference on reservoir quality between dolofloatstone (3) and dolorudstone (5).

Upper Maastrichtian HFS M40 to M00: Highstand Systems Tract The upper Maastrichtian HST is composed of five shallowing-upward HFS sequences commonly capped with hardgrounds and organic-rich shale. Flugel (2004, p. 206) describes hardgrounds as subtidal cementation at the water-sediment interface as ‘‘related to a

combination of nondeposition or low accumulation rates, and condensation.’’ Hardgrounds are surfaces that are cemented before the next sedimentation event. The core data show that the HST is typically characterized by thin cycles, indicating less accommodation space, and also exhibits a greater variability in facies with relative sea level changes. The lack

FIGURE 39. Graph of porosity and permeability for the seven lithofacies showing that the more grain-dominated rocks have higher permeability. The best reservoir rock is the graindominated lithofacies shown in red (7 = dolograinstone; 6 = mud-lean dolopackstone) and green (5 = dolorudstone; 4 = mud-rich dolopackstone). The poor reservoir rock and potential baffles or barriers to flow are shown in blue (3 = dolofloatstone; 2 = dolowackestone; 1 = dolomudstone).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 267

of accommodation space and resulting thin cycles in the upper Maastrichtian is reflected in the number and types of hardgrounds. Two types of hardgrounds dominate the upper Maastrichtian: a subtidal firmground to incipient hardground and an intertidal to supratidal hardground as described by Hillgartner (1998). Hillgartner (1998), p. 1095, described the subtidal firmground as ‘‘low accumulation rate in a lagoonal environment, which favors consolidation and incipient cementation of the sediment at the water-sediment interface.’’ The upper Maastrichtian firmgrounds show characteristics similar to those described by Hillgartner (1998): highly bioturbated with overlying sediment infilling the burrows and an irregular surface probably caused by weak erosion and/or differential cementation. The overlying sediment infilling the burrows in this case is organic-rich lagoonal shale as shown in Figure 41. The intertidal to supratidal hardgrounds tend to be flat with a great deal more of associated anhydriteinfilling burrows, fenestral pores, and evidence of exposure such as tension gashes. The intertidal to supratidal hardgrounds are occasionally found above microkarsts. The recognition of hardgrounds of low porosity and permeability in the upper Maastrichtian is important not only for cycle correlation but for potential impact on reservoir compartmentalization. The incipient subtidal and intertidal hardgrounds as described by Hillgartner (1998) were found to have lateral extents from 0.1 to 1 km (0.06 to 0.6 mi). Some of the hardground-capped cycles cannot be correlated in the upper Maastrichtian, suggesting that the hardgrounds may not act as continuous barriers to flow throughout the entire reservoir. Floatstones and rudstones dominate the middle part of the HST from the M50 to the M20 marker. The floatstones-rudstones are composed of broken to intact rudists (Figure 42), bivalves, and gastropods. The floatstones commonly contain the large benthic foraminifer, Loftusia (Figure 43). The floatstonerudstone less commonly contains corals, stromatoporoids, and bryozoans. The deposition of the rudist and skeletal debris is interpreted to be biostromal, with no evidence of mounding. An abrupt facies change is indicated from the HFS M20 to M10 from restricted lagoon to more open-marine deposition of peloidal and bioclastic packstones. A subareal exposure surface occurs at the top of HFS M10 (Figure 12) that is porosity destructive and provides the seal on the oil-productive interval. This is shown on the cores as low porosity and permeabil-

ity and on the well logs as a sharp change from low to high water saturation (Figure 44). The HFS M10 surface separates the oil-productive interval from known water production from the HFS M00. A marine transgression across the top of the Maastrichtian formation overlies the unconformity surface M00 that marks the top of the Maastrichtian. HFS M00 in cored well 5 shows a karsted interval capped with an intertidal hardground that is overlain with marine shale (Figure 45). The palynological and x-ray diffraction data from the shale for the HFS M50 to HFS M00 suggest that the shale deposition transitioned from a quiet lagoonal setting to marine settings that are interpreted as localized marine incursions that resulted from eustatic sea level rise.

GEOSTATISTICAL RESERVOIR MODEL A geostatistical model for the Maastrichtian reservoir was generated to assess OOIP and reservoir development potential. Figure 3 shows the area modeled and the distribution of wells with appropriate well-log data (e.g., the synthetic facies curve FACIESG described above along with porosity and Sw curves). The modeled area covers 144 km2 (55.6 mi2). The modeled interval includes the entire Maastrichtian reservoir and the top of the Maastrichtian (HM00) to the top of Hartha. The overall model grid dimensions are 106  136  622 cells (8.97 million cells total). The areal cell size is 100  100 m (330  330 ft), and the vertical layering is nominally 0.3 m (1 ft) between the HM00 and HM90 markers. The nominal vertical layering is 0.6 m (2 ft) from the HM90 marker to the base of the model. The geostatistical property model was generated using an industry standard workflow (Deutsch, 2002; Kelkar and Perez, 2002; Goovaerts, 1997). Porosity was distributed by sequential Gaussian simulation (sGs) constrained by sequence-stratigraphic layers. Sw was distributed using a colocated cokriging with sGs algorithm also constrained by sequence-stratigraphic layers. Tables 3 and 4 summarize the input data and semivariograms used to build the reservoir property model. Although not used to constrain porosity or Sw distribution, several additional reservoir properties, such as FACIESG and Pe (photoelectric effect), were also distributed using sGs constrained by a stratigraphic layer to assist in understanding the development potential of the reservoir, particularly the sequences below M20.

268 / Dull et al.

FIGURE 40. Thin-section photomicrographs illustrating the seven dolomite lithofacies that compose the Maastrichtian arranged by reservoir quality. (a) Argillaceous lime mudstone with a few bioclasts and composed primarily of silt-size micrite and micrite that has been neomorphosed to microspar then etched, creating the microporosity. Organic matter is present as fine filamentous algal matter. Porosity occurs as micropores and reduced molds. The permeability is poor because the connected pore system is dominated by the very fine intercrystalline pores (Phi = 17.9%, k = 0.16 md). (b) Bioclastic dolowackestone or grain-poor argillaceous and organic dolomite composed of fine-grained dolomite matrix cut by coarse-grained dolomite laminae or burrows. The matrix is composed of organic matter and detrital clays but the dolomite crystals form the framework. Organic matter is common as pellets and flattened tabular grains. Porosity occurs in micropores, and permeability is poor because of the replacement of the fine carbonate matrix with tightly interlocking dolomite crystals (Phi = 4.5%, k < 0.10 md). (c) Bioclastic dolofloatstone with the most common grains are benthonic foraminifera (Rotalids) and Loftusia. Other bioclasts include ostracods, bivalve debris, gastropods, and echinoderm plates. Organic matter is found as reddish brown pellets and possibly bitumen. Uncommon angular fragments of bones are also present. The original matrix is not preserved and has been replaced or etched. Dolomite is the major mineral. Porosity is a combination of molds and secondary interparticle pores. Permeability is low as result of the variations in porosity and the frequency of organic laminae (Phi = 14.4%, k < 2.21 md). (d) Mud-rich dolopackstone in which the fabric-destructive dolomitization has largely destroyed the original depositional fabric. No evidence of carbonate mud matrix is preserved, and the organic matter is preserved as pellets. The finer dolomite crystal laminae suggest that the original fabric contained some lime mud. Porosity occurs as secondary intercrystalline pores with some local intracrystalline pores. Permeability is low because of the generally interlocking mosaic of the dolomite crystals. Larger isolated pores exist, but are connected by finer pore throats (Phi = 17.6%, k < 5.00 md).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 269

FIGURE 40. (cont.). (e) Dolorudstone is a mixture of fine-to medium-grained dolomite that has partly replaced grains and relatively coarse dolomite, destroying much of the original rock fabric. The fabric suggests an original bioclastic grainstone that was overpacked. No matrix appears to have been present. The framework grains are commonly rudist debris, bivalve debris fragments of Orbitoides with possible stromatoproids and/or Loftusia. The porosity is all secondary as intercrystalline pores. Calcite cement has reduced some of the finer and more isolated pores. Permeability is higher because of the occurrence of fine interparticle pores (Phi = 19.8%, k < 80.0 md). (f ) Mud-lean peloidal dolopackstone (Phi = 15.3%, k < 9.5 md) that has undergone fabric-destructive dolomitization and dissolution of the original rock. Large tabular molds may have been molluskan debris. Only ghosts of what appear to be medium-grain-size peloids, ooids, or other rounded grains are found. Organic matter is present as fibrous or brown amorphous pellets. Porosity is composed exclusively of secondary pores, primarily intercrystalline pores, although reduced molds are also found. Some intracrystalline pores exist but are uncommon and ineffective. Permeability is good probably because of the uneven distribution of pores and appears to be associated with carbonaceous laminae, suggesting that the original fabric may have been a poorly sorted grainstone. (g) Peloidal dolograinstone with a mixture of fine- to mediumcrystalline dolomite that has partly replaced grains and coarsely crystalline dolomite, which has largely destroyed the rock fabric. Extensive pores created by leaching of undolomitized limestone have been extensively cemented by calcite. No matrix appears to have been present. Partly mimetic-replaced bioclasts suggest small benthonic foraminifera, molluskan debris, algae, and fragments of large foraminifera. Very uncommonly are grain outlines complete and almost always rounded or tabular. Porosity occurs as secondary intercrystalline pores. Permeability is high because of late-stage etching and an open dolomite framework (Phi = 19.9%, k = 1108 md).

270 / Dull et al.

Table 2. Average core porosity and permeability by lithofacies. Lithofacies*

Mean Permeability (md)

Mean Porosity

1 2 3 4 5 6 7

0.022 1.445 20.550 11.477 81.037 114.817 294.868

0.086 0.080 0.114 0.134 0.182 0.200 0.202

*1 = dolomudstone; 2 = dolowackestone; 3 = dolofloatstone; 4 = mud-rich dolopackstone; 5 = dolorudstone; 6 = mud-lean dolopackstone; 7 = dolograinstone.

locations as well as select exploration targets in the Maastrichtian reservoir’s deeper portions. Figure 48 shows the variation in Pe and, hence, limestone and dolomite throughout the reservoir. Note that the dolomite is much more abundant in the more productive, upper parts of the Maastrichtian.

MAASTRICHTIAN DEVELOPMENT POTENTIAL The Maastrichtian reservoir contains more than 1 billion bbl of oil in place. The degree of compartmentalization of the Maastrichtian can be shown by

As shown in Table 3 and Figures 46 and 47, a general increase in porosity with depth is present, along with an increase in Sw. Porosity is very irregularly distributed in the upper parts of the reservoir, but is generally best developed in the M10 and M20 intervals. The Sw distribution shown in Figure 47 provides the best depiction of the distribution of productive and potentially productive parts of the Maastrichtian reservoir at Wafra. The middle parts of the Maastrichtian show isolated regions of high porosity and low Sw that may be near-term exploration targets. Developing a proper sequence-stratigraphic framework and using it to constrain 3-D reservoir models is key if the 3-D models are to be used to prioritize infill

FIGURE 41. Core slab photo from cored well 2 of black coaly shale in the upper Maastrichtian deposited on an uneven subaerial hardground. (12816 ft [858 m]).

FIGURE 42. Core slab photo of rudist that appears to be intact, center filled with blue anhydrite (cored well 5, 2722 ft [829 m]).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 271

area of the model also shows significant HCPV development in the M40 that trends northeast to southwest along the paleohigh. The northeast trend of prospective reservoir development appears to continue outside the model area beyond well control. No Maastrichtian well completions exist in this area, and the model suggests significant exploration potential to the north of the model area along the Maastrichtian paleohigh and along paleodepositional strike. Note that if the reservoir property models used to generate the HCPV distributions were not constrained by stratigraphy, but instead by lithology, the trend shown in Figure 50 probably would not have been captured, much less be used, to assist development and exploration planning. Additional studies, incorporating seismic attributes, are in progress to further refine the 3-D reservoir property maps and to enhance the usefulness of the models to assist reservoir development.

SUMMARY The methodology for developing the sequencestratigraphic framework for the Maastrichtian at Wafra field included a workflow to integrate core and well-log data enabling the construction of the 3-D sequence-stratigraphic framework. The workflow for the Maastrichtian consisted of   

FIGURE 43. Core slab photo of Loftusia dolowackestone (cored well 4, 2826 ft [861 m]).

maps of hydrocarbon pore volume (HCPV) for the HFS M10 and M40 shown in Figures 49 and 50. The trend of the highest HCPV for HFS M10 is coincident with present-day structural trends and represents a peloidal dolopackstone shoal complex of the late HST. The M10 is the most extensive and productive interval of the Maastrichtian. Whereas, the HFS M40 shows a northeast to southwest linear trend that likely parallels paleotopography. The northeast

using the five cores to define the depositional environment and cycle hierarchy correlating the five cores to define the ten HFSs developing synthetic facies curves from the welllog data that enabled the use of the 123 well logs with FACIESG and VCLGR curves to construct the 3-D stratigraphic framework

The sequence-stratigraphic framework is essential to the geostatistical modeling and to reservoir simulation because it controls the spatial distribution of petrophysical properties and identifies the chronostratigraphic surfaces that act as barriers and baffles to flow that compartmentalize the reservoir. The modeling of the Maastrichtian reservoir has revealed   

a large oil resource of more than 1 billion bbl of oil in place that entrapment of hydrocarbons is primarily stratigraphic a highly compartmentalized reservoir

272 / Dull et al.

FIGURE 44. A stratigraphic cross section of the cored wells shows the varying water saturation observed in the reservoir interval and the grain-dominated lithofacies that typically have low water saturation. The M00 at the top of the Maastrichtian from production tests is known to be water productive. Besides the M00, the M90 interval is the only Maastrichtian sequence that shows nearly 100% water saturation. Depth in feet. 

multiple oil-water transition zones no single definitive oil-water contact a water zone above and below the oil reservoir additional development potential associated with paleohighs and facies trends

and vertical hydrocarbon distribution of the Maastrichtian that will have a substantial impact on the reservoir development.

The sequence-stratigraphic framework and geostatistical modeling have quantified a very significant asset and provided great insight into the areal

The authors thank the Ministry of Petroleum and Mineral Resources, Kingdom of Saudi Arabia; Saudi Arabian Texaco; and Chevron Energy Technology

  

ACKNOWLEDGMENTS

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 273

Table 3. Summary of the well-log porosity and water-saturation data used to construct the reservoir property models listed by HFS. Stratigraphic Interval

Average Porosity

Average Sw

M00 M10 M20 M30 M40 M50 M60 M70 M80 M90

0.166 0.150 0.137 0.131 0.130 0.125 0.127 0.135 0.165 0.202

0.795 0.531 0.635 0.611 0.670 0.761 0.607 0.676 0.779 0.896

FIGURE 45. Core slab photo of top of HFS M00 in cored well 5 (2567 – 2568 ft; 782.4 – 782.7 m) showing hardground capped with shaly dolomudstone and a dark organic-rich marine shale marking a marine incursion in the late Maastrichtian. Note the large lithoclast above the irregular hardground surface.

274 / Dull et al.

Table 4. Summary of the variograms used in the property distribution of the geostatistical models. Interval

Range x (m)

Range y (m)

Azimuth

Range z (m)

Nugget

Sill

Form

M00 M10 M20 M30 M40 M50 M60 M70 M80 M90

2460 1660 820 1830 1600 1100 2060 1720 1760 1860

1350 1220 660 1260 1110 710 1140 1380 1020 1000

N1258E N608E N458E N258E N258E N408E N108E N358E N408E N358E

6.0 2.7 4.6 4.4 3.4 2.6 4.3 5.5 6.6 9.8

0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 1 1 1

Exponential Exponential Exponential Exponential Exponential Exponential Exponential Exponential Exponential Exponential

FIGURE 46. Perspective view of Wafra Maastrichtian reservoir model showing porosity distribution. Note that porosity distribution is very discontinuous, even in the generally higher porosity, lowermost stratigraphic zones. Within the uppermost stratigraphic zones, only the M10 and M20 intervals show consistent high-porosity distribution across large areas of the field. Vertical exaggeration = 5. Map shows the structure on top of the Hartha Formation (base of Maastrichtian).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 275

FIGURE 47. Perspective view of Wafra Maastrichtian reservoir model showing Sw distribution. Note the general increase in Sw with depth. Also note that some of the middle stratigraphic zones (e.g., M40) have locally low Sw regions. Vertical exaggeration = 5. Map shows structure on the top of the Hartha Formation (base of Maastrichtian).

FIGURE 48. Perspective view of Wafra Maastrichtian reservoir model showing Pe (photoelectric effect) distribution. Note the general increase in Pe with depth showing the irregular distribution of limestone ( yellow) and dolomite (blue). Note, however, that in general, there is much more dolomite in the upper parts of the Maastrichtian. Vertical exaggeration = 5. Map shows structure on the top of the Hartha Formation (base of Maastrichtian).

276 / Dull et al.

FIGURE 49. Hydrocarbon pore volume (HCPV) map for the M10 to the M20 interval. The area shown in orange is greater than 10 ft (3 m) of HCPV. The wells that have been completed in the Maastrichtian are also shown. Contour interval is 2 ft (0.6 m).

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 277

FIGURE 50. Map of hydrocarbon pore volume (HCPV) for M40. The greatest of amount of HCPV is coincident with the maximum dolograinstone thickness. This observation strongly supports the premise that the hydrocarbon trapping is primarily stratigraphic. The contour interval for M40 HCPV map is 1 ft (0.3 m).

278 / Dull et al.

Company for their permission to share the results of this study. We also thank the reviewers for their insight and guidance to this study: Steven L. Bachtel, ConocoPhillips Company; Sherry Becker, ExxonMobil Upstream Research Company; and P. Mitch Harris, Chevron Energy Technology Company.

REFERENCES CITED Abdel-Kireem, M. R., A. M. Samir, and H. P. Luterbacher, 1994, Planktonic foraminifera from the Kolosh Formation (Paleogene) of the Sulaimaniah – Dokan region, northeastern Iraq; Tuebingen University Neues Jarhrbuch Geol.: Paleontology Monatsh, v. 9, p. 517 – 527. Alsharhan, A. S., and S. J. Y. Nasir, 1996, Sedimentological and geochemical interpretation of a transgressive sequence: The Late Cretaceous Qahlah formation in the western Oman Mountains: United Arab Emirates University Sedimentary Geology, v. 101, no. 3 – 4, p. 227 – 242. Alsharhan, A. S., and C. G. St. C. Kendall, 1995, Facies variation, depositional setting and hydrocarbon potential of the Upper Cretaceous rocks in the United Arab Emirates: Cretaceous Research, v. 16, p. 435 – 449. Alsharhan, A. S., and A. E. M. Nairn, 1990, A review of the Cretaceous formations in the Arabian Peninsula and Gulf: Part III. Upper Cretaceous (Aruma Group) stratigraphy and paleontology: Journal of Petroleum Geology, v. 13, p. 247 – 266. Alsharhan, A. S., and A. E. M. Nairn, 1997, Sedimentary basins and petroleum geology of the Middle East: New York, Elsevier, p. 843. Beydoun, Z. R., 1991, Arabian plate hydrocarbon, geology and potential — A plate tectonic approach: AAPG Studies in Geology 33, 77 p. Beydoun, Z. R., 1993, Evolution of the northeastern Arabian plate margin and shelf: Hydrocarbon habitat and conceptual future potential: Revue de l’Institut Franc¸ais du Pe´trole, v. 48, p. 311 – 345. Danielli, H. M. C. D., 1988, The Eocene reservoirs of Wafra field, in A. J. Lomando and P. M. Harris, eds., Giant oil and gas fields — A core workshop: SEPM Core Workshop 12, p. 119 – 154. Deutsch, C. V., 2002, Geostatistical reservoir modeling: New York, Oxford University Press, 376 p. El-Nakhal, H. A., and Z. R. El-Naggar, 1994, Review of the biostratigraphy of the Aruma Group (Upper Cretaceous) in the Arabian Peninsula and surrounding regions: Cretaceous Research, v. 15, p. 401 – 416. Flugel, E., 2004, Microfacies of carbonate rocks: Analysis, interpretation and application: Berlin, Spring-Verlag, p. 976. Golonka, J., M. I. Ross, and C. R. Scotese, 1994, Phanerozoic paleogeographic and paleoclimatic modeling maps, in A. F. Embry, B. Beauchamp, and D. J. Glass, eds., Pangea global environments and resources: Cana-

dian Society of Petroleum Geologists Memoir 17, p. 1 – 47. Goovaerts, P., 1997, Geostatistics for natural resources estimation: New York, Oxford University Press, 483 p. Harris, P. M., S. H. Frost, G. A. Seiglie, and N. Schneidermann, 1984, Regional unconformities and depositional cycles, Cretaceous of the Arabian Peninsula, in J. S. Schlee, ed., Interregional unconformities and hydrocarbon accumulation: AAPG Memoir 36, p. 67 – 80. Hillgartner, H., 1998, Discontinuity surfaces on a shallowmarine carbonate platform (Berriasian, Valanginian, France and Switzerland): Journal of Sedimentary Research, Section B: Stratigraphy and Global Studies, v. 68, no. 6, p. 1093 – 1108. Kelkar, M., and G. Perez, 2002, Applied geostatistics for reservoir characterization: Society of Petroleum Engineers, 264 p. Kerans, C., and W. S. Tinker, 1997, Sequence stratigraphy and characterization of carbonate reservoirs: SEPM Short Course Notes 40, 130 p. Lawrence, T. D., A. Kohar, B. Sukamto, and H. Pramono, 1997, Sonic density well log data editing with pseudocurve generation — Indonesian examples using a multiple crossproduct non-linear method: World Petroleum Congress, p. 163 – 167. Morris, A., M. W. Anderson, A. H. F. Robertson, and K. AlRiyami, 2002, Extreme tectonic rotations within an eastern Mediterranean ophiolite (Baer-Bassit, Syria): Earth and Planetary Science Letters, v. 202, no. 2, p. 247 – 261. Nelson, P. H., 1968, Wafra field— Kuwait-Saudi Arabia neutral Zone. Second Regional Technical Symposium, Society of Petroleum Engineers, Saudi Arabia Section, Dhahran, 16 p. Philip, J. M., J. Roger, D. Vaslet, F. Cecca, S. Gardin, and A. M. S. Memesh, 2002, Sequence stratigraphy, biostratigraphy and paleontology of the Maastrichtian – Paleocene Aruma Formation in outcrop in Saudi Arabia: GeoArabia, v. 7, no. 4, p. 699–718. Sayed, M. G. S. A., and M. A. Mersal, 1998, Surface geology of Jebel Rawdah, Oman Mountains: GeoArabia, v. 3, no. 3, p. 401 – 414. Schreurs, G., and A. Immenhauser, 1999, West-northwest directed obduction of the Batain group on the eastern Oman continental margin at the Cretaceous – Tertiary boundary: Tectonics, v. 18, no. 1, p. 148 – 160. Sharland, P. R., R. Archer, D. M. Casey, S. H. Hall, A. P. Heward, A. D. Horbury, and M. D. Simmons, 2001, Arabian plate sequence stratigraphy: GeoArabia Special Publication 2, 371 p. Skelton, P. W., S. C. Nolan, and R. W. Scott, 1990, The Maastrichtian transgression onto the northwestern flank of the Proto-Oman Mountains: Sequences of rudist-bearing beach to open shelf facies, in A. H. F. Robertson, M. P. Searle, and A. C. Ries, eds., The geology and tectonics of the Oman region: Geological Society (London) Special Publication 49, p. 521 – 547.

The Sequence Stratigraphy of the Maastrichtian Reservoir at Wafra Field / 279 Smith, A. B., N. J. Morris, A. S. Gale, and W. J. Kennedy, 1995, Late Cretaceous carbonate platform faunas of the United Arab Emirates – Oman border region: Oxford University Bulletin of the Natural History Museum (Geology Series), v. 51, no. 2, p. 91–119. Van Wagoner, J. C., H. W. Posamentier, R. M. Mitchum, P. R. Vail, J. F. Sarg, T. S. Loutit, and J. Hardenbol, 1988, An overview of the fundamentals of sequence stratigraphy and key definitions, in C. K. Wilgus, B. J.

Hastings, H. Posamentier, J. C. Van Wagoner, C. A. Ross, and C. G. St. C. Kendall, eds., Sea-level change: An integrated approach: SEPM Special Publication 42, p. 39 – 46. Van Wagoner, J. C., R. M. Mitchum, K. M. Campion, and V. D. Rahmanian, 1990, Siliclastic sequence stratigraphy in well logs, cores, and outcrops: Concepts for high-resolution correlation of time and facies: AAPG Methods in Exploration Series 7, 55 p.

8

Porter, M. L., A. R. G. Sprague, M. D. Sullivan, D. C. Jennette, R. T. Beaubouef, T. R. Garfield, C. Rossen, D. K. Sickafoose, G. N. Jensen, S. J. Friedmann, and D. C. Mohrig, 2006, Stratigraphic organization and predictability of mixed coarse- and fine-grained lithofacies successions in a lower Miocene deep-water slope-channel system, Angola Block 15, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From Rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 281 – 305.

Stratigraphic Organization and Predictability of Mixed Coarse- and Fine-grained Lithofacies Successions in a Lower Miocene Deep-water Slope-channel System, Angola Block 15 M. L. Porter

C. Rossen

ExxonMobil Production Company, Houston, Texas, U.S.A.

ExxonMobil Development Company, Houston, Texas, U.S.A.

A. R. G. Sprague

D. K. Sickafoose

ExxonMobil Upstream Research Company, Houston, Texas, U.S.A.

ExxonMobil Exploration Company, Houston, Texas, U.S.A.

M. D. Sullivan

G. N. Jensen

Chevron Energy Technology Company, Houston, Texas, U.S.A.

ExxonMobil Production Company, Houston, Texas, U.S.A.

D. C. Jennette

S. J. Friedmann

Apache Oil, Houston, Texas, U.S.A.

Lawrence Livermore Laboratory, Livermore, California, U.S.A.

R. T. Beaubouef D. C. Mohrig

ExxonMobil Exploration Company, Houston, Texas, U.S.A.

Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Boston, Massachusetts, U.S.A.

T. R. Garfield ExxonMobil Production Company, Houston, Texas, U.S.A.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215880M883273

281

282 / Porter et al.

ABSTRACT

R

egional seismic mapping identified lower and middle Miocene slope channels as significant exploration targets for Angola Block 15. Seventeen exploration wells, followed by four appraisal wells, established these slope-channel complexes as a world-class development opportunity. ExxonMobil’s current development activity targets stacked turbidite-dominated reservoirs in long-reach, high-angle wellbores tied back to tension leg platform (TLP) and close-moored floating production, storage, and offshore loading facilities. One of the major development targets in Block 15 is the Burdigalian-aged (Bur1) slope-channel reservoirs. The Bur1 slope-channel system was one of many lower Miocene sediment fairways that provided a mechanism for the delivery of coarse-grained turbidites and mixed muddy and sandy debrites into the Lower Congo basin. This slope-channel system traverses across the block in an east–west direction and can be continuously mapped on adjacent seismic data sets across a 30–40-km (18–25-mi) reach. Three conventional cores and 28 well penetrations calibrate updip to downdip changes in lithofacies type and channel architecture. Map patterns of the Bur1 slope-channel system show distinctive changes in sinuosity, channel confinement, and degree of amalgamation broadly related to concurrent growth of salt-related structures. Channel-complex confinement is more pronounced, and vertical amalgamation is better developed in segments that cross structural highs. The Bur1 channel system shows weaker lateral amalgamation, greater sinuosity, and less erosional confinement in structural lows. The episodic fill of the Bur1 slope-channel system can be better understood by a hierarchical arrangement of unconformity-bounded stratal units. Within these unconformity-bounded channel sets, nested channels form composite channel complexes that show distinctive trends in lithofacies type and vertical facies succession. Compared to other offshore Angola slope-channel systems, this Bur1 system is noteworthy because of the relatively coarse granule to cobble grain sizes encountered. Well logs and high-resolution seismic data calibrated to conventional cores show that the lower parts of the channel complexes are dominated by sandy-muddy debrites, slumps, and injected sandstones. These facies are typically overlain by coarse-grained, gravelly, and well-amalgamated sandy turbidites. The overlying facies succession is more variable, but commonly consists of interbedded sandy and muddy turbidites, injected sandstones, and a range of both muddy and sandy debrites.

BURDIGALIAN DEEP-WATER CHANNEL SYSTEMS Introduction The past 5 yr have seen a remarkable expansion of offshore exploration and development activity in Angola. The highly productive hydrocarbon system in the Lower Congo basin, coupled with large structures, well-established sediment fairways, and excellent seismic coverage has made this part of west Africa a world-class development opportunity. Total’s Girassol field development showed the huge potential that exists for deep-water production from Angola’s

slope-channel systems (Bouchet et al., 2005), and the growth of ExxonMobil’s Kizomba development to more than 550 thousand bbl/day of production in 3 yr shows the prolific nature of deep-water depositional systems that exist in Block 15 (Figure 1). Seismic stratigraphy applied to conventional and high-resolution three-dimensional (3-D) data sets offered a compelling method to understanding the Tertiary history of the Angolan continental margin (Beaubouef et al., 1998; Schollnberger and Vail, 1999; Gottschalk, 2002). Recent studies of the Zaire Fan show a compelling view of successive fan development driven by erosion, sediment bypass, and punctuated

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 283

FIGURE 1. Location map of the Bur1 slope-channel system, offshore Angola. The Bur1 channel system forms a part of the greater Kizomba development. Water depth ranges from 750 to more than 1300 m (2460 to more than 4265 ft) in the study area. A 50-km (31-mi) transect of the Bur1 slope-channel system is possible because of the contiguous 3-D seismic coverage in the 400-km2 (154-mi2) study area. The distribution of oil and gas fields in the Lower Congo basin is shown by green polygons. The inset photo shows one of the two close-moored tension leg platform (TLP) and floating production, storage, and offloading vessel (FPSO) facilities producing hydrocarbons from the Bur1 system. deposition in the updip submarine canyon and leveechannel system (Babonneau et al., 2002; Droz et al., 2003). The Pleistocene to Holocene history of the Zaire Fan shows a linked shelf-slope-basin depositional system and provides an analog for interpreting the genetic packaging of complex deep-water depositional fan elements through sequence-stratigraphic concepts (e.g., Garfield et al., 1998; Raposo and Sykes, 1998; Goulding et al., 2000; Temple and Broucke, 2004). For reservoir-scale studies, placement of subseismic-scale stratigraphy in sinuous deep-water channel fills is critical in developing a depositional model suitable for development drilling (e.g., Kolla et al., 2001). A hierarchical approach to the physical stratigraphy of deep-water systems allows the smallest stratigraphic elements (beds) to be placed in increasing larger composite stratal packages (complex

system sets) commonly mapped with seismic and sequence-stratigraphic methods (Sprague et al., 2002, 2005). This chapter shows the seismic expression, map trends, and reservoir elements of a major lower Miocene slope-channel system in offshore Angola. Development drilling reveals a complex, multistage lithofacies succession of coarse- and fine-grained turbidites and mixed debrites. These lithofacies form predictable fining-upward channel fills in larger, unconformity-bounded channel-complex sets that can be mapped across tens of kilometers. Regional mapping of the Bur1 system shows the position and extent of reservoir unit changes along depositional dip, and these variations can be attributed to sea-floor gradient changes associated with salt movement. A depositional model is developed from the superposition of

284 / Porter et al.

FIGURE 2. General stratigraphy of the Lower Congo basin. The Bur1 slope-channel system is contained within the Oligocene – Miocene Malembo formation. The Malembo and the underlying Landana Formation contain the bulk of postrift, sag-phase sedimentation in the Lower Congo basin. The lower-right diagram shows a generalized depositional model of a linked shelf-slope-basin depositional system (e.g., Garfield et al., 1998). The Bur1 reservoirs discussed in this chapter represent only a small segment of a very large, deep-water depositional system. The study area is approximately located at cross section AA0 on this profile and represents a deep-water confined channel complex system in a midslope depositional setting.

unconformity-bounded packages that shows the relationships of turbidites and debrites in a nested stratal hierarchy of deep-water channel complexes.

GEOLOGIC SETTING Deep-water slope-channel systems are one of the important reservoir types in the hydrocarbon system of the Lower Congo basin. The Lower Congo basin is a Cretaceous to Holocene sedimentary basin developed on the west African continental margin. The early history of the basin is represented by a thick, Aptian-aged salt succession that formed over rift basin deposits associated with the Early Cretaceous breakup of Gondwana (Scrutton and Dingle, 1976;

Brice et al., 1982; Emery and Uchupi, 1984; Burke, 1996) (Figure 2). A postrift, early sag phase filling of the basin is represented by marine deposits of the Cretaceous Pinda Formation. Source rocks are mostly in the Upper Cretaceous Iabe Formation, which overlies the Pinda, and these shale-dominated rocks accumulated under low sedimentation rates. Fully marine conditions developed with continued sea-floor spreading of the south Atlantic, and this early Tertiary drift phase is represented by marine shales, sandstones, and carbonates of the Landana Formation. Middle to late Tertiary basin fill is contained in the Oligocene to Miocene Malembo formation. The Malembo formation is a 6-km (3.7-mi)-thick, time-transgressive lithostratigraphic unit that contains strata deposited

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 285

FIGURE 3. Seismic traverse through the Bur1 slope-channel system. The upper panel shows the seismic expression and structural configuration of the lower and middle Miocene along BB0. The Bur1 channel system is marked by dashed horizons along a 55-km (34-mi) transect. Some of the well penetrations in the Bur1 are shown with corresponding log responses. The lower panel is an interval amplitude extraction of the Bur1 slope-channel system. Note the confinement of the system near or at structural highs and more weakly confined, sinuous map pattern developed in the intervening synclines. Well penetrations, including cored wells, are shown by symbols on the seismic amplitude map. in nearshore to deep-water slope and basinal environments (Temple and Broucke, 2004). At least three major episodes of clastic input across the slope and into the basin are recognized in the Oligocene and Miocene of the basin (Ardill et al., 2002). The Bur1 deep-water slope system discussed in this chapter forms a part of the second major influx of coarse-grained clastic sediments in the lower Miocene. Biostratigraphic analysis of cores and cuttings in exploration wells shows that this slope-channel system is early Burdigalian in age. Nannofossil assemblages in bracketing hemipelagic slope mudstone successions, tied to aged-dated European reference sections, show the Bur1 slope-channel system developed ca. 20 Ma. Internal age subdivision of the Bur1 channel system is problematic because of a lack of age-diagnostic faunal assemblages and multiple episodes of erosion, bypass, and deposition. Correlated ages of capping mudstone sequences that bracket the overall slope-channel system suggest about 2 m.y. of deposition, ending at about 18 Ma.

Regional tectonic studies and structural analysis of the Lower Congo basin have established a linked extensional-contractional structural style for the Angolan continental margin (Larson and Ladd, 1973; Rabinowitz and LaBreque, 1979; Duval et al., 1992; Hartman et al., 1998; Gottschalk, 2002). Structures have developed as a result of salt-related deformation associated with detachment on the Aptian salt. Updip parts of the basin-margin slope are characterized by extensional faults and tilted fault blocks, and downdip sections show a full suite of salt structures, compressional anticlines, and thrust faults (e.g., Gottschalk et al., 2004).

THE BUR1 SLOPE-CHANNEL SYSTEM Sinuosity and Width A seismic amplitude extraction across the reservoir interval shows the large-scale map pattern of the Bur1 slope-channel system (Figure 3). The Bur1 system shows low to moderate sinuosity across a 40-km (25-mi)

286 / Porter et al.

reach and a well-developed east-to-west orientation. This depositional trend is remarkably consistent, presumably related to the regional slope gradient into the Lower Congo basin during deposition. Salt diapirs, extensional faults, and compressional folds segment the slope profile into several large, north– south-oriented anticlines and intervening synclines. These structural elements, early in their growth history, clearly influenced the width of this deep-water channel system. The Bur1 channel system is more confined, narrow, and straight as it courses over structural highs (Figure 3). Widths of the confined channel system are 700 –2200 m (2300 – 7200 ft) across positive structural elements. Channel system sinuosities, measured from interpreted thalweg segments across three anticlines in the study area, range between 1.06 and 1.15 (n = 5). In each intervening syncline, the channel system shows less erosional confinement and widths three to six times greater than over structural highs. Higher channel sinuosities (range 1.12 – 1.73, n = 5) are developed in these laterally expanded parts of the Bur1 channel system (Figure 3).

System Depth and Relief of the Thalweg High-resolution 3-D seismic data show that the composite fill thickness of the Bur1 slope system varies along a 40-km (25-mi) depositional fairway in Block 15. This variation is a product of erosional and depositional confinement of the channel complexes, proximity to growing salt structures, and accommodation provided by extensional fault networks. Average Bur1 thickness in the Kizomba development is about 120 m (390 ft), with a range between 70 and more than 220 m (230 and more than 720 ft). Paleorelief that developed on channel margins above the channel thalweg can be estimated from the thickness of channel fill, stratal onlap positions, dimensions of rotated slump blocks and their related displacement lengths, and cross sections from well-imaged, individual channel profiles. These aggregate data types suggest that the relief of Bur1 channels during deposition never exceeded 50 m (164 ft), and in some of the more weakly confined parts of the system, relief may have been as little as 5–10 m (16–33 ft). Thus, the Bur1 system is not a submarine canyon fill or the product of a single erosional event followed by deposition. Instead, the composite Bur1 sediment fill is a product of multiple episodes of sediment gravity flow scour, headward erosion, and channel-margin slumping, alternating with episodic sedimentation by turbidity and debris flows. Updip to downdip variations in Bur1 channel system fill were controlled by vari-

able sediment influx to the basin, local sea-floor gradients associated with active salt movement, and a regional depositional gradient that crosses linked extensional and contractional tectonic domains of a rapidly prograding shelf-slope system.

Levee Confinement The Bur1 channel system is bounded by moderateto low-amplitude seismic reflections that dip away at very low angles (<18) from the reservoir-prone axis (Figure 4A). Downlap positions of these dipping strata can be kilometers away from the trend of the channel system (M. Grove, 2005, personal communication). Well penetrations show that these seismic facies are low-net-to-gross silty mudstones interpreted here as flanking levees. Relief from the thalweg to levee crest is on the order of 10–120 m (33–393 ft). Multiple episodes of levee growth and erosion are evident from the stratal truncations observed on seismic traverses and comparison of levee thickness with the laterally adjacent, more erosionally bounded channel complexes. Flanking levees are best developed at the base of the channel system. Levee geometries are laterally more restricted in the younger parts of the system and show classic geometries well documented in Pleistocene successions from the Amazon and Gulf of Mexico (Pirmez et al., 1997).

Changes in Slope-channel System Confinement across Structural Highs and Synclinal Lows Erosional confinement of the Bur1 channel system is most pronounced near or across the current positive structural elements. Channel segments that cross positive structural elements are relatively straight in plan view, and this straight geometry sharply contrasts with the more highly sinuous channel patterns in the intervening synclines. Significant positive relief at the sea floor was not a factor during Bur1 deposition, as the Bur1 system crosses each of the saltcored anticlines or north – south extensional fault systems at right angles, without channel diversion or deflection. Net erosion of the sea floor by highenergy turbidity currents and levee growth was sufficient to maintain the channel system’s position and gradient across growing positive tectonic elements.

Factors Controlling the External Form of the Confined Channel System Extensive 3-D seismic coverage shows that lower Miocene slope-channel systems are large-scale depositional features that encompass more area than the well-imaged, high-amplitude channel packages

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 287

FIGURE 4. Seismic cross section through the Bur1 slope-channel system. (A) The seismic expression of the Bur1 is varied because of multiple rock types and a stratigraphy built up through multiple episodes of erosion and deposition. The axial parts of the system are seismically characterized as multicyclic, high-amplitude, semicontinuous seismic facies. Off-axis and channel-margin successions show moderate- to low-amplitude response, and these strata onlap the erosional boundary of the system. An Earth model of mixed-impedance channel fills and nonchannelized depositional elements (B) is convolved with a seismic wavelet to produce a modeled seismic response (C) that is similar to the actual seismic traverse. commonly illustrated in reservoir studies. Lateral dimensions are large, on the order of 2 – 20 km (1.2 – 12 mi). These low-amplitude seismic facies represent an integral component of the depositional system and are fundamentally related to the genesis of the associated channel belts. Establishing the exact chronostratigraphic relationships between these deposits and those of the channel fills is difficult, but it is clear that they contain a stratigraphic record of protracted overbank processes. Thickness patterns and geometries suggest that these are net depositional units formed via aggradation adjacent to evolving channel complexes that were deposited from numerous turbidity currents with strong lateral gradients in velocity, concentration, and depositional rates. Recognition of these units is important in understanding the evolution of the slope-channel system because these widely dispersed, low-net facies establish the long-

term channel system trend across the slope and into the basin. The paucity of coarse-grained lithofacies (i.e., coarse-grained sandstones, gravels, and conglomerates) in these flanking levee packages suggests largescale sediment bypass down the channel system fairway. Flow stripping of fine-gained suspension clouds, sea-floor gradients modified by mass transport complex emplacement, and shelf-edge control of sediment delivery access points determine the offset spacing of the channel system fairways.

KIZOMBA FIELD DEVELOPMENT Field History The early geoscience work and drilling history leading up to the Kizomba development is presented by Reeckmann et al. (2001). The phased development

288 / Porter et al.

FIGURE 5. Reservoir subdivision of the Bur1 slope-channel system. The Bur1 channel system can be subdivided into four unconformity-bounded channel sets. This reservoir subdivision is based on the physical relationships of strata in each channel set and contains lithofacies deposited in both channel and intrachannel areas. This approach avoids overestimation of reservoir connectivity in assuming that all sandstones are time equivalent and recognizes lateral facies changes into more shale-prone strata that baffle the reservoirs. Depth in meters.

consists of Kizomba A, a combination 36-slot TLP with a close-moored, 2.2-MMBO capacity FPSO (floating production, storage, and offloading vessel) (Figure 1). First oil was produced in August 2004, and the facility in mid 2006 is producing more than 250,000 bbl/day. Drilling activity in Kizomba A continues, with pressure support wells from five subsea drill centers. The Kizomba B development, a virtual twin of Kizomba A, began production in July 2005. In addition to reservoirs accessible from the TLP, Kizomba B takes production from the Dikanza subsea development 10 km (6 mi) to the northwest. Current production from Kizomba B exceeds 250,000 bbl/day. Productionfocused activities in the Kizomba field include drilling pressure support wells, establishing production in unpenetrated compartments, and shooting a fourdimensional, high-resolution seismic survey to manage field depletion.

Bur1 Well Penetrations in Block 15 A total of 28 wells penetrate or are completed in the Bur1 deep-water channel system (Figure 3). More than half of the wells target multitiered, gravelly sandstone reservoirs in the Bur1 slope-channel system. The high-angle, long-reach well design used in Kizomba is optimized by volume-based interpretation workflows in high-resolution 3-D seismic cubes.

Distribution of Cored and Uncored Wells Four conventional cores (554 m [1817 ft] total) establish the vertical lithofacies arrangement in the Bur1 channel-complex system. Three of the cored wells establish a 27-km (16-mi) updip to downdip transect of channel system sets in the Kizomba development (Figure 5). The cores are described in detail,

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 289

FIGURE 6. Gravel-prone lithofacies in the Bur1 cores. Gravel-prone lithofacies in the Bur1 channel system are typically thick bedded, moderate to poorly sorted bedsets bounded by sharp, erosive boundaries. The coarsest examples are R1 or R3 types that contain greater than 30% granule, pebble, or small cobble clasts. S1 and S3 bedsets contain 5 – 30% granule to pebble clasts. Low-angle cross-stratification and clast imbrication are well developed in the R1 and S1 bedsets. Color bar is 20 cm (8 in.) long. and core results are used for seismic facies calibration, rock property studies, and inputs to geologic modeling. Forty kilometers (25 mi) to the north, wells in Xikomba field produce from a time-equivalent but geographically distinct Burdigalian deep-water channel system, and these cores show a comparative channel-fill succession of turbidites and debrites.

DEEP-WATER LITHOFACIES IN CORED WELLS Four facies associations are observed in cores from the Bur1 channel-complex system. These facies associations are distinct lithofacies successions of genetically related bedsets, each characterized by distinctive sedimentary textures, bed thicknesses, and petrologic constituents. All the deep-water facies associations in the Bur1 slope system are common to other confined

channel systems studied in subsurface and outcrop data sets (Raposo and Sykes, 1998; Sykes et al., 1998; Campion et al., 2000).

Facies Association 1 This is a gravel-prone facies association consisting of clast-supported sandy gravels that display massive, ungraded to normally graded clast fabrics (Figures 6, 7).

Description Bed thickness averages 1.3 m (4.2 ft), with a range between 0.3 and 8 m (1 and 26 ft) thick. Gravels and sandy conglomerates show a framework-supported texture and contain greater than 30% granule- to pebble-size clasts. Elongate pebble- to cobble-size clasts show well-developed planar to inclined, imbricated clasts between sharply erosional bed boundaries.

290 / Porter et al.

FIGURE 7. Gravel-prone, traction-dominated lithofacies associations. A vertical progression of lithofacies in gravel-prone, traction-deposited bedsets consists of a lower part of very large shale rip-up clasts that are overlain by clast-supported gravels and very coarse sandstones. These lithofacies associations are vertically well amalgamated and form the bases of thick channel complexes as represented by a blocky well-log response. Core segment is 5 m (16 ft) long. Gammaray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

Very coarse sandstones are massive in texture and marked by thin, basal accumulations of granule conglomerate with 5–30% granule and pebble clasts. Normal grading is common, and overall sorting is characterized as poor to very poor. More than 80% of clasts described in this facies association are well-rounded, quartzofeldspathic, igneous, and schistose metamorphic clasts, with minor amounts of well-indurated sandstone and metavolcanic lithologies. Isolated calcite-cemented concretions show convex upper and lower boundaries (Figure 8). Bed thickness, textures, and grain-size trends compare favorably with coarsegrained deposits categorized as R1 and R3 bed types in Lowe (1982).

Interpretation The gravel-rich beds in this facies association represent bed-load –dominated deposition from highconcentration turbidity currents. The imbricated clast

fabric, mapped distribution in high-resolution seismic data, and relation to channel-defined erosional cuts suggest deposition as large gravelly bedforms (e.g., Winn and Dott, 1979). Gravel units developed during maximum sediment throughput in active channel thalwegs by the accumulation of coarse bed load from multiple flow events.

Facies Association 2 This is a sand-prone facies association of thickbedded, well- and poor-sorted structureless to rippled sandstones (Figure 9).

Description Structureless sandstones are medium to very fine grained, and bed thickness ranges between 0.06 and 0.6 m (0.2 and 2 ft) thick, with an average bed thickness of 0.2 m (0.6 ft). Lower bed contacts are planar

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 291

FIGURE 8. Gravel-prone, traction-dominated lithofacies associations. Pebble- to granule-size lithofacies show massive to weakly stratified textures in medium to thick (0.3 – 2-m; 1 – 6.6-ft) bedsets. This lithofacies association may contain discrete, 0.5 – 1-m (1.6 – 3.3-ft) intervals of calcite cements. Convex, curved upper and lower boundaries defined by the cementation front suggest an ellipsoidal, concretionary shape to these cements. Well-log response is blocky. Core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

and erosional; upper bed contacts are planar to gently curved. Sedimentary fabric is structureless and well sorted, although sparse swirled lamination is present in thicker bedded examples. These beds typically overlie the gravel-prone lithofacies. Current-rippled sandstones include fine- to very fine-grained sandstone and interbedded silty mudstone. Bed thickness ranges from 0.06 to 0.5 m (0.2 to 1.6 ft) and averages 0.2 m (0.6 ft) thick. This facies association commonly caps the massive sandy turbidites, and contacts are conformable and generally planar in core. Sedimentary structures are dominated by 1–5-cm (0.4–2-in.)thick laminasets of current ripples. Minor amounts of bioturbation and soft-sediment deformation are present in the muddy sandstone beds. A distinctive suite of more poorly sorted sandstones with a silty or muddy matrix is also developed in the sand-prone facies association. These medium to fine sandstones are 0.3 –1.5 m (1– 5 ft) thick, and bed-planar bed contacts are well defined by erosion-

al truncation of dewatering and fluid-escape structures (Figure 9). Grain fabric is typically massive and homogeneous. Most of these thick-bedded sandstones show a slight fining-upward grain-size trend. Pebble-size floating shale clasts, swirled flow fabric, and small granules are common components of these beds. The macroscopic presence of clay-size matrix is inferred by the abundance of vertical dewatering pipes, swirled matrix flow structures, and floating, matrix-supported clasts. Particle size analysis shows 1 – 8% clay content in the sandy matrix of these bedsets.

Interpretation The vertical facies transition from underlying sandy conglomerates to these thick-bedded, well-sorted sandstones suggest rapid suspension deposition by sand-rich turbidity currents in a channel-axis setting. Rapid deposition is suggested by the thick-bedding,

292 / Porter et al.

FIGURE 9. Sand-prone, suspension-dominated lithofacies associations. Well-amalgamated sandstones show massive to normal grading and sharp, conformable bed contacts. Minor amounts of granule or shale clast-rich lags define the bases of many bedsets in this lithofacies association. Muddy sandstones of debrite origin are developed locally in the core suite. Core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

conformable bed contacts, and uniform grain-size distribution (Arnott and Hand, 1989). Massive sandstones are comparable to the Ta/Tb turbidite subdivision of Bouma (1962). In contrast to the well-sorted sandstones, mixed grain sizes and sedimentary structures in more poorly sorted, clay-bearing sandstone beds suggest a sandy debris-flow origin. The dispersed clast fabric, meterscale bedding, and abundant fluid-escape structures suggest that matrix strength was a profound control on the movement and deposition of these sandy debrites (e.g., Middleton and Hampton, 1976; Iverson, 1997). The matrix strength imparted by the small amount of clay dampened fluid turbulence and hindered flow-related textural size segregation. Numeric and physical process modeling suggests that some thick-bedded sandstones previously described as the product of amalgamated Ta beds may, in fact, be the product of sandy debrite deposition (Angevine et al., 1990; Shanmugam, 1996).

Facies Association 3 This is an interbedded sand and mud lithofacies association. Massive, current-rippled, and laminated bed types are identified on the basis of internal sedimentary structures and presence of grading (Figures 10–12).

Description Interbedded sandy and muddy facies consist of medium- to very fine-grained sandstone, siltstone, and mudstone in fining-upward, 0.1 – 0.3-m (0.3 – 1-ft) beds. These beds are comparable to classic Bouma (1962) bed types and include current-rippled sandstone, wavy to parallel-laminated sandy siltstone, and parallel-laminated mudstone (Figure 10). Sandstones are typically very well sorted. Structureless mudstones occur in 0.1 – 4-m (0.3 – 13-ft)-thick beds with planar to highly contorted bed contacts. The matrix has a swirled or deformed appearance and commonly shows shear banding and

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 293

FIGURE 10. Spectrum of sandy and interbedded sandy-muddy lithofacies in Bur1 cores. This lithofacies suite spans a complete range of sand-to-mud ratios and is best described as classic turbidites. Bed thickness is strongly related to sand content. Well-amalgamated Ta and Tb beds are medium to thick bedded, whereas Tc , Tcd, and Tde are mud dominated and thin bedded. Color bar is 20 cm (8 in.) long.

internal fabric variations on a centimeter scale. Floating sand-size quartzose grains and larger organic detritus are common in these mudstones (Figures 13, 14).

Interpretation The current-rippled, heterolithic nature and diversity of lamina types suggest that these mud-prone beds were deposited from suspension from turbidity currents (Arnott and Hand, 1989). Bed fabric in muddy turbidites succession is comparable to Bouma Tde beds (e.g., Bouma, 1962). Rippled sandy turbidites were deposited under waning-flow conditions in channelmargin and proximal overbank settings. Dilute sediment plumes associated with each turbidity current spread laterally from the axial parts of deep-water

channels. Accumulation and preservation of these fine sandstones and mudstones were enhanced by their off-axis position, removed from the erosive effects of subsequent high-concentration flows in the channel thalwegs. It is likely that sediment gravity flows varied substantially in strength, sediment volume, and texture during the active phase of channelcomplex development. This shale-prone succession accumulated in the waning stages of channel fill and abandonment, when channel avulsions likely changed the locus of channel deposition to another part of the slope complex.

Facies Association 4 This is an injected and slumped facies association.

294 / Porter et al.

FIGURE 11. Interbedded sandy-muddy lithofacies associations. Mud-prone lithofacies show parallel to wavy lamina sets and a shaly log response in the intervening intervals between channelized reservoir packages. The core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

Description Concordant (sills) and discordant (dikes) sandstones occur in a wide variety of shapes, thicknesses, and geometries in the Bur1 cores (Figure 15). Sandstones are well sorted, and grain size is very restricted, in the very fine- to fine-sand range. The internal structure is commonly massive, and some of the thicker discordant beds have small shale rip-up clasts at the upper and lower contacts (Figure 16). Injected sandstone thickness ranges from millimeter scale to 0.75 m (2.5 ft) in dikes, and 0.4–1 m (1.3–3.3 ft) thick in sills. In some sections, more or less isolated sills of sandstone pass upsection into a complex network of vertical and horizontal injected sandstones throughout a 10–15-m (33–49-ft) interval. Stratigraphically below coarse-grained reservoir sections, many cored intervals contain 0.5–5-m (1.6– 16-ft) blocks of folded and contorted sandstone and mudstone blocks. Internal fabrics are always deformed and never parallel to master bedding surfaces. Large

clasts commonly show internal bedding or lamination to be truncated at block margins. The blocks are embedded in sandy or muddy matrix.

Interpretation The injected sandstones are locally remobilized sands derived from well-sorted, sandy bedforms. Sandstone dikes and sills in interchannel shales may be related to sediment loading under and adjacent to rapidly deposited gravelly and sandy bedforms. Later sand remobilization into overbank and capping abandonment shale successions likely occurred during shallow burial of the sandy channel complexes. Lithologyrelated permeability contrast between the sandy channel fills and mud-prone overbank facies with compaction led to overpressure in porous sands. Hydrostatic pressure release forced these well-sorted sands into confining mudstones as a series of dikes and sills. The origin of the slumped strata is more problematic and may represent more than one depositional

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 295

FIGURE 12. Interbedded sandy and muddy lithofacies associations. Muddy debrites form the upper part of a welldeveloped fining-upward well-log motif (arrow). The muddy debrites contain 2 – 20% sand in the clay-rich matrix. These muddy debrites are overlain by thin-bedded Tc , Tcd, and Tde turbidites. Thickness of this lithofacies association is commonly 2 – 8 m (6.6 – 26 ft). The core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

process and episode of deposition. Proximity of slump deposits to major erosional surfaces implies channel instability and undercutting by energetic turbidity currents. Crosscutting relations of deformed strata with injected sandstones suggest that slump blocks failed from oversteepened cut banks and were buried prior to compaction by deposition of bed-load – dominated bar deposits.

FACIES ARCHITECTURE OF THE BUR1 DEEP-WATER CHANNEL SYSTEM Vertical and Lateral Facies Stacking Arrangements High-resolution seismic data sets and numerous well penetrations establish some general trends of vertical and lateral facies relationships in the Bur1 slope-channel system. In most areas where current development activities are focused, gravel-prone lithofacies are common in mapped channel-axis positions. The lower one third to one half of the channel

fill consists of thick-bedded gravels and very coarse sandstones (Figure 17). Upper parts of the channel fills consist of interbedded sand-prone turbidites and mixed muddy and sandy debrites, thin-bedded muddy turbidites, and injected sandstones. Multiple, stacked channel-fill deposits can be grouped with laterally associated channel-margin and overbank deposits to form genetically distinct channel complexes. Seismic mapping of erosion surfaces, together with correlation of low-net facies related to channel-complex abandonment and lateral migration, provides a fourfold, unconformity-bounded subdivision of the Bur1 into channel-complex sets (e.g., Sprague et al., 2002) (Figures 5, 18).

Coarse-grained Gravels in Confined Channel Complexes The gravel-prone facies succession shows a highimpedance seismic character that allows detailed geobody extractions in the Bur1 channel system (Figure 19).

296 / Porter et al.

FIGURE 13. Spectrum of debrite types in the Bur1 cores. A complete spectrum of lithofacies with floating, outsized lithoclasts, swirled or deformed primary bedding geometries, dewatering pipes, and interstitial clay is present in the Bur1 system. Completely sand-prone or completely mud-prone examples occupy the ends of a true spectrum of bed types and are interpreted here as sandy debrites and muddy debrites. Most of the debrites cored in the Bur1 fit into the middle part of the bed-type spectrum and are pragmatically separated into sandy-muddy debrites (poorer reservoir quality) and muddy-sandy debrites (better reservoir quality) on the basis of sand in the matrix, water saturation, permeability values, and bed thickness. Color bar is 20 cm (8 in.) long.

Seismic facies mapping in high-resolution 3-D cubes indicates these geobodies to be an excellent proxy for the size, distribution, and shape of both tractiondeposited gravel bedforms and intervening intrachannel units, such as thick- to thin-bedded turbidites, and mixed sandy-muddy debrites (e.g., Goulding et al., 2000; Beaubouef, 2004). Horizon-keyed amplitude extractions along the erosional base of the Bur1 channel system show interpreted gravel bars and coarse-grained thalweg deposits to be elongate and paddle shaped (Figure 19). Downflow-directed lengths are 5– 10 times longer than corresponding widths. The long-axis orientation of the gravel bars is never parallel to the thalweg po-

sition as determined from isochron maps. Instead, gravel bars splay 30 –808 off the downcurrent flow direction. Interpretation of seismic amplitudes in the channel system suggests that gravel-prone units are not uniformly deposited across the width of the Bur1 system. Strong stratigraphic control of these units by channel migration and incision is inferred by sinuous map patterns and cross-sectional geometries in channel complexes. Although late-stage incision or unrelated subsequent erosion by turbidity currents cannot be discounted, the oblique orientation of the preserved gravelly bars to the overall trend of the sinuous channel suggests a primary, preserved depositional extent to these deposits.

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 297

FIGURE 14. Mud-prone lithofacies associations. Mixed muddy debrites form highly varied successions of thick- and thin-bedded, contorted facies successions. Log response is serrate to bell shaped. Thin, interbedded Tc and Tcd turbidites commonly punctuate the thicker debrite-rich intervals. Permeability contrast between turbidite-deposited sandstones and muddy debrites is indicated by the degree of hydrocarbon saturation. The core segment is 5 m (16 ft) long. Gammaray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

Implications of the Gravel-prone Facies Successions The abundance of gravel-prone lithofacies in channel complexes of the Bur1 system illustrates some important geologic features and processes that operate within large slope-channel systems. These gravelly sediments are clearly associated with channel erosion and fill, organized in bedforms, and do not represent eroded coarse-grained remnants of older strata. Framework clast petrology is essentially that of Proterozoic igneous and metamorphic rock suites of the Angolan shield, and this implies an updip, direct shelfal input of gravels or cannibalization of older, coarse-grained shelf-margin fluvial-mouth bar deposits by shelfedge failure and headward erosion up the channel systems. A relatively narrow shelf and Tertiary-aged regional tectonic uplift were important factors in keeping the lower Miocene slope-channel systems of Angola charged with coarse-grained sediments (Brice et al., 1982).

The lower Miocene shelf margin, identified on regional seismic lines, was at least 40–50 km (25–31 mi) to the east of the study area. The competency of transporting sediment gravity flows was considerable. Clast sizes of as much as 45 cm (17 in.) in diameter are present in the cored suite, and pebbles to small cobbles comprise most of the gravel-prone bedsets. Larger framework clast sizes likely exist in these deposits but are not represented in the cored lithofacies because of the mechanical difficulty of coring in largely unconsolidated reservoirs by extended-reach wells. Low-angle cross-stratification, imbricate clast fabrics, and grain-size trends suggest that the gravelly units were deposited as traction carpets by highconcentration turbidites (e.g., Winn and Dott, 1977; Lowe, 1982). Effective transport of such coarse-grained lithofacies more than 50 km (31 mi) from the timeequivalent shelf margin implies tremendously energetic sediment gravity flows into the Lower Congo basin.

298 / Porter et al.

FIGURE 15. Mud-prone lithofacies associations. Thick successions of muddy debrites are uniformly gray in color and may show discrete, well-sorted disconformable intervals of fine-grained sandstone. Crosscutting relations of the sandstone with the bedding geometry of muddy debrites suggest a postdepositional injection origin of the sandstones. The shaly log response is dependent on the amount of injected sandstone; those intervals with abundant sandstone injections show a highly serrate log pattern. Core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log.

Deposition of the gravel units was focused in the deeper thalweg part of channel complexes. Erosional bedset boundaries suggest episodic development of the barforms, and numerous high-concentration flows likely contributed to the volume and arrangement of the gravels preserved in the channel fills. Because the gravelly units may represent only a small fraction of the total sediment load in large, turbulent sediment gravity flows, the preservation of thick, amalgamated gravelly packages implies that much greater amounts of sand and mud have completely bypassed the study area. It is likely that these missing sediments built downdip depositional elements of the large, linked deep-water slope and basin depositional system. It is without a doubt that these depositional elements of the lower slope and basin deep-water systems will be better understood as exploration pushes into the ultradeep parts of the Lower Congo basin.

Depositional History of the Bur1 Slope-channel System The integration of high-resolution 3-D data, cores, and well logs demonstrates a composite sedimentary record for the Bur1 slope-channel system. Cored lithofacies show both turbidity current and debris-flow deposits in the composite fill of the channel system (Figure 20). Multiple episodes of erosion and deposition produced at least four unconformity-bounded channel-complex sets over the span of 2 m.y. Each of these channel sets shows both lateral and vertical connectivity and can be subdivided into an axis dominated by gravel-rich traction deposits, highly amalgamated sandstones, and laterally flanking strata, which contain thin-bedded sandstone and mudstones, injected sandstones, and slumps deposited in channel-margin, levee, and overbank areas of the Bur1 slope-channel system.

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 299

FIGURE 16. Injected sandstone and debrite lithofacies associations. This segment shows some of the facies organization at the base of the Bur1 system. The slope-channel system boundary (white dashed line) can be tied to seismic data as a regional mapping horizon. The contact is sharp and uncorformably separates the Bur1 channel system from older slope mudstones (gray lithologies). The lithofacies succession that sits directly on this unconformity is highly variable. In this core, sandy debrites, slump blocks, and injected sandstones overlie the contact. Core segment is 5 m (16 ft) long. Gamma-ray (GR) and deep induction resistivity (ILD) log curves are depicted on the well log. The general similarity of channel stacking pattern and constituent channel-fill elements in a 40-km (25-mi) reach of the Bur1 channel system suggests that sediment gravity flow erosion and deposition extend over long distances (tens to hundreds of kilometers). The Bur1 succession in the Kizomba development gives a glimpse into the variations of facies architecture constructed by highly energetic and competent sediment gravity flows traversing the midslope of the Lower Congo basin.

ACKNOWLEDGMENTS The authors thank ExxonMobil, together with Angola Block 15 coventurers British Petroleum, Statoil, and ENI for permission to publish our work on this

lower Miocene deep-water system. We also express our gratitude to Sociedade Nacional de Combustiveis de Angola (Sonangol) for the opportunity to study and develop these world-class reservoirs. Our work includes substantial contributions from numerous prospect and development geoscientists in the Exploration and Development companies of ExxonMobil. We thank John Ardill, Dana Butters, Tim Fahrer, E-Chien Foo, Matt Grove, Randy Kissling, Dave Mason, Lisa McBee, Jose Sequeira, Mark Rosin, and Bill Tate for their insight to the geology of Bur1 slope-channel system. Reviews by Laurence Droz (University of Brest) and Carlos Pirmez (Shell Oil) greatly improved the manuscript. The views expressed in this study are those of the authors and ExxonMobil and do not necessarily reflect those of the concessionaire or the Block 15 contractor group.

300 / Porter et al.

FIGURE 17. Depositional environments interpreted from the vertical succession of lithofacies association in the Bur1 channel system. Each of the channel complexes in the Bur1 shows a progression of mixed debrites, turbidites, and slumps that are overlain by turbidite-dominated facies associations of well-amalgamated gravel- and sand-prone traction deposits. Because of the admixture of muddy debrites in the channel complexes, log response would suggest a nonamalgamated, disconnected prediction of channel-fill connectivity. Erosion associated with the gravelly lithofacies produces lateral and vertical amalgamation of the channel complexes, and dynamic well performance demonstrates better connectivity than would be inferred by log response. Gamma-ray and Vshale log response of stacked facies associations is on the leftmost track with core photo annotations; resistivity, porosity, density, and water-saturation curves are shown on the right log tracks. Depth in meters.

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 301

FIGURE 18. Vertical lithofacies succession and stratal hierarchy in the Bur1 slope-channel system. Detailed facies description of Bur1 cores shows the relationship of bed and bedset elements to larger stratal elements in a deep-water hierarchy. In this example, yellow bedsets are gravel- or sand-prone lithofacies deposited by high- and low-concentration turbidity currents. Brown-colored units are sandy and muddy sand debrites. Gray-colored lithologies are an admixture of muddy turbidites, debrites, and hemipelagic shales. The vertical facies log shows the axial expression of three stacked channel complex sets in this part of the Bur1 slope-channel system. Leftmost well-log curves are gamma ray and Vshale ; shallow and deep resistivity curves are displayed on the right track. The length of the core description is 110 m (360 ft).

302 / Porter et al.

FIGURE 19. Seismic expression of gravel-prone units in the Bur1 slope-channel system. Four seismic horizon-keyed extractions of the high-impedance, lower parts of Bur1 channel complexes show the geometry and dimensions of gravel-rich units. (A) Scrollwork pattern of gravel-rich units deposited within sinuous channels of the Bur1 system. These units represent the lower half of the composite channel complex fills and are overlain by sand- and mixed sand- and mud-prone lithofacies. (B) Stratigraphically younger channel complex shows smaller proportion of gravel units in an overall more sand-prone channel complex set. Scrollwork pattern is developed from laterally migrating channels, but gravelly units are more isolated than in the channel complex set illustrated in (A). (C) Strongly confined part of the Bur1 system shows the abundance of gravel thalweg units in green. Well penetrations that establish intervening areas are sandstones deposited by high-concentration turbidity currents. (D) Volume seismic sculpt shows relationship of gravels with overlying sandstones in a single channel complex set of the Bur1. Well penetrations show blue units as highimpedance gravels and the red units as lower impedance sandstones.

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 303

FIGURE 20. Depositional model for the Bur1 slope-channel system. The vertical and lateral relationships of turbidite and debrite deposits are represented in this cross-sectional depositional model of the Bur1 confined slope-channel system. Core-calibrated logs define the axial facies succession. Debrite and gravelly channel fills define the deepest parts of erosionally based channel sets, marked by light-blue lines. The overall vertical succession shows most of the gravelprone lithofacies in the lower half of the channel system, and the upper fill succession contains more mixed sandstone, mudstones, and muddy debrites. Seismic facies mapping is used to place depositional elements in the off-axis and margin areas of the channel system. These lateral facies are thin-bedded mixed sandstones and mudstones deposited in channel-related, stacked levee margins.

304 / Porter et al.

REFERENCES CITED Angevine, C. L., P. L. Heller, and C. Paola, 1990, Quantitative sedimentary basin modeling: AAPG Continuing Education Course Note Series 32, 133 p. Ardill, J. A., T. C. Huang, and O. McLaughlin, 2002, The stratigraphy of the Oligocene to Miocene Malembo formation of the Lower Congo basin, offshore Angola (abs.): AAPG Annual Meeting Expanded Abstracts, p. 9. Arnott, R. W. C., and B. M. Hand, 1989, Bedforms, primary structures and grain fabric in the presence of suspended sediment rain: Journal of Sedimentary Petrology, v. 59, p. 1062 – 1069. Babonneau, N., B. Savoye, M. Cremer, and B. Klein, 2002, Morphology and architecture of the present canyon and channel system of the Zaire deep-sea fan: Marine and Petroleum Geology, v. 19, p. 445 – 467. Beaubouef, R. T., 2004, Deep-water leveed-channel complexes of the Cerro Toro Formation, Upper Cretaceous, southern Chile: AAPG Bulletin, v. 88, p. 1471 – 1500. Beaubouef, R. T., T. R. Garfield, and F. J. Goulding, 1998, Seismic stratigraphy of depositional sequences: High resolution images from a passive margin slope setting, offshore west Africa (abs.): AAPG Bulletin, v. 82, p. 1980. Bouchet, R., B. Levallois, G. Mfonfu, and J.-F. Authier, 2005, Optimizing development of Angola’s Girassol field: World Oil, v. 226, no. 3, p. 45 – 48. Bouma, A. H., 1962, Sedimentology of some flysch deposits: Amsterdam, Elsevier, 162 p. Brice, S. E., M. D. Cochran, G. Pardo, and A. D. Edwards, 1982, Tectonics and sedimentation of the south Atlantic rift sequence, Cabinda, Angola, in J. S. Watkins and C. L. Drake, eds., Studies in continental margin geology: AAPG Memoir 34, p. 518. Burke, K., 1996, The African plate: South African Journal of Geology, v. 99, p. 341 – 409. Campion, K. M., A. R. Sprague, D. Mohrig, R. W. Lovell, P. A. Drzewiecki, M. D. Sullivan, J. A. Ardill, G. N. Jensen, and D. K. Sickafoose, 2000, Outcrop expression of confined channel complexes: Gulf Coast Section SEPM Foundation 20th Annual Research Conference, Houston, Texas, p. 127 – 150. Droz, L., T. Marsset, H. Ondreas, M. Lopez, B. Savoye, and F.-L. Spy-Anderson, 2003, Architecture of an active mud-rich turbidite system: The Zaire Fan (Congo – Angola margin southeast Atlantic): Results from ZaiAngo 1 and 2 cruises, AAPG Bulletin, v. 87, p. 1145 – 1168. Duval, B., C. Cramez, and M. P. A. Jackson, 1992, Raft tectonics in the Kwanza Basin, Angola: Marine and Petroleum Geology, v. 9, p. 389 – 404. Emery, K. O., and E. Uchupi, 1984, The geology of the Atlantic Ocean: New York, Springer-Verlag, 1050 p. Garfield, T. R., D. C. Jennette, F. J. Goulding, and D. K. Sickafoose, 1998, An integrated approach to deepwater reservoir prediction (abs.): AAPG Bulletin, v. 83, p. 1314.

Gottschalk, R. R., 2002, The Lower Congo basin, deepwater Congo and Angola: A kinematically linked extensional/contractional system (abs.): AAPG Bulletin, v. 86, p. 66. Gottschalk, R. R., A. V. Anderson, J. D. Walker, and J. C. Da Silva, 2004, Modes of contractional salt tectonics in Angola Block 33, Lower Congo basin, west Africa, in 24th Annual Gulf Coast Section SEPM Research Conference, Salt-sediment Interactions and Hydrocarbon Prospectivity: Concepts, Applications, and Case Studies for the 21st Century, Houston, Texas, December 5 – 8, 2004, 30 p. Goulding, F. J., T. R. Garfield, K. W. Rudolph, G. N. Jensen, and R. T. Beaubouef, 2000, Seismic/sequence stratigraphy of deep-water reservoirs: 1. Seismic facies recognition criteria: Past experience and new observations (abs.): AAPG Annual Meeting Program, v. 9, p. A56. Hartman, D. A., W. A. Swanson, P. R. Smith, F. J. Goulding, and C. A. Kelly, 1998, Structural development of the continental margin of Congo and northern Angola (abs.): AAPG Bulletin, v. 82, p. 1923. Iverson, R. M., 1997, The physics of debris flows: Reviews in Geophysics, v. 35, p. 245 – 296. Kolla, V., P. Bourges, J. M. Urruty, and P. Safa, 2001, Evolution of deep-water Tertiary sinuous channels offshore Angola (west Africa) and implications for reservoir architecture: AAPG Bulletin, v. 85, p. 1371 – 1405. Larson, R. L., and J. W. Ladd, 1973, Evidence from magnetic lineations for the opening of the South Atlantic in the Early Cretaceous: Nature, v. 246, p. 209 – 212. Lowe, D. R., 1982, Sediment gravity flows: II. Depositional models with special reference to the deposits of highconcentration turbidity currents: Journal of Sedimentary Petrology, v. 52, p. 279 – 297. Middleton, G. V., and M. A. Hampton, 1976, Subaqueous sediment transport and deposition by sediment gravity flows, in D. J. Stanley and D. J. P. Swift, eds., Marine sediment transport and environmental management: New York, John Wiley and Sons, p. 197 – 218. Pirmez, C., R. N. Hiscott, and J. D. Kronen Jr., 1997, Sandy turbidite successions at the base of channel-levee systems of the Amazon Fan revealed by FMS logs and cores: Unraveling the facies architecture of large submarine fans, in R. D. Flood, et al., eds., Proceedings of the Ocean Drilling Program, Scientific Results, v. 155, p. 7 – 33. Rabinowitz, P. D., and J. LaBreque, 1979, The Mesozoic South Atlantic Ocean and evolution of its continental margins: Journal of Geophysical Research, v. 84, p. 5973– 6002. Raposo, A. J. M., and M. A. Sykes, 1998, Exploration for deep-water reservoirs, offshore Angola (abs.): AAPG Bulletin, v. 82, p. 1956. Reeckmann, S. A., D. K. S. Wilkin, and J. Flannery, 2001, Kizomba, a deep-water giant field, Block 15, Angola, in M. T. Halbouty, ed., Giant oil and gas fields of the decade 1990 – 1999: AAPG Memoir 78, p. 227 – 236.

Lower Miocene Slope-channel System in Block 15, Offshore Angola / 305 Schollnberger, E., and P. R. Vail, 1999, Seismic stratigraphy of the Lower Congo, Kwanza, and Benguela basins, offshore Angola, Africa (abs.): AAPG Bulletin, v. 83, p. 1338. Scrutton, R. A., and R. V. Dingle, 1976, Observations on the processes of sedimentary basin formation at the margin of southern Africa: Tectonophysics, v. 36, p. 143– 156. Shanmugam, G., 1996, High density turbidity currents: Are they sandy debris flows?: Journal of Sedimentary Research, v. 66, p. 2 – 10. Sprague, A. R. et al., 2002, The physical stratigraphy of deep-water strata: A hierarchical approach to the analysis and genetically related stratigraphic elements for improved reservoir prediction (abs.): AAPG Annual Meeting Expanded Abstracts, p. 167. Sprague, A. R. et al., 2005, Integrated slope channel depositional models: The key to successful prediction of

reservoir presence and quality in offshore west Africa: E-Exitep Proceedings 2005, Veracruz, Mexico, 13 p. Sykes, M. A., D. Mohrig, C. Rossen, and T. R. Garfield, 1998, Lithofacies associations within complex slope channel reservoirs: Debrites and turbidites (abs.): AAPG Bulletin, v. 82, p. 1973. Temple, F., and O. Broucke, 2004, Sedimentological models of the Oligocene and Miocene Malembo formation in offshore Angola (Lower Congo basin) (abs.): 1st Nigerian Association of Petroleum Explorationists– American Association of Petroleum Geologists West Africa Deepwater Conference, Abuja, Nigeria, p. A46. Winn, R. D., and R. H. Dott Jr., 1977, Large-scale traction produced structures in deep-water fan-channel conglomerates in southern Chile, Geology, v. 5, p. 41 – 44. Winn, R. D., and R. H. Dott Jr., 1979, Deep-water fanchannel conglomerates of Late Cretaceous age, southern Chile: Sedimentology, v. 26, p. 203 – 228.

9

Dubois, M. K., A. P. Byrnes, G. C. Bohling, and J. H. Doveton, 2006, Multiscale geologic and petrophysical modeling of the giant Hugoton gas field (Permian), Kansas and Oklahoma, U.S.A., in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 307 – 353.

Multiscale Geologic and Petrophysical Modeling of the Giant Hugoton Gas Field (Permian), Kansas and Oklahoma, U.S.A. Martin K. Dubois, Alan P. Byrnes, Geoffrey C. Bohling, and John H. Doveton Kansas Geological Survey, University of Kansas, Lawrence, Kansas, U.S.A.

ABSTRACT

R

eservoir characterization and modeling from pore to field scale of the Hugoton field (central United States) provide a comprehensive view of a mature giant Permian gas system and aid in defining original gas in place and the nature and distribution of gas saturation and reservoir properties. Both the knowledge gained and the techniques and workflow employed have implications for understanding and modeling reservoir systems worldwide that have similar geologic age and reservoir architecture (e.g., Gwahar and North fields, Persian Gulf). The Kansas–Oklahoma part of the field has yielded 34 tcf (963 billion m3) gas throughout a 70-yr period from more than 12,000 wells. Most remaining gas is in lower permeability pay zones of the 557-ft (170-m)-thick, differentially depleted, layered reservoir system. The main pay zones represent 13 shoaling-upward, fourth-order marinecontinental cycles comprising thin-bedded (6.6–33-ft; 2–10-m), marine carbonate mudstone to grainstone and siltstones to very fine sandstones and have remarkable lateral continuity. The pay zones are separated by eolian and/or sabkha red beds having low reservoir quality. Petrophysical properties vary among 11 major lithofacies classes. Neural network procedures, stochastic modeling, and automation facilitated building a detailed full-field three-dimensional (3-D) 108-million-cell cellular reservoir model of the 10,000-mi2 (26,000-km2) area using a four-step workflow: (1) define lithofacies in core and correlate to electric log curves (training set); (2) train a neural network and predict lithofacies at noncored wells; (3) populate a 3-D cellular model with lithofacies using stochastic methods; and (4) populate model with lithofacies-specific petrophysical properties and fluid saturations.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215881M883274

307

308 / Dubois et al.

INTRODUCTION The focus of this chapter is the definition and integration of core-defined lithofacies, core-derived petrophysical properties, wire-line-log response, and estimation of reservoir properties to characterize a giant reservoir system at the core, well, and field scale. Central to the effort is the use of core-defined lithofacies to train a neural network to predict lithofacies at wells without core. We discuss each step of the workflow, many aspects of which could be applied in other settings. The results and summary are an early view of work in progress that is part of an ongoing collaborative project. The importance of the Hugoton field study extends beyond the borders of Kansas and Oklahoma. Both the knowledge gained and the techniques employed have implications for understanding and modeling reservoir systems worldwide that have similar geologic age, reservoir architecture, production characteristics, problems in determining water saturation, large data sets, split ownership, or maturity. The fullfield model of the 10,000-mi2 (26,000-km2) reservoir area provides a detailed three-dimensional (3-D) view of 13 shoaling-upward cycles vertically stacked in a low-relief shelf setting. The nature of the model and its construction provides a good analog for similar thin, stacked-cycles reservoir systems, including the Aneth field in the Paradox basin (Weber et al., 1994; Grammer et al., 1996), fields in the prolific Permian basin of west Texas (Dutton et al., 2005), and the Khuff Formation in Gwahar and North fields in the Persian Gulf (McGillivray and Husseini, 1992; Konnert et al., 2001). Fine-scale cellular models are particularly important for modeling thin-layered, differentially depleted reservoir systems, and methods used in building the model demonstrate the construction of a cellular petrophysical model for a giant field. The study also demonstrates the benefits of pooling proprietary geologic and engineering data in settings having split ownership (Sorenson, 2005). As the world’s fields mature, high-resolution modeling at the full-field scale in data-rich environments will become increasingly important, and we present a largescale example for developing such models.

Background The combined Kansas Hugoton and Panoma, Texas Hugoton, and West Panhandle fields, with an estimated ultimate recovery of 75 tcf (2.1 trillion m3) gas (Sorenson, 2005) represent the largest gas field in

North America. Covering southwest Kansas and parts of the Oklahoma and Texas panhandles, these fields are situated in the Hugoton embayment of the Anadarko basin (Figure 1). Since its discovery in 1922 and development in the 1950s, 34 tcf gas (963 billion m3) have been produced from greater than 12,000 wells across 6200 mi2 (16,000 km3) in the Kansas and Oklahoma part of the Hugoton field (Figure 2). Unless otherwise noted, the term ‘‘Hugoton’’ in this chapter combines the Hugoton (Kansas), Panoma (Kansas and Oklahoma), and the Guymon-Hugoton (Oklahoma) fields. Production is from the Lower Permian Chase Group and Council Grove Group (Figure 3). In most areas inside the Panoma boundary, the gas column is continuous between the two stratigraphic intervals (Pippin, 1970; Parhman and Campbell, 1993) and reaches a maximum thickness of 500 ft (150 m) in the west-central part of the study area. One exception may be in a relatively small part of the field near the western margin in Morton County, Kansas, that is described by Olson et al. (1997) as being compartmentalized by faults. In Oklahoma, production outlined as ‘‘other Council Grove’’ (Figure 1) is from intervals in the Council Grove that are as much as 300 ft (100 m) below the lowest perforations in the Chase. The reservoir is shallow, with depth to the top of the Chase ranging from 2100 to 2800 ft (640 to 850 m) and lower and upper productive limits, referenced to sea level, of approximately +100 ft (+30 m) on the east and +1250 ft (+380 m) on the western updip margin, respectively. Original wellhead shut-in pressure in Kansas was 437 psi (3013 kPa) (Hemsell, 1939), significantly less than half of a seawater gradient, and similar, anomalous initial pressures were recorded in Oklahoma (Sorenson, 2005). Average 72-h wellhead shut-in pressure in Kansas in 2003 was 32 psi (221 kPa). Annual production in 2004 was 265 bcf (7.5 billion m3). Early completions in the Chase were commonly open hole with a slotted liner followed by a large acid treatment. After 1960, typical completions commonly involve casing, perforation, and acidizing as many as six zones separately, followed by a large hydraulic sand fracture treatment, sometimes exceeding 200,000 lb (91,000 kg) of sand, to the entire perforated interval (Hecker et al., 1995). Although much has been published on the Hugoton throughout the 70-yr life of the field, most of the studies were broad in scope (Hemsell, 1939; Mason, 1968; Pippin, 1970). Sorenson’s (2005) recent article that presents a paleostructural and pressure history for the reservoir system stretching from the Texas Panhandle to west-central Kansas provides a good recent

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 309

FIGURE 1. Regulatory boundaries for Permian (Wolfcampian) gas and oil fields, Kansas, Oklahoma, and Texas. Wolfcampian structure in feet, after Pippin (1970) and Sorenson (2005).

overview of the field history and prior work. Detailed studies involving reservoir characterization have been limited geographically and stratigraphically. For example, Seimers and Ahr (1990) investigated the Chase in the Oklahoma Panhandle; Olson et al. (1997) studied the Kansas Chase; and Heyer (1999) focused on the Council Grove in a small area of the Okla-

homa Panhandle. Following the Kansas Corporation Commission proration order permitting a second well in each unit, several studies on reservoir characterization (Seimers and Ahr, 1990; Caldwell, 1991; Olson et al., 1997) and reservoir simulation (Fetkovitch et al., 1994; Oberst et al., 1994) were published. Past studies by industry have been generally confined to areas where they have assets and data. This chapter provides details of the most comprehensive reservoir characterization effort to date, both geographically and stratigraphically (the entire reservoir system, Chase and Council Grove groups). Work related to this study is part of a collaborative, multidisciplinary study of the Hugoton field, supported by 10 industry partners having gas resources in southwest Kansas and the Oklahoma Panhandle (Dubois et al., 2005). The primary purpose of the 2-yr study is to develop a comprehensive, fieldwide geologic model that can be used to quantify and locate remaining gas for improved reservoir management. No field-scale model exists primarily (1) because of the immense size of the field and (2) because the most critical petrophysical data were proprietary. The current project was facilitated by the creation of a cooperative geologic and engineering data set amassed from participating companies, which allowed a comprehensive global view of the entire reservoir system.

310 / Dubois et al.

is critical. Because only a relatively small number of wells were cored and the reservoir volume is immense, a methodology was developed for predicting lithofacies from wire-line-log response at wells not cored and then between wells.

RESERVOIR GEOLOGY Regional Geology

FIGURE 2. Gas production from the Wolfcampian (Hugoton and Panoma fields) in the Kansas part of the study area through 2004. The Oklahoma part of the study area produced 7 tcf (198 billion m3) Wolfcampian gas in the same period that 27 tcf (765 billion m3) was produced in Kansas. The spike in production beginning in the early 1980s was caused by infill drilling the Hugoton field in Kansas.

Although the field is very mature, individual companies possess excellent modern wire-line-log, core, and engineering data in their asset areas because of drilling for deeper production and the infill drilling program that took place in the late 1980s and early 1990s.

The Hugoton field lies on the west side of the Hugoton embayment of the Anadarko basin and is bounded to the northwest by the Las Animas arch and to the northeast by the central Kansas uplift. The Anadarko basin is an asymmetric foreland basin associated with the early Pennsylvanian Ouchita–Marathon orogeny, and the Hugoton embayment and the rest of the Kansas shelf form the flatter side of the asymmetry (Figure 4). The Anadarko basin was initiated and had its greatest subsidence in Pennsylvanian-Morrowan time, with subsidence rates decreasing through the Permian. The basin was nearly filled by the end of the Wolfcampian when the Anadarko basin was covered by shelf carbonates (Kluth and Coney, 1981; Rascoe and Adler, 1983; Kluth, 1986; Perry, 1989). Marine carbonate reservoirs thin toward the updip margin and many pinch out at, or just west of, the field margin, particularly in the Council Grove.

Problem and Approach A principal goal of the project is to quantify the nature and distribution of gas saturation and reservoir properties and original and remaining gas in place. Determining the gas in place and its distribution is hampered by three significant obstacles: (1) accurate determination of water saturations using conventional wire-line logs is not possible because of deep invasion by filtrate for typical drilling programs (Olson et al., 1997; George et al., 2004); (2) wellhead shut-in pressures are strongly influenced by high-permeability interval properties and do not accurately represent all interval pressures; and (3) the reservoir is layered, differentially depleted, and pressure data for individual layers are minimal. To define the remaining gas in place, we used an integrated approach using core, core-derived lithofacies-specific petrophysical relationships, and engineering data. Because petrophysical property relationships (e.g., permeability-porosity, capillary pressure) differ among lithofacies, the construction of a geomodel with appropriate lithofacies

FIGURE 3. Stratigraphic column, Hugoton field area, with the names of gas fields in Kansas and Oklahoma adjacent to the intervals from which they produce (compiled from Zeller, 1968; Pippin, 1970; Baars, 1994; D. P. Merriam, 2006, personal communication). The combined Hugoton and Panoma fields in Kansas and the Guymon-Hugoton field in Oklahoma are lumped as Hugoton in this chapter.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 311

FIGURE 4. Distribution of major lithofacies in the mid-continent during the late Wolfcampian (modified from Rascoe, 1968; Rascoe and Adler, 1983; Sorenson, 2005; used with permission from the Rocky Mountain Association of Geologists and AAPG). Approximate paleolatitude was 38N (Scotese, 2004).

Red continental rocks, primarily very fine to coarse siltstones, are thickest at the margin and thin basinward across the shelf (Figure 5). The red beds have been thought by many to be the lateral seal that, when accompanied by a Leonardian-age evaporite top seal, created a giant stratigraphic trap (Garlough and Taylor, 1941; Mason, 1968; Pippin, 1970; Parhman and Campbell, 1993). However, laterally continuous, high-porosity and high-permeability, marine, and continental sandstones are common at the updip margin in the northwest part of the field. These rocks are gas productive inside the field boundaries and water saturated outside the field despite being in a higher structural position and without evidence of a physical barrier. These conditions argue against these rocks being a lateral seal and suggest that mechanisms other than lithofacies change alone are responsible for trapping (Dubois and Goldstein, 2005). Present-day structure of Wolfcampian-age rocks was strongly influenced by a Laramide-age eastward tilt (Figure 6), whereas the Wolfcampian isopach (Figure 7) better reflects the shelf geometry. From the west field margin, the Wolfcampian strata thicken

basinward at a rate of approximately 1.2 ft/mi (0.24 m/km) to location on the shelf where the rate of thickening increases by a factor of 10. The axis of thickening is coincident with an area of present-day steep dip and may mark a shelf margin or the axis of a steepened slope. It is also nearly coincident with the edge of a Virgillian-age starved basin and the transition from marine carbonate to marine shale (Rascoe, 1968; Rascoe and Adler, 1983). Dubois and Goldstein (2005) estimated the maximum relief across the Kansas part of the shelf during Council Grove deposition to have been 100 ft (30 m), with a slope of approximately 1 ft/mi (0.2 m/km). Notable is the absence on the Hugoton shelf of dark, fissile shale, a common deep-water lithofacies in the Wolfcampian in outcrop in eastern Kansas and northeast Oklahoma (Mazzullo et al., 1995; Boardman and Nestell, 2000), suggesting that water depths on the Hugoton shelf were less than those at the present-day outcrop 300 mi (480 km) to the east. The closest equivalent to the typical deep-water lithofacies in Hugoton core are dark, marine siltstones found near the base of the marine carbonate intervals in four cycles, the Grenola (C_LM), Funston (A1_LM), Wreford, and Ft. Riley. For this chapter, we will refer to most of the extremely gently sloping area as shelf and the area of steeper dip and stratigraphic thickening as the shelf margin.

Reservoir Lithofacies The Hugoton in Kansas and Oklahoma produces gas from 13 fourth-order marine-continental (carbonatesiliciclastic) sedimentary cycles (Figure 8), six in the Chase and seven in the Council Grove, reflecting rapid glacioeustatic sea level fluctuations (Olson et al., 1997; Heyer, 1999; Boardman et al., 2000; Olszewski and Patzkowsky, 2003). The marine and continental lithologic units are laterally continuous and can

are those predicted by neural network models (small well symbols) or from core (large symbols) and are interpolated in Geoplus Petra between wells. The lumped lithofacies include continental sandstone (L0), continental siltstone (L1-2), mud-supported carbonate and marine siltstone (L3-5), grain-supported carbonate and dolomite (L6-9), and marine sandstone (L10). The Council Grove is thinnest at a midfield (midshelf) position. Log curves are gamma ray (left) and corrected porosity (right).

TM

FIGURE 5. Stratigraphic cross section of the Chase and Council Grove groups with the top of the Council Grove as the datum. At the wells, lumped lithofacies

312 / Dubois et al.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 313

FIGURE 6. (A) The present-day structure of the top of the Wolfcampian reservoir (top of Chase) is mostly a function of eastward tilt during the Laramide orogeny. Note the shelf margin or area of steepened slope at the southeast margin of the Hugoton field outline. (B) Three-dimensional view of the same area. The present-day structure on the top of the Chase and base of Council Grove.

be traced across the shallow shelf to the outcrop in eastern Kansas. The main pay zones are 13 thin (mean thickness 6– 70 ft [2 –21 m]) marine, mainly carbonate intervals, deposited during sea level highstands. Pay zones are separated by continental, mainly siltstone (red-bed) intervals (mean thickness 6–25 ft [2 – 8 m]) deposited during sea level lowstands, when most of the shelf was exposed. The siltstones generally have poor reservoir quality and vertically isolated, or restricted communication among, the 13 pay intervals (Seimers and Ahr, 1990; Oberst et al., 1994; Ryan et al., 1994; Olson et al., 1997). The principal factor in determining the reservoir storage and flow capacity (hydrocarbon pore volume and permeability) of Hugoton reservoir rock is primary depositional texture. Diagenesis, both early and after burial, including leaching of grains and cements and early and late dolomitization, are important factors in enhancing or reducing porosity (Seimers and Ahr, 1990; Olson et al., 1997; Luczaj and Goldstein, 2000). However, the dominant reservoir rocks are marine carbonate with grain-supported textures and, to a lesser

extent, siliciclastic sandstone (Seimers and Ahr, 1990; Caldwell, 1991; Olson et al., 1997; Heyer, 1999; Dubois et al., 2003).

Depositional Model Climate, shelf geometry, and glacially forced sea level changes all influenced sediment supply, depositional patterns, accommodation, and stabilization of both the marine and continental sediments. The Hugoton shelf was near the paleoequator (Figure 4), and monsoonal climate conditions are likely to have prevailed at the annual scale (Parrish and Peterson, 1988). Generally, arid conditions accompanied glacially induced sea level lowstands with more humid conditions and high sea level present during interglacial periods (Rankey, 1997; Soreghan, 2002). Prevailing winds are thought to have been from the present-day west during winter and east during summer (Parrish and Peterson, 1988). Extremely low relief enabled rapid migration of the shoreline position and changes in shelf hydrodynamics, with only

314 / Dubois et al.

FIGURE 7. Isopach of the Wolfcampian reservoir (top of Chase Group to base of Grenola Limestone, Council Grove Group). Wolfcampian rate of thickening increases by a factor of 10 at the shelf margin.

minimal absolute changes in sea level (as little as 100 ft [30 m]). These conditions set the stage for the vertical succession of lithofacies repeated from one sedimentary cycle to the next, as well as the remarkable lateral continuity of thin lithofacies units in each cycle. The cyclical nature of the Council Grove and Chase is widely recognized (Seimers and Ahr, 1990; Caldwell, 1991; Puckette et al., 1995; Olson et al., 1997). Vertical succession of lithofacies in a shoalingupward pattern in both the Council Grove and Chase (Figure 9) is a result of depositional environments changing across the shelf in response to rapid sea level fluctuation. Differences in the style (symmetry) and pattern (lithofacies) among the Chase cycles recognized by Olson et al. (1997) are confirmed, for the most part, in our study. Exceptions are that we do see fine-grained sandstone of marginal-marine origin at both the top and base of the Towanda and

Winfield near the updip margin of the field, although it is more common to find the situation as they depicted elsewhere (sandstone at the base of the Towanda and top of the Winfield). Although similar in many respects, the Council Grove cycles are typically more asymmetric than the Chase cycles and tend to have better developed, thin, packstone-grainstone lithofacies at the base of the marine half-cycle. Figure 10 presents a composite of the vertical distribution of lithofacies from model node wells that also help illustrate the difference in symmetry between a Council Grove and Chase cycle. Intervals were defined within a simple cyclic framework instead of a sequence-stratigraphic framework. Existing formation or member tops that are halfcycle boundaries between marine and continental intervals represent a sequence boundary and flooding surface. Because the transgressive systems tract (flooding surface to maximum flooding) is relatively thin and consistent in most of the cycles, little is gained by correlation of an additional surface for sequencestratigraphic classification. Idealized depositional models (Figure 11) for the Council Grove Chase and Chase are generally similar, but differences exist because of gradual changes in climate, ambient sea level position, and sea level fluctuation rate. Differences may be related to a shift from more icehouse to more greenhouse conditions in the Permian (Parrish, 1995; Olszewski and Patzkowsky, 2003). The entire Hugoton shelf was above sea level during maximum lowstand for all studied Council Grove cycles, and continental red-bed siliciclastics accumulated and were stabilized by vegetation and built relief preferentially near the field’s west updip margin (Dubois and Goldstein, 2005). Accommodation for the carbonate sediments of the overlying marine half-cycle was reduced, leading to nondeposition, or pinch-outs, of several marine intervals in the Council Grove at that position. At the end of each lowstand, a relatively rapid sea level rise resulted in deposition of a thin (1 –4-ft; 0.3– 1.2-m) transgressive carbonate-siliciclastic interval at the base of each marine half-cycle. Only in the Funston (A1-LM) and Neva (C-LM) cycles are welldeveloped marine siliciclastics (shaly siltstone) deposited during maximum flooding. After maximum flooding, shallowing, accompanied by conditions that fostered increased carbonate production, resulted in a shoaling-upward lithofacies stacking pattern. A fall in absolute sea level caused progradation of broad facies belts (e.g., carbonate sand shoals), resulting in laterally extensive lithofacies bodies. With continued

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 315

sea level fall, continental sabkha, coastal-plain, and savannah environments followed the retreating shoreline and covered the carbonate surface. Evidence for prolonged direct subaerial exposure and erosion of the carbonate surface is absent in all seven Council Grove cycles in the nine cores examined. Instead of calcretes, microkarst, erosion, or other indicators of prolonged exposure in the upper part of the marine carbonate, there is a vertical succession of lithofacies that suggests continuous sedimentation that

accompanied a sea level fall and withdrawal: subtidal carbonate, tidal-flat carbonate, red siltstone and muddy siltstone with anhydrite (sabkha), and finally, red siltstone with paleosols (coastal plain or savannah). Although Chase deposition was similarly influenced by absolute sea level, it differs from the Council Grove in significant ways. Specifically, during Chase lowstand, the lateral extent of subaerial exposure on the shelf was generally more limited, and in some continental intervals, tidal-flat siltstone and very fine-grained sandstone are prevalent, particularly in positions lower on the shelf. Fine-grained eolian sandstone present in nearly all continental half-cycles in the Council Grove is nearly absent in the Chase. Marine transgressions in the Chase generally extended farther landward than they did during Council Grove deposition, with marine sediments extending beyond the updip margin of the field in all six cycles, whereas they pinch out in four of the seven Council Grove cycles studied. During the maximum highstand and the subsequent fall in sea level, carbonate sand shoals of the Chase tend to be coarser grained, and constituents include bioclasts and, occasionally, ooids, instead of oncoids and peloids (in the Council Grove), indicating more open-marine conditions. Another significant difference between the Chase and Council Grove is the presence of fine-grained sandstone deposited in tidal-flat and marginal-marine

FIGURE 8. Formation- and member-level stratigraphy correlated to the wire-line well log in the A-1 Flower well, Stevens County, Kansas. Commonly used formation and member letter-number combinations are shown for the Council Grove. Twelve of the thirteen marine-continental (carbonate-siliciclastic) sedimentary cycles that are gas productive are shown (Grenola Limestone, C_LM is not logged). Stratigraphic names that include ‘‘limestone’’ are marine half-cycles when combined with an adjacent continental half-cycle, intervals with stratigraphic names that include ‘‘shale’’ form a complete cycle. Color-coded lithofacies are derived from core. Three were deposited in a continental setting (L0 = sandstone; L1 = coarse siltstone; and L2 = shaly siltstone) and eight in a marine environment (L3 = siltstone; L4 = carbonate mudstone; L5 = wackestone; L6 = very fine-crystalline dolomite; L7 = packstone; L8 = grainstone and phylloid algal bafflestone; L9 = fine-medium-crystalline moldic dolomite; and L10 = sandstone). L0 is absent in this well. Wire-line-log abbreviations are caliper (CALI), gamma ray (GR), corrected porosity (PHI_GM3), photoelectric effect (PEF), density porosity (DPHI), neutron porosity (NPHI), core permeability (K_MAX), and core porosity (CORE_POR). Logged interval is 520 ft (160 m).

316 / Dubois et al.

FIGURE 9. Idealized Chase and Council Grove Groups cycles. Chase cycles are from Olson et al. (1997), used with permission from the AAPG, and our Council Grove cycles are similarly formatted. One exception is that we extend the cycle and approximate sea level curve through the continental half-cycle based on earlier work (Dubois and Goldstein, 2005). Five cycle types distinguished the basis of lithofacies stacking pattern and inferred relative sea level curve.

with normal-marine assemblages are absent in most of the Council Grove cycles at or near the west margin of the field. Both the Chase and Council Grove cycles exhibit gradual changes through time that may be related to third-order cyclicity (Boardman et al., 2000) and the overall shift from icehouse to greenhouse conditions that began in late Pennsylvanian and continued until the end of glacial conditions in the Permian (Parrish, 1995). Most likely, as a consequence of the climate change trend, Chase marine carbonate intervals tend to be three to five times thicker than their Council Grove counterparts, at least in the Crouse through Cottonwood interval (B1-LM–B5-LM) on the Hugoton shelf. settings at either the top, base, or top and base of all cycles above the Fort Riley in the northwest part of the field (Winters et al., 2005). In all Chase or Council Grove intervals, the nature of the lithofacies present in a succession are a function of the position on the shelf: the farther west and updip, the greater the volume of siliciclastics, whether in the marine or continental setting. Marine carbonate tends to be muddier to the west and northwest, and grain-supported carbonate tends to be finer toward the west, with the dominant grains being hardened pellets (round, very fine grained, micritic) and peloids (subrounded, fine grained, micritic) instead of oncoids, bioclasts, or ooids (found in upper Chase only). Marine environments become more restricted in a westerly direction, and rocks

Layered Reservoir and Differential Depletion The Hugoton and Panoma fields appear to be one large reservoir system that may have filled and changed pressure in stages (Sorenson, 2005). However, the six pay zones in the Chase are being depleted by production at different rates, as indicated by different interval pressure tests and reservoir simulation (Fetkovich et al., 1994; Oberst et al., 1994; Ryan et al., 1994). Table 1 lists zonal pressures showing differential depletion in both the Chase and Council Grove. Composite tests covering multiple zones exhibit pressures that primarily represent the produced zone(s) having the lowest pressure and the highest permeability as measured from core and drillstem test (DST) analysis. The more recent of the wells tested (2005) showed pressures as low as 19 psi (131 kPa) in

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 317

FIGURE 10. Vertical histograms showing the average relative distribution of lithofacies in two Wolfcampian marine half-cycles from wells having predicted lithofacies data (node wells). Data for the Crouse (B1_LM), Council Grove Group, are from 1146 wells, and for the Krider, Chase Group, data are from 1069 wells. Histograms and probabilities demonstrate the difference in symmetry in vertical lithofacies distribution between the Chase and Council Grove. Probability distributions were used to condition lithofacies modeling by sequential indicator simulation between node wells. Layer annotations refer to layering in the halfcycle respective models (discussed later). Abbreviations are fine-grained (Fg), fine crystalline (Fxln), and fine to medium crystalline (F-mxln).

the main pay zones, the Herington and Krider halfcycles. This explains why comingled wellhead shutin pressures (32 psi [220 kPa] field average in Kansas) do not provide the necessary data for determination of overall depletion.

STATIC MODEL WORKFLOW Although many details exist, the workflow for developing the Hugoton geomodel can be described simply in four steps: (1) define interval tops from logs, lithofacies in core, petrophysical properties for lithofacies, and accurate log correction algorithms to obtain a log database; (2) train a neural network and predict lithofacies at node wells; (3) populate a 3-D cellular model with lithofacies and porosity;

using stochastic methods and (4) populate model with lithofacies-specific petrophysical properties and fluid saturations. Figure 12 provides a more comprehensive overview. The following is a discussion of each of the steps for integrating core-defined lithofacies, core-derived petrophysical properties, wireline-log response, and prediction of reservoir properties to characterize the reservoir system at the core and well scale and at the field scale.

Lithofacies Classification Because petrophysical properties vary among lithofacies, fundamental to reservoir characterization and the construction of a cellular reservoir model is the population of cells with lithofacies. Determining the number of lithofacies classes and the criteria for defining classes involved four standards: (1) maximum number of lithofacies recognizable by neural networks using petrophysical wire-line-log curves and other variables; (2) minimum number of lithofacies needed to accurately represent lithologic and petrophysical heterogeneity; (3) maximum distinction

318 / Dubois et al.

FIGURE 11. Idealized depositional models for the Council Grove (A) and Chase (B) showing the distribution of dominant lithofacies on the Hugoton shelf. Depicted are approximate depositional environments and associated lithofacies for typical Chase and Council Grove cycles at maximum sea level lowstand and during the falling sea level stage of the marine highstand.

Table 1. Pressures by zone for two relatively closely spaced wells.* Group

Zone

1994 Science Well DST-Sip psi (kPa)

Chase Group

Council Grove Group

Herington Krider Winfield SS Winfield LS Towanda U. Ft. Riley Florence Wreford A1_LM B1_LM B2_LM B3_LM B4_LM B5_LM

120 (830) 88 (610) 105 121 230 >400 398 372 400 350 131 368 215 160

(720) (830) (1590) (2750) (2740) (2570) (2760) (2410) (900) (2540) (1480) (1100)

2005 Replacement Well

Composite psi (kPa) 104 104 104 104 104 104 104 104 104 156 156 156 156 156 156

(720) (720) (720) (720) (720) (720) (720) (720) (720) (1080) (1080) (1080) (1080) (1080) (1080)

TM

XPT -SIP* psi (kPa) 19 21 30 141 217 165 192 265 219

(130) (145) (210) (970) (1500) (1140) (1320) (1830) (1510) nt nt nt 386 (2660) nt 348 (2400)

*The well drilled in 1994 was a research well (Flower A1, Figure 8), drilled with a foam fluid to limit filtrate invasion and formation damage. Pressures are 24-h shut-in pressures from drillstem tests. The well drilled in 2005 is located 6 mi (10 km) north of the earlier well, and pressures were recorded in the open hole by Schlumberger’s XPT TM tool, a repeat formation tester.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 319

FIGURE 12. Workflow for field-scale Hugoton model. Workflow can be divided into three broad tasks: (1) gather and qualify data; (2) process data to provide basic geomodel input files (Develop/Define/Properties/Algorithms); and (3) build the geomodel. The figure suggests the process is linear, whereas in reality, there are more feedback loops, multiple iterations at subtask level, and testing and validation at smaller scales.

320 / Dubois et al.

of core petrophysical properties among classes; and 4) the relative contribution of a lithofacies class to storage and flow. An optimal solution using these criteria resulted in 11 lithofacies distinguished on the basis of rock type (siliciclastic or carbonate), texture (Folk, 1954, grain size for siliciclastics; Dunham, 1962, classification for carbonates), and principal pore size (visual estimate). In classifying dolomite rocks, we did not consider depositional texture, but instead, the present texture and pore size that is primarily a function of crystal size and the presence or lack of molds of leached carbonate grains. Classes based on differences in core petrophysical properties coincided well with major lithofacies classes of rocks and have fairly distinctive wire-line-log response to petrophysical properties, the principal variables used for neural network prediction of lithofacies. Although defining more classes might have improved petrophysical prediction accuracy, the inability of neural networks to effectively recognize and distinguish finer lithofacies classes discouraged finer class distinctions (e.g., discriminating between fine-grained packstone and coarse-grained packstone). A quantitative, digital lithofacies description system (Table 2) was used in describing core at 0.5-ft (0.15-m) intervals. Three of the five factors illustrated in the tables were sufficient to segregate lithofacies classes, although other digits were considered initially in the process of determining class boundaries. For each interval, as much as 12 variables were recorded. Factors recorded in addition to those in Table 2 included the degree of consolidation and fracturing, subsidiary pore size, cement mineralogy, bedding, water depth, faunal assemblage, and color. Classifying lithofacies in a digital form facilitated changes in classification criteria and correlation of lithofacies with core and log petrophysical properties involved in the iterative process of determining optimal lithofacies class boundaries. This digital system is designed to provide a continuous numerical classification that corresponds to the continuum in lithological and petrophysical properties. In using this system, instead of a mnemonic system, error in classifying a given sample is generally only one class up or down, and therefore, the predicted property values are within a class step up or down from the true value. Once an object is numerically classified, mapping to alternate classification schemes can be performed automatically. Fourteen of approximately one hundred continuous cores were selected for lithofacies analysis on the basis of length (longest selected), geographic position (sampling distribution), and availability of core

analysis and wire-line-log data (Figure 13). In most cases, selected cores included either the entire Chase (five) or Council Grove (six) interval or covered both intervals (three). Examples of the 11 major lithofacies classes are shown in Figures 14 and 15. Common subclasses are represented, but the examples illustrated do not show the range exhibited by a lithofacies. For example, the marine carbonate packstone, packstonegrainstone class (L7), includes rocks having a variety of principal grain types and grain size but were deposited in a variety of environments (e.g., fine-grained pellets = tidal flat and lagoon; peloid and oncoids = restricted shelf and shoals; bioclasts = open shelf and shoals). Relative proportions of the 11 lithofacies in 4250 ft (1295 m) of core described in this study are shown in Figure 16. Continental lithofacies comprising finegrained sandstones (L0), coarse siltstones (L1), and fine or shaly siltstones (L2) represent 42% of the rock volume, whereas marine carbonates and marine siliciclastics represents 45 and 13%, respectively. Lithofacies with the greatest storage and flow capacity in marine rocks include L6 to L10, consistent with the principal reservoir lithofacies defined in previous studies (Siemers and Ahr, 1990; Olson et al., 1996; Heyer, 1999). These lithofacies represent 31% of the rock volume (Figure 16) and include very finecrystalline dolomite (L6), fine-medium-crystalline dolomite (L9) with grain-moldic porosity, and lithofacies with grain-supported texture, including packstone (L7), grainstone (L8), and marine sandstone (L10). Continental, fine-grained sandstones represent only 6% of the rock volume, but are important to flow and storage in the Council Grove. All 11 lithofacies are present in the Chase Group, but continental sandstones and very fine-crystalline, sucrosic dolomites are less common than in the Council Grove. Coarser crystalline dolomite with grain-moldic porosity, typically dolomitized bioclastic or ooid grainstone or packstone, is absent, and marine sandstone is uncommon in the Council Grove. The very fine-crystalline dolomite is interpreted to have originally been mud-rich carbonate for the most part.

Lithofacies Prediction To predict lithofacies using neural network analysis, we used a standard, single-hidden-layer neural network (Hastie et al., 2001) based on wire-line well logs in 1364 node wells throughout the Hugoton and Panoma fields. As illustrated in Figure 17, the input vector to the neural network included two computed

Lithofacies Class

Frac-fill 10 – 50% Frac-fill 5 – 10% Shale > 90% Shale 75 – 90% Shale 50 – 75% Shale 25 – 50% Shale 10 – 25% Wispy 5 – 10% Trace 1 – 5% Clean < 1%

Argillaceous Content

5

NM sandstone NM siltstone NM shaly siltstone Mar shale and siltstone Mdst/Mdst-Wkst Wkst/Wkst-Pkst Vfxln sucrosic (Dol) Pkst/Pkst-Grnst Grnst/PA Baff F-Mxln sucrosic moldic Dol Marine sandstone

cavern vmf (>64 mm) med – lrg vmf (4 – 64 mm) sm vmf (1 – 4 mm) crs (500 – 1000 Mm) med (250 – 500 Mm) fn (125 – 250 Mm) pin-vf (62 – 125 Mm) Pinpoint (31 – 62 Mm) Microporous (<31 Mm) Nonporous

Principal Pore Size

4

*After Dubois et al. (2003). (A) Five-digit classification system used for core descriptions at 0.5-ft (0.15-m) intervals, gathered by visual observation with the aid of binocular microscope. A total of seven other variables were recorded but were not used in determining lithofacies. (B) Digital code for 11 lithofacies. An example, 13323, is a continental siliciclastic, very fine-grained sandstone (203 – 410 ft; 62 – 125 m), with pinpoint porosity and wispy clay laminations (5 – 10% clay). Abbreviated are nonmarine (NM), marine (Mar), carbonate mudstone (Mdst), wackestone (Wkst), packstone, (Pkst), grainstone (Grnst), phylloid algal bafflestone (PA Baff), dolomite (Dol), very fine crystalline (Vxln), and fine to medium crystalline (F-Mxln).

0 1 2 3 4 5 6 7 8 9 10

vcrs rudite and cobble congl (>64 mm) m-crs rudite/pebble congl (4 – 64 mm) fn rudite/vcrs sand (1 – 4 mm) Arenite/crs sand (500 – 1000 Mm) Arenite/med sand (250 – 500 Mm) Arenite/fn sand (125 – 250 Mm) Arenite/vfn sand (62 – 125 Mm) crs lutite/crs silt (31 – 62 Mm) fn-med lutite/vf-m silt (4 – 31 Mm) Clay (<4 Mm)

1/>2 1/2 1/0-1 0.2/<3 3-8/01 3-8/2-3 6-8/8/<3 3-8/4-5 3-8/6-7 7-8/8/>2 2/>2

Cobble conglomerate Sucrosic/pebble conglomerate Baffle-boundstone/vcrs sandstone Grainstone/crs sandstone Packstone-grainstone/med Ss Packstone/fn sandstone Wackestone-packstone/vfn Ss Wackestone/crs siltstone Mudstone-wackestone/vf-m silt Mudstone/shale/clay

3 Grain Size

Lithofacies

Evaporite Dolomite Dolomite-limestone Dolomite-siliciclastic Limestone Carbonate-siliciclastic Siliciclastic-carbonate Marine siliciclastic Continental siliciclastic Shale

9 8 7 6 5 4 3 2 1 0

Dunham/Folk Classification

2

Variables

Digital Description

Rock Type

Code

1

Table 2. Digital lithofacies description system.*

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 321

322 / Dubois et al.

FIGURE 13. Distribution of Hugoton cores (continuous) for which lithofacies were defined at 0.5-ft (0.15-m) intervals.

geologic variables, a depositional environment indicator (MnM) and a stratigraphic cycle relative position (RelPos), in addition to the following wire-line-log parameters: gamma ray, logarithm of deep induction resistivity, average of neutron and density porosity, neutrondensity porosity difference, and photoelectric factor (PE, where available). No adjustment was made for thinbed or boundary effects. For each input vector, the network computed a vector of output values representing the corresponding lithofacies membership probabilities, and the most probable lithofacies was assigned for the logged interval. The network was trained based on associations between core-defined lithofacies and the log and geologic constraining variables. A comparison of core-defined lithofacies, lithofacies membership probabilities, and predicted discrete lithofacies is shown in Figure 18. The two geologic variables were derived from a 25-formation (or member) set of tops (Figure 8), which are also the tops of marine or nonmarine (continental) half-cycles. Values for the depositional environment indicator were (1) nonmarine (continental) and (2) marine and (3) intertidal, for the Herington and Holmesville, where intertidal environments are dominant. The MnM variable helps to distinguish between lithofacies with similar petrophysical properties but developed in different broad depositional environments. Values for the stratigraphic cycle relative position parameter (RelPos) range linearly with depth from 0 at the bottom of each half-cycle interval to 1 at the top, indicating position in each interval. Including this curve allowed the network to encode information regarding the fairly regular succession of lithofacies commonly exhibited within each interval and,

FIGURE 14. Major lithofacies in Chase and Council Grove, lithofacies code 0-5. (A) Continental sandstone (L0): Example: Blue Rapids (Council Grove, B1_SH), Cross H Cattle 1-6, 2652 ft (808 m). Coarse silt to very fine-grained sandstone, mostly quartz, massive bedded, adhesive meniscate burrows (S. Hasiotis, 2005, personal communication). Low-relief migrating eolian system. Digital classification: 13322. (B) Continental coarse siltstone (L1): Example: Stearns (Council Grove, B4_SH), Newby 2-28R, 2963 ft (903 m). Coarse quartz silt, rhizolith (Rz) and root traces with reduction haloes (Ho). Savannah, slow accumulation of silt by airfall, stabilized by vegetation and soil processes. Digital classification: 12213. (C) Continental shaly siltstone (L2): Example: Hooser (Council Grove, B3_SH), Newby 2-28R, 2944 ft (897 m). Fine- to medium-grained quartz silt and clay, caliche (Ca), rhizolith (Rz), and root traces with reduction haloes (Ho). Coastal plain, slow accumulation of silt by airfall, stabilized by vegetation and soil processes. Digital classification: 11114. (D) Marine siltstone and shale (L3): Example: Funston (Council Grove, A1_LM), Newby 2-28R, 2872 ft (875 m). Very finegrained shaly siltstone. Siliciclastic-dominated shelf during maximum flooding. Plug $ = 4.6%; k = 0.0001 md. Digital classification: 21104. (E) Mudstone and mudstone-wackestone (L4): Example: Crouse (Council Grove, B1_LM), Alexander D-2, 2962 ft (903 m). Silty mudstone-wackestone, wispy laminations, and ministylolites (Ms), burrowed in part (Bh), sparse normal-marine fauna, including fusulinids (Fs). Low-energy shelf at a time close to maximum flooding. Plug $ = 3.1%; k = 0.00239 md. Digital classification: 41113. (F) Wackestone and wackestone-packstone (L5): Example: Fort Riley (Chase), Flower A-1, 2700 ft (823 m). Slightly dolomitized wackestone, normal-marine faunal assemblage includes echinoids, brachiopods, bryozoan, and fusulinids. Intercrystalline micropores (blue in thin section) in dolomitized mud matrix is dominant porosity (core slab and thin section stained with alizarin red). Low-energy normal-marine shelf. Full-diameter $ = 15.2%; k = 0.413 md. Digital classification: 52111.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 323

thus, transfer some of that character to the sequence of predicted lithofacies in each well. The two curves were computed for the node wells using Visual Basic code in an ExcelR spreadsheet run in a batch-processing routine and exported as log curves in a Log Ascii Standard (LAS) file format. They were then combined with the wire-line-log curves to complete the feature vector.

The neural network code has been added to Kipling.xla, an ExcelR add-in for nonparametric regression and classification (Bohling and Doveton, 2000). For this study, the network was trained to match observed associations between logs and lithofacies identified in core from a set of key wells shown in Figure 13 (8 Chase wells with 3952 0.5-ft [0.15-m]

324 / Dubois et al.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 325

FIGURE 16. Relative proportions of 11 lithofacies in 4250-ft (1300-m) core from Chase and Council Grove groups, Hugoton field. Eleven lithofacies are identified by a code (L0–L10): L0 = sandstone; L1 = coarse siltstone; L2 = shaly siltstone; L3 = siltstone; L4 = carbonate mudstone; L5 = wackestone; L6 = very fine-crystalline dolomite; L7 = packstone; L8 = grainstone and phylloid algal bafflestone; L9 = fine-medium-crystalline moldic dolomite; and L10 = sandstone.

intervals in the training set, and 10 Council Grove wells with 4593 0.5-ft [0.15-m] intervals). Fundamental parameters controlling the network behavior are the number of nodes in the hidden layer (network size) and a damping or decay parameter. Increasing the network size allows the network to match the training set more closely, but using too many hiddenlayer nodes leads to the network becoming precisely

tuned to the training data and unable to generalize. Increasing the damping parameter counteracts precise tuning but can result in the network developing smoother representations of the boundaries between lithofacies. We used cross-validation to estimate the optimum values for network size and damping parameter. The cross-validation was done in two ways: (1) splitting the entire training data set, regardless of well, into random subsets, with two thirds of the cases used for training and the other one third for testing (comparison of predicted and actual lithofacies); and (2) taking out each well in turn, training on the remaining wells, and testing on the removed well. Training and testing were repeated several times for each parameter combination to account for the random variation between different realizations of the network. This computationally intensive cross-validation was performed using scripts in the R statistical computing language (R Development Core Team, 2005) using the nnet function developed by Venables and Ripley (2002). The scripts computed several measures of correspondence between actual and predicted lithofacies, including the average lithofacies misallocation cost for all intervals in the test well. This value is computed from a misallocation cost matrix that assigns a cost for misallocation that is proportional to the distance in lithofacies code units in the lithofacies spectrum from actual. For example, calling a packstone (L7) a marine siltstone (L3) carries a higher cost than confusing it with a grainstone (L8) with similar

FIGURE 15. Major lithofacies in Chase and Council Grove, lithofacies code 6-11. (A) Very fine-crystalline sucrosic dolomite (L6): Example: Cottonwood (Council Grove, B5_LM), Beatty E-2, 2800 ft (853 m). Finely crystalline, sucrosic dolomitized mudstone, locally with anhydrite cement and replacement in nodules and along fracture (An). Porosity (blue in thin section) is microporous (intercrystalline) and pinpoint (molds-Mo). Restricted, protected lagoon. Plug $ = 13.9%; k = 1.37 md. Digital classification: 88120. (B) Packstone and packstone-grainstone (L7): Example: Winfield (Chase), Flower A-1, 2579 ft (768 m). Medium- to coarse-grained bioclastic-oncoid packstone, patchy anhydrite cement (An). Most porosity (blue in thin section) is intergranular. Carbonate sand shoal on open shelf. Full-diameter $ = 16.4%; k = 5.98 md. Digital classification: 54520. (C) Grainstone or phylloid algal bafflestone (L8): Lithofacies have similar core and wire-line-log properties and were lumped because of their small populations. Example 1: Cottonwood (Council Grove, B5_LM), Alexander D-2, 3024 ft (922 m). Medium-coarse grained oncoid-peloid grainstone. Well-connected intergranular porosity is blue in thin section. Carbonate sand shoal on restricted shelf. Full-diameter $ = 18.8%; k = 39.0 md. Digital classification: 56540. Example 2: Cottonwood (Council Grove, B5_LM), Newby 2-28R, 2992 ft (912 m). Phylloid algal bafflestone. Phylloid algal blade molds (Pm) partially filled with anhydrite cement (An). Matrix is largely peloid-pellet packstone (Pp). Phylloid algal mound on slightly restricted shelf. Full-diameter $ = 20.6, k = 1141 md. Digital classification: 57770. (D) Fine to medium crystalline moldic dolomite (L9)-Example: Krider (Chase), Flower A-1, 2516 ft (767 m). Fine-medium crystalline moldic dolomite. Large molds (Mo), possibly ooids and bioclasts, dominate the wellconnected pore system in a dolomitized medium-coarse grained grainstone. Patchy anhydrite cement (An) occludes some porosity. Carbonate sand shoal on an open shelf. Full-diameter $ = 22.3%; k = 275 md. Digital classification: 88550. (E) Marine sandstone (L10): Example: Herington (Chase), Flower A-1, 2485 ft (757 m). Planar (Px) and ripple (Rx) crossbedding and vertical burrows (Bv). Tidal flat. Very coarse silt to very fine-grained sandstone, well-sorted, subarkose (83% of detrital fraction is quartz, by x-ray diffraction), well-connected intergranular porosity (blue), patchy anhydrite cement (An). Full-diameter $ = 20.8%; k = 48.2 md. Digital classification: 23321.

326 / Dubois et al.

FIGURE 17. Schematic representation of single hiddenlayer neural network used to predict lithofacies from wire-line logs and geologic constraining variables. Inputs two geologic constraining variables (MnM = depositional environment indicator; RelPos = relative position in stratigraphic interval) and includes an array of nuclear and electrical wire-line-log curves: gamma ray, (GR); logarithm of deep induction log (LogILD); average of neutron and density porosity (%N + %D)/2); difference between neutron and density porosity (%N  %D); and photoelectric effect (PE). Outputs are lithofacies occurrence probabilities.

petrophysical properties. Absolute accuracy in lithofacies prediction, although desirable, is unrealistic, and our goal was to limit error to the nearest lithofacies class. Plots of median average misallocation cost versus damping parameter and network size (panel variable) for the well-by-well cross-validation, including the photoelectric wire-line log (PE) in the Council Grove (Figure 19), illustrate the median computed over 40 average misallocation costs for each parameter combination: five trials per well for each of the six Council Grove wells with PE logs. Cross-validation plots for the Council Grove case without PE, and for the Chase, with and without PE, were similar (not illustrated). Although the network performed reasonably well throughout a range of parameter values, we chose to use a network size of 20 hidden-layer nodes and a damping parameter of 1.0. The damping parameter chosen (1.0) exhibited consistently low misallocation values, and the number of hidden-layer nodes (20) was chosen over configurations with fewer nodes that tended to overgeneralize. Initially, we trained four neural networks: Chase with and without PE logs and Council Grove with and without PE logs. Predictions using the model including PE were used wherever possible. The Chase models included all 11 lithofacies, but the Council Grove models included only 9 because the fine- to medium-crystalline dolomite and marine sandstone do not occur in the Council Grove in sufficient volume to be considered as separate classes. Later, we

added two additional models for the Chase below the Towanda (with and without PE) to better represent the distribution of marine sandstone in that interval and region. These six neural network models were then applied as appropriate to produce predicted lithofacies versus depth logs at 0.5-ft (0.15-m) intervals in the 1364 node wells distributed throughout the field (Figure 20A). The batch prediction capability of the Kipling program was used in this case, with logs being read from LAS files and the predicted lithofacies curves being written out to LAS files. Following the construction of predicted lithofaciesdepth curves, the predicted lithofacies curves were read TM into Petrel (Schlumberger 3-D modeling software) and upscaled to the resolution of the model layers (roughly 2 ft [0.6 m] in marine intervals and 4 ft [1.3 m] in nonmarine intervals) by majority vote: the lithofacies for each model cell intersecting a well is taken to be the most frequently occurring lithofacies in that layer in the well. Voxel-based methods were chosen over object-based methods for facies and property modeling because of the relatively dense well control and geometry of the lithofacies bodies being modeled (thin and laterally extensive). Sequential indicator simulation (Deutsch and Journel, 1998) as implemented in Petrel was used to generate lithofacies values in all model cells conditioned on the upscaled lithofacies values in the node well cells. The development of variograms for the lithofacies by data analysis on a zone-by-zone basis (24 zones and 11 lithofacies) is

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 327

FIGURE 18. Comparison of predicted lithofacies versus core-defined lithofacies. Illustrated vertical plots of lithofacies membership probabilities, predicted discrete lithofacies (most probable), and lithofacies in core at the 0.5-ft (0.15-m) scale for the Chase and Council Grove from two separate wells in the training set (Youngren and Stuart, respectively). Neural networks were those used in estimating lithofacies in node wells (trained on all wells). Probabilities were not used as an input for modeling, but they do illuminate some of the misallocations (actual lithofacies is commonly in second place).

difficult and not justified considering the well control and lithofacies geometry. Regions were therefore classified using a limited number of variograms for each lithofacies and large horizontal ranges. Using large horizontal ranges encouraged the stochastic simulation to produce facies bodies that are as laterally extensive as possible, but still subject to conditioning data at the node wells. This provided facies bodies consistent with geologic depositional models but constrained by node well data. Lithofacies bodies are laterally extensive because of both the width of the lithofacies spectrum used in determining class (lumped lithofacies) and the geologic conditions under which they were deposited. Node well spacing is much closer than the width of the lithofacies bodies. For the marine half-cycles, the major axis was set at 30,000 ft (9144 m) and minor axis at 25,000 ft (7620 m) and the azimuth equal to 118, the approxi-

mate depositional strike, whereas in the continental and tidal-flat half-cycles, a 30,000  30,000-ft (9144  9144-m) range was used. Vertical range was set at approximately two times the mean layer height in the zone, and nuggets ranged from 0.1 to 0.22 on the basis of limited data analysis. The very large number of conditioning data helped reduce the sensitivity of the simulated results to the variogram model parameters, making the predictions more deterministic. Table 3 shows the proportional distribution of lithofacies defined in core in the training set (keystone wells) and predicted lithofacies at the node wells and in the lithofacies model in the study area. Considering bias by the geographic distribution of sampling and the scale change from 0.5 ft (0.15 m) to much thicker layering, consistency in the distribution at each scale can be characterized as good. Most discrepancies are between the core-defined lithofacies

328 / Dubois et al.

FIGURE 19. Example results for cross-validation analysis to determine optimal values of neural network size (number of hidden-layer nodes) and damping parameter. Results shown are the median over eight Council Grove wells with core-defined lithofacies and wire-line photoelectric curve. The procedure was to perform five trials per well; leave out each well in turn, train on the other seven wells, and predict on the subject well. Median average misallocation cost versus damping parameter and network size for all wells was then plotted. A network size of 20 hiddenlayer nodes and a damping parameter of 1.0 were chosen.

(training wells) and the lithofacies predicted by neural networks (node wells) and are caused by the position of the training wells on the shelf. Training wells are skewed to the west (because of core availability), whereas the node wells are more evenly distributed. An exception is grainstone (L8), where the neural networks as trained are not effectively predicting this facies. Fortunately, this facies represents only a small part of the volume (2.3% in the training wells) and is commonly mistaken as packstone, a close neighbor, in the neural network training exercises.

Core Petrophysics Previous petrophysical studies of the Hugoton and Panoma fields have generally used average properties assigned to formations (e.g., Siemers and Ahr, 1990; Olson et al., 1996). Porosity and permeability characteristics of reservoir-quality wackestone, packstone, and grainstone lithofacies in the Council Grove, Texas County, Oklahoma, were reported by Heyer (1999). Byrnes et al. (2001) and Dubois et al. (2003) presented lithofacies-specific petrophysical properties for the Council Grove in Panoma field and illustrated the similarities between low-permeability carbonates and sandstone. Fundamental to construction of the reservoir geomodel is the population of cells with the basic lithofacies and their associated petrophysical properties—porosity, permeability, and

fluid saturation. Petrophysical properties vary among the 11 major lithofacies. Principal lithofacies-specific petrophysical properties analyzed and discussed here include routine helium and in-situ porosity, routine air and in-situ Klinkenberg gas permeability, grain density, capillary pressure, and gas-water relative permeability. Data for routine porosity, permeability, and grain density were compiled for more than 6000 full-diameter core and plug samples from measurements performed by commercial laboratories and the Kansas Geological Survey. Lithofacies were determined for more than 3500 samples. Core-plug sampling was designed to represent the range in porosity, permeability, and lithofacies in the study area. Routine air permeability was generally measured under a confining pressure of approximately 400 psi (2.8 MPa); in-situ Klinkenberg gas permeability (high-pressure gas or liquid equivalent) and in-situ porosity were measured under a hydrostatic confining pressure of 800 psi (5.5 MPa) and a hydrostatic in-situ stress equal to the net effective stress. A correlation between routine helium and insitu porosity was determined for 245 core samples representing the range in porosity and lithofacies. In addition, capillary-pressure curves were obtained for 252 samples using air-mercury capillary-pressure intrusion analysis. Data for 32 gas-water drainage-relative permeabilities were compiled, and effective gas permeability at critical water saturation was measured on more than 200 cores.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 329

FIGURE 20. (A) Map illustrating the location of 1364 node wells used for the static model construction. Fourteen of the wells have core-defined lithofacies (circled) and the balance has lithofacies predicted by neural networks. The 1364 is a mix of wells with Council Grove (1146) and Chase (1069) wells with lithofacies defined by neural net (854 of the 1364 have both Council Grove and Chase lithofacies). Only wells with lithofacies defined at least to the top of the Fort Riley (Chase) or the Florena Shale, B5_SH (Council Grove), were considered. (B) Map showing 8765 wells with formation-member level tops was used for building the structural and stratigraphic framework for the Hugoton geologic model.

Porosity and Grain Density

Permeability

Routine (unconfined) helium porosity ($He) values range from 1 to 26% (Figure 21), with in-situ porosity ($i) values averaging 0.54 porosity units less than routine porosity values (i.e., $i = $He  0.54) for all lithofacies and porosity rocks. Comparing porosity among lithofacies, average porosity decreases with decreasing grain size (i.e., very fine sandstone to very fine-medium siltstone) in the continental siliciclastics and from grainstone to mudstone in the carbonates. Continental rocks exhibit an average grain density of 2.70 ± 0.08 g/cm3 (error expresses 2 standard deviations), limestones exhibit a grain density of 2.72 ± 0.06 g/cm3, and dolomites exhibit a grain density of 2.82 ± 0.08 g/cm3.

Core-measured in-situ Klinkenberg permeabilities (kik) range from 0.00002 to 400 md (2  108 –0.4 mm2). Permeability is a function of several variables, including primarily pore-throat size, porosity, grain size and packing (which controls pore body size and distribution), and bedding architecture. Lithofacies-specific equations were developed to predict permeability using porosity as the independent variable because porosity data are the most economic and abundant and because porosity is well correlated with the other variables for most lithofacies. More than approximately 75% of all rocks in the Hugoton exhibit an in-situ Klinkenberg permeability of less than 1 md (0.001 mm2). Obtaining accurate

330 / Dubois et al.

Table 3. Relative distribution of 11 lithofacies in core, node wells, and cellular model.* Height Source

0.5 ft (0.15 m) Actual (%)

0.5 ft (0.15 m) NNet Predicted (%)

Variable** Upscaled (%)

Variable** Modeled (SIS) (%)

Lithofacies Code

Training

Node Wells

Node Wells

All Cells

0 1 2 3 4 5 6 7 8 9 10 Count (N)

5.6 23.3 12.9 7.5 5.4 14.5 2.8 14.7 2.3 5.6 5.4 8545

2.2 19.7 9.6 9.6 4.3 20.1 4.9 24.7 0.2 1.4 3.4 993,146

1.0 17.0 7.1 9.0 3.6 22.2 3.9 25.9 0.2 3.8 6.3 183,949

1.1 16.7 6.7 9.1 3.9 22.5 3.8 25.2 0.2 3.8 7.1 107,765,147

*Core-defined lithofacies for 14 wells were used in neural network Training for lithofacies prediction in 1364 Node Wells. Lithofacies 0.5 ft (0.15 m) in node wells were upscaled to model layer thickness (Variable Upscaled). Sequential indicator simulation (SIS) was used to populate the cellular model (All Cells) between the node wells. **Model layer h. Average of mean h = 3.3 ft (1 m). Range of mean h = 1.9 – 5.2 ft (0.57 – 1.58 m). Lithofacies 0 – 2 tend to be in thicker layers.

permeability values for rocks with low permeability requires correction for the influence of confining stress and care to avoid core data influenced by stressrelease microfractures. Figure 22 shows that both continental and carbonate rocks in the study area exhibit a significant decrease in permeability below approximately 1 md (0.001 mm2) because of the influence of Klinkenberg gas slippage and confining stress combined. This decrease is consistent with other tight gas rocks (Byrnes et al., 2001). Cores with identified macrofracturing exhibit increasingly greater permeability with decreasing matrix permeability (Figure 23). The permeability of the fractured cores with matrix permeability below 0.5 md (0.0005 mm2) can be attributed to the core permeability measurement reflecting the permeability of fractures in the sample with the matrix contribution being small or negligible. Full-diameter analysis, generally performed at confining pressures less than 400 psi (2.8 MPa), commonly exhibit significant difference from plug values for permeabilities below 0.5 md (0.0005 mm2), even for samples where fractures were not identified but microfractures may have been present. To obtain accurate matrix permeabilityporosity correlations, full-diameter permeability data for permeabilities less than 0.5 md (0.0005 mm2) that were more than 2 standard deviations outside the plug-defined matrix permeability-porosity trend were not used in final matrix correlations. The correlation

between matrix and well test permeability is discussed in a following section. As with many sedimentary rocks, the relationship between permeability and porosity can be approximated by power-law functions, although the relationship changes slightly in some lithofacies at porosities below approximately 6%. Each lithofacies exhibits a relatively unique kik  $i correlation that can be represented using a power-law equation of the form (Figure 24) kik ¼ AfBi where kik is in millidarcys (md); porosity is in percent (%); and values for A and B are shown in Table 4. Subparallel trends are apparent for the continental siliciclastics, the sucrosic dolomites, and the mudstoneto-grainstone limestones. For these trends, the standard error of prediction ranges from a factor of 3.3–9.1. At $i > 6%, permeability in a grainstone-bafflestone can be 35 times greater than mudstone and 150 times greater than marine siltstone of similar porosity. These differences illustrate the importance of identifying lithofacies to correctly predict permeability from wireline-log porosity. Differences in permeabilities between continental very fine sandstone and coarse siltstone and between coarse siltstone and very finemedium siltstone are approximately 2.5 times, whereas the difference between marine sandstone

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 331

FIGURE 21. Histogram of routine helium porosity for Chase and Council Grove nonmarine continental (NM) sandstones and siltstones (A) and limestones (B). Lithofacies codes and descriptions are provided in text. Porosity generally increases with increasing grain size in siliciclastics and with decreasing mud content from mudstone through grainstone (Baff = bafflestone; Grst = grainstone; Pkst = packstone; Wkst = wackestone; Mdst = mudstone).

and marine siltstone is approximately 10 times. Although models for permeability prediction using porosity were developed for utility, we recognized that the dominant control on permeability is the pore-throat size (Figure 25).

Water Saturation and Capillary Pressure It is important to consider the presence of water in the pore space of low-permeability reservoirs both for accurate volumetric calculations and because water occupies critical pore-throat space and can greatly diminish gas permeability, even in rocks at irreducible water saturation (Swi). In the Hugoton,

determination of formation water saturation from electric wire-line-log response is problematic because of deep mud filtrate invasion with conventional mud programs because of the low reservoir pressure (Olson et al., 1997; George et al., 2004). Because water saturations could not be reliably determined for most wells using logs, saturations were estimated based on matrix capillary-pressure properties and determination of the free-water level (level at which gas-brine capillary pressure is zero). Air-mercury capillary-pressure data were compiled and measured for 252 samples ranging in porosity, permeability, and lithofacies, and relationships developed that allowed the prediction of a capillary-pressure curve for any given lithofacies and porosity. To examine the lithofacies dependence of threshold-entry pressure, gas-column height, and porethroat size (Figure 26), laboratory capillary-pressure data were converted to reservoir gas-brine capillarypressure data using the standard equation (Purcell, 1949; Berg, 1975): Pcres = Pclab (scosures/scosulab); where Pcres is the gas-brine capillary pressure (psia) at reservoir conditions, Pclab is the laboratory-measured capillary pressure (psia), and scosures and scosulab is the interfacial tension (s, dynes/cm) times the cosine of the contact angle (u, degrees) at reservoir and laboratory conditions, respectively. For the Hugoton and Panoma fields, the interfacial tension is approximately 63–65 dynes/cm for the initial reservoir gas pressures of about 400–450 psi (2.8–3.1 MPa) and temperatures of 90–1008F (32–388C) (Hough et al., 1951; Jennings and Newman, 1971). Conversion of capillary pressure to height above free-water level to determine the water saturation in any given rock type as a function of height above the free-water level requires the conversion of capillary-pressure data to height above free-water level. This conversion was performed using the standard relation (Hubbert, 1953; Berg, 1975): H = Pcres/(C(rbrine  rgas)), where H is the height (ft) above free-water level, Pcres is the capillary pressure (psia) at reservoir conditions, rbrine and rgas are the density of brine and gas at reservoir conditions (rbrine = 1.16– 1.19 g/cm3 and rgas = 0.025–0.035 g/cm3, which are reasonable intermediate values for these fields, and C is a constant (0.433 [psia/ft]/[g/cm3]) for converting density to pressure gradient in psia/ft. From the airmercury capillary-pressure data, pore-throat diameter was calculated using the modified Washburn (1921) relation: d = 4Cscosu/Pc, where Pc = capillary pressure (psia); C = 0.145 psiacmmm/dyne; u = contact angle (1408); s = interfacial tension (484 dynes/cm); and d = pore-throat diameter (mm). This relation

332 / Dubois et al.

FIGURE 22. Crossplot of routine air permeability (kair) versus in-situ Klinkenberg gas permeability (kik) for Council Grove rocks (gray circles) and Chase rocks (black triangle). Influence of both confining stress and Klinkenberg correction increase with decreasing permeability. Values of kik can be predicted approximately from kair using log10kik = 0.059 (log10kair)3  0.187 (log10kair)2 + 1.154 log10kair  0.159, where permeabilities are in millidarcys (md). Variance is caused by both differing routine conditions and rock response.

assumes that the nonwetting phase (i.e., gas) enters the pores through circular pore throats. Figure 26 illustrates selected capillary-pressure curves for samples of different permeability. Differences among capillary-pressure curves for the various lithofacies correspond to variations in threshold-entry pressure, pore-throat diameter, and water saturation for various gas-column heights above the free-water level, including the thickness of the transition zone from Sw = 100% to approximately Swi. Capillary pressures and corresponding water saturations (Sw) vary among lithofacies and with porositypermeability and gas-column height. Threshold entry pressures and corresponding heights above free-water level are well correlated with permeability (Figure 26). This is consistent with the relationship between pore-throat size and permeability. The figure shows that for rocks with in-situ Klinkenberg gas permeability below approximately 0.003 md (3  106 mm2), threshold entry heights are greater than the gascolumn heights available in the Hugoton, and therefore, the samples have Sw = 100%.

Synthetic capillary-pressure curves were constructed based on capillary pressure-porosity-permeability-lithofacies relationships exhibited by the 252 cores analyzed, representing the range in lithofacies and permeability. Capillary pressures in each lithofacies can be represented to be a function of porosity. Modeled capillarypressure curves for two important lithofacies (Figure 27) illustrate that with decreasing porosity (and associated permeability), threshold entry heights and transition zone heights increase. Example capillary-pressure curves for different lithofacies at a given 10% porosity (Figure 28) illustrate the significant differences in Sw that can exist among lithofacies at any given height above free-water level. Because differences decrease with increasing height, saturations for all lithofacies approximately approach a similar irreducible saturation at gas-column heights above approximately 300 ft (90 m), except for lowporosity rocks where saturation differences are still evident. Using the capillary-pressure model, it was possible to predict water saturation for any given lithofacies and porosity at any given height above freewater saturation and, thus, populate every grid cell in the 3-D geomodel with water saturation values.

Relative Permeability Gas and water drainage relative permeability curves reveal several characteristics similar to other lowpermeability rocks. Water permeability, even at 100% Sw, is less than Klinkenberg gas permeability and decreases with decreasing permeability. Gas relative permeability is less than the absolute gas permeability at

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 333

FIGURE 23. Crossplot of in-situ Klinkenberg permeability versus in-situ porosity for whole core identified as fractured (asterisk), whole core that were not identified as fractured but may contain microfractures (gray circle), and unfractured core plugs (black triangle). Permeabilities shown are either measured in-situ values or routine values corrected to in-situ conditions using the equation presented in Figure 22. Whole core (full-diameter) values diverge from matrix (plug) values at porosities less than about 10% and matrix permeability of approximately 0.5 md, reflecting the influence of microfracture(s) on permeability in whole core samples with porosity less than 10%. Above 10% porosity, influence of microfracture(s) is small.

all water saturations greater than zero and gas relative permeability decreases significantly as Sw increases above 50%. Relative permeabilities can be reasonably modeled using Corey-type equations (Figure 29), similar to other low-permeability rocks (Byrnes, 2003).

Permeability at Different Scales Fundamental to modeling the permeability distribution in the Hugoton is the need to understand the relative function of matrix and fracture flow and the possible scale dependence of permeability. Figure 23

showed that for rocks below approximately 8% porosity, or approximately 0.5 md (0.0005 mm2), microfractures in core significantly increased permeability. A fundamental question for these data is as follows: are the microfractures present in the subsurface, or are they a stress release or coring-induced phenomenon? This question can only be answered by comparing upscaled matrix permeabilities with unfractured full-diameter permeabilities and with DST- or well test-calculated permeabilities. Comparing carefully examined unfractured full-diameter permeability values with core-plug values measured on plugs

334 / Dubois et al.

Table 4. Power-law coefficients (A) and exponents (B) by lithofacies for in-situ Klinkenberg (md) versus in-situ porosity (%) trendlines illustrated in Figure 24.

FIGURE 24. Crossplot of in-situ Klinkenberg permeability (kik) versus in-situ porosity ($i) for the continental siliciclastics (A), limestones (B), and dolomites (C). Lithofacies and codes are discussed in the text. Each lithofacies exhibits a relatively unique kik  $i correlation that can be represented using a power-law equation of the form kik = A$ Bi ; values for A and B are shown in Table 4. For some samples, routine permeability values were converted to in-situ values using the equation in Figure 22, and routine porosity was converted to in-situ porosity using the equation in the text. Trend lines exhibit a standard error of prediction ranging from 3.3 to 9 times depending on the lithofacies.

Lithofacies Code

In-situ Klinkenberg Permeability Coefficient kik = A$iB A

In-situ Klinkenberg Permeability Exponent kik = A$iB B

0 1 2 3 4 5 6 7 8 9 10

1.000E – 09 3.715E – 10 1.585E – 10 1.995E – 11 2.088E – 10 2.967E – 08 1.967E – 09 1.527E – 07 3.631E – 09 2.553E – 07 1.995E – 10

7.90 7.90 7.90 8.31 7.98 6.26 7.10 6.17 8.24 6.30 8.31

taken from the full-diameter cores (Figure 30) indicates that for homogenous samples, matrix properties apply to the full-diameter core scale. The ability to compare well-scale permeability with matrix permeability is limited because so few wells have DST or well test data for thin intervals for which core data are available and which were tested prior to hydraulic fracturing, which complicates artificial fracture-enhanced permeability with reservoir permeability. In four key research wells, permeability was measured using DST for multiple intervals for which core analysis was also performed. To compare with core permeabilities, full-diameter and plug permeabilities were arithmetically averaged (representing parallel-flow contribution from each depth interval) to determine average interval permeabilities. Correlation between DST and upscaled full-diameter and plug core permeabilities (Figure 31) shows good correlation for intervals with permeability greater than about 0.5 md (0.0005 mm2). For interval permeabilities below 0.5 md (0.0005 mm2), full-diameter permeabilities exhibit nearly constant permeability between 0.5 and 3 md (0.0005 and 0.000033 mm2), characteristic of microfracture-influenced permeability. Matrixscale plug permeabilities can be either higher or lower than DST permeabilities. Variance in the DST-matrix permeability correlation is partially or predominantly related to the limited vertical sampling of the core plugs and difficulty

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 335

FIGURE 25. Crossplot of principal pore-throat diameter (PPTD, micrometers) versus in-situ Klinkenberg permeability (kik) for lithofacies in Chase and Council Grove groups and sandstones from outside the Hugoton embayment for comparison. The good correlation through eight orders of magnitude shows the predominant influence that pore-throat size exerts on permeability and explains permeability changes with grain size and Dunham classification at a given porosity. Second Y-axis shows corresponding threshold entry heights necessary for gas column to enter sample for gas pressure and temperature conditions in the Hugoton area and discussed in text. The correlation between kik and PPTD can be expressed: PPTD = 2.2 kik0.42. Standard error of prediction for this correlation is a factor of 1.7.

FIGURE 26. Selected capillary-pressure curves for rocks of different permeability illustrating general curve characteristics. Capillary pressure has been converted to height above free-water level (at which Pc = 0) using equations in text. These curves illustrate how threshold entry height and transition zone height increase with decreasing permeability.

336 / Dubois et al.

ability for rocks with permeability greater than 0.5 md (0.0005 mm2), both full-diameter and plug data reflect matrix properties, and the good correlation with DST permeabilities indicates that the reservoir is not fractured at the scale of investigation of the DST. The better correlation of plug and DST permeabilities for intervals with permeability below 0.5 md, and the fact that upscaled permeabilities from plug data are greater than or equal to DST permeabilities for three of four intervals, can be interpreted to indicate that these intervals are also unfractured. These data, and less precise data from other wells, indicate that the production characteristics of many wells in the Hugoton are consistent with matrix properties control of flow, without significant natural fracture contribution. Data and statistics on the fraction of wells that exhibit production greater than what would be predicted from matrix properties have not yet been compiled and calculated.

Log Petrophysics

FIGURE 27. Model capillary-pressure curves converted to height above free-water level for nonmarine coarse siltstone (A) and packstone and packstone-grainstone (B) for different porosities. Threshold entry heights and transition zone heights increase with decreasing porosity for all lithofacies. The siltstones exhibit both greater entry heights and higher transition zones than corresponding porosity packstones. Threshold entry heights for coarse siltstones less than 6% porosity exceed existing closure in the field, indicating these rocks are water saturated. Curves for packstones less than 6% porosity are not modeled.

in representing some pore properties that are larger in scale than core plugs. The single, phylloid algal bafflestone interval exhibits significantly lower matrix permeability because core plugs did not sample the larger scale vuggy nature of this lithofacies, which exhibits high permeability. Because microfractures do not contribute significantly to measured perme-

The application and validation of statistical and petrophysical concepts to porosity estimation is an important factor in reserve calculations. As always, accuracy is a major concern, but special attention must be paid to potential bias introduced by gas effects and lithology variation. Cumulative effects of the factors can result in significant under- and overestimation of total hydrocarbons in place. As discussed earlier, the Chase and Council Grove are invaded extensively by mud filtrate during drilling with conventional mud programs, so that water saturations estimated from resistivity logs are adversely overestimated. Pore volumes remain unchanged regardless of invasion, so that density and neutron logs can be used with confidence for porosity estimation in the Hugoton. However, these logs must be evaluated carefully to eliminate, first, borehole environmental effects, and then the factors associated with variable mineralogy and variable gas effects. Because we used only relatively modern log suites (post-1980) and only neutron and density porosity for estimating porosity, there was no need to calibrate, normalize, or otherwise preprocess the logs other than checking quality. Because the density and neutron measurements are from devices that make contact with the borehole wall, the highly anomalous porosities caused by washouts must be removed prior to detailed analysis. Porosity cutoffs of 20–22.5% were applied by lithofacies to eliminate washout effects. This was followed by a

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 337

FIGURE 28. Model capillary-pressure curves converted to height above free-water level for lithofacies L0-L10 at a constant porosity of 10%. Figure 27 illustrates how curves change as a function of porosity for two lithofacies. In this figure, for a given porosity, comparison between curves at any given height above free water shows that water saturations are generally greater for the siliciclastics than the carbonates. The differences among curves illustrate the importance of knowing lithofacies to accurately predict water saturation.

remedial elimination of shoulders caused by the truncation. Porosities substituted into these washout intervals corresponded to average porosities of the lithofacies assigned to the interval. This procedure was found to be effective in removing washout effects and was performed using an automated process with limited manual intervention to preserve intervals with high, but real, porosities such as those that occur in more coarsely crystalline dolomites (L9). Following environmental correction, the calculation of porosities from logs could then be developed as a reliable procedure, particularly when validated through the use of porosities from the extensive core database as the calibration standard.

Porosity Log Calibration among Lithofacies The subdivision of the Chase and Council Grove into lithofacies allowed a statistical strategy of porosity-log calibration that was sensitive to mineralogy effects while simultaneously accommodating the effects of gas saturation in residual or greater volumes. Initial regression analysis that related core porosities to log measurements in the Council Grove showed a distinctive difference in calibration model between the siliciclastic lithofacies and the carbonate mudstone lithofacies (L0–L4) and the carbonate lithofacies (L5–L9). The best correlation of log and core porosities in the siliciclastic lithofacies was exhibited for the density

log alone, because the neutron log appears to be adversely and erratically affected by clay minerals in the fine fraction. Siliciclastic lithofacies were calibrated separately, and the relation between each one’s density porosity ($d) to core porosity is shown in Figure 32. A striking distinction is present between the continental lithofacies (L1 and L2) and the marine siliciclastic lithofacies (L3), which exhibits relative homogeneity. It should be noted here that the carbonate mudstone lithofacies (L4) typically has high silt content, and its population can be characterized as exhibiting properties more like a fine-grained siliciclastic than a pure carbonate mudstone. By contrast, in the carbonate lithofacies L5 to L9, the average of the neutron ($n) and density ($d) porosities, calibrated to a limestone matrix, was found to be the best estimator of porosity, compensating for changes in lithology between limestone and dolomite, as well as minor saturations of residual gas.

Compensation for Gas Effects Although gas effects are minor in the Council Grove and can be compensated for by the averaging of neutron and density porosities, they can be significant in the Chase (Figure 33) and must be accommodated in an expanded equation set. The equation commonly applied to estimate the porosity with compensation for gas takes the form $ = (($n2 + $d2)/2)0.5. The equation closely approximates the formula derived by Gaymard and Poupon (1968) from a petrophysical model of a gas-filled reservoir. An alternative and empirical equation that is also widely used is a simple weighted average of the neutron and density porosities with a weighting of one third applied to the neutron porosity and a two thirds weighting of the density porosity (Asquith and Krygowski, 2004). This

338 / Dubois et al.

rosities related to neutron and density fractional log porosities resulted in an equation of the form $ = 0.37$n + 0.62$d for limestone (lithofacies L5, L7, and L8) and $ = 0.04 + 0.31$n + 0.53$d for dolomite (L6 and L9). The extra term in the dolomite equation accommodates the lithology correction required for logs calibrated to a limestone matrix. Porosities in the

FIGURE 29. Relative gas permeability (A) and water relative permeability (B) curves for 32 samples of various lithofacies. Black curves represent predicted values for the Corey-type equation model: krg = (1  (Sw  Swc,g)/(1  Sgc  Swc,g))p (1  ((Sw  Swc,g)/(1  Swc,g))2), krw = ((Sw  Swc)/ (1  Swc))q (kw/kik), where Swc is the critical water saturation, Swc,g is the critical water saturation for gas flow, and all saturation terms are in fractions. Black curves represent mean values of exponents p = 1.3 for gas curve and q = 8.3 for water curve, whereas gray-bounding curves represent outer limits of curves using exponents p = 1.3 ± 0.4 and q = 8.3 ± 3, which represent the range exhibited by the sample set, which had 0.1 < kik < 50 md.

empirical equation closely emulates the gas correction shown on service company neutron-density crossplots (Figure 34). A regression analysis of core po-

FIGURE 30. Crossplot of full-diameter core porosity versus plug porosity (A) and permeability (B) for samples in which the full-diameter cores did not exhibit any apparent microfracturing. Good correlation indicates that matrix-scale properties apply to full-diameter scale. Variance can be attributed to full-diameter core sampling multiple lithofacies or a range in porosity not sampled by the corresponding core plug.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 339

FIGURE 31. Crossplot of calculated interval drillstem test (DST) formation permeability versus average interval permeability calculated from full-diameter core for four wells and from core plugs in well 1. Routine core data were corrected for confining stress, Klinkenberg, and relative permeability effects so as to correspond to reservoir-condition values. Good correlation down to about 0.5 md shows matrix-scale control of flow in the region of DST investigation. Below 0.5 md, microfractures in fulldiameter core result in permeabilities higher than in the unfractured reservoir. Higher DST than core-plug permeabilities can be interpreted to indicate that formation is not fractured in the range of investigation, and that plug sampling density was probably not adequate to properly sample lower range of permeability.

FIGURE 32. Porosity calibration of siliciclastic zones based on regression of core measurements on density-log porosity in limestone-equivalent units for continental facies (L1 and L2) and marine facies (L3 and L4).

340 / Dubois et al.

FIGURE 33. Example of strong (and atypical) gas effect on the neutron and density-porosity logs in the Towanda Limestone for a well in Stevens County, Kansas.

marine sandstone (F10) were calibrated to log porosities with the equation $ = 0.05 + 0.15$n + 0.47$d. Although there are comparatively fewer core measurements in the continental sandstone (F0), the use of the averaged neutron and density porosities provided an excellent match with core porosities.

Free-water Level and Trapping Mechanism Estimating the free-water level (FWL) position is critical for calculating water saturations using capillary pressures and the height above FWL. It has been recognized that the Hugoton field has a sloped gas-water contact, and we interpret a sloped FWL that is several hundreds of feet higher at the west updip margin than on the east downdip limits (Garlough and Taylor, 1941; Hubbert, 1953, 1967; Pippin, 1970; Sorenson, 2005). In this study, we have defined the gas-water contact as the lowest position in the reservoir that a well can produce gas economically, without substantial water, and the free-water level as the

FIGURE 34. Neutron-density porosity log crossplot of gas zones in the Towanda Limestone from the example Stevens County well (see Figure 33).

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 341

datum where gas-brine capillary pressure is zero. As shown in the section on petrophysics, initial reservoir desaturation may not occur for some lithofacies until several tens or hundreds of feet above the freewater level (threshold entry height). For typical reservoir rocks in the study area, packstone-grainstone 8–10% porosity, the FWL ranges from 50 to 70 ft (9 to 21 m) below the gas-water contact, a point at which the water saturation is approximately 70% (see Figure 27). The Hugoton gas reservoir is a dry-gas, pressuredepletion reservoir with very little or no support from the underlying aquifer. Vertical water flow is constrained by low vertical water permeabilities through low-porosity siltstone layers (k < 106 md [109 mm2] for $ < 4%) and by low water relative permeability in carbonates with low water saturation. However, below the transition zone, water can be produced freely, and reservoir pressures (600 – 700 psi; 4.1 – 4.8 MPa) approach regional hydrodynamic pressures for the depth (Sorenson, 2005). The low reservoir gas pressures (450 psi; 3.1 MPa) and subhydrostatic water pressures below the transition zone were proposed by Sorenson (2005) to be the result of water pressure equilibrating with reservoir rocks exposed at outcrop in eastern Kansas and gas cap expansion, and consequent pressure decrease. The Hugoton has long been considered a classic example of a giant stratigraphic trap (Garlough and Taylor, 1941; Parhman and Campbell, 1993) because of updip changes in lithofacies and petrophysical properties associated with these changes. However, dips on the apparent gas-water contact and FWL that cross stratigraphic boundaries cannot be fully explained by lateral heterogeneities. Hubbert (1953, 1967) proposed a conceptual model for the Hugoton being a hydrodynamic trap with trapping resulting from a hydraulic gradient coupled with permeability changes at the updip margin of the field. Pippin (1970) cited Hubbert’s hydrodynamics and updip pinch-outs of reservoir rock as the trapping mechanism. Olson et al. (1997) suggested that sealing faults, at least in the western part of the field in Stanton and Morton counties, Kansas, compartmentalize the lower Chase reservoirs with the compartments having dramatically different gas-water contacts that rise to the west. Sorenson (2005) suggested that the downdip flow of gas during expansion of the Hugoton gas bubble might be responsible for the gas-water contact geometry. Determining the mechanism for an uneven FWL was not an objective of our investigations, but FWL had to be established for the calculation of water

saturations using capillary pressure. Although others have presented general descriptions of the gas-water contact datum (e.g., Garlough and Taylor, 1941; Parhman and Campbell, 1993), it has not been rigorously defined by earlier workers. Our estimation of the FWL (Figure 35) was derived using a combination of four indicators: (1) base of lowest perforations; (2) formation fluid resistivity estimated from wire-line logs; (3) calculation of the FWL for an estimated original gas in place (OGIP); and (4) pressure measurements of deep-water productive intervals. Within the limits of the Panoma field, we based the depth of FWL on the average lowest reported productive perforations in the Council Grove (FWL = base of perforations + 70 ft [20 m]), assuming that operators have been efficient at identifying pay and avoiding water production. Along the eastern and western margin of the Hugoton in Kansas, where there is no underlying Council Grove production, and the western margin in Oklahoma, we used a formation fluid resistivity method for estimating the FWL at the field boundary. Here, FWL was estimated to be 30 ft (9 m) below the structural datum of the point in the Chase pay zones where the average apparent formation fluid resistivity (mix of all fluids) estimated from logs was at or near the resistivity of formation water. Limited data in the Oklahoma Panhandle required that FWL be estimated by backcalculating the FWL necessary to calculate an OGIP that would allow total gas production to equal 70% of OGIP. This method assumed that the Panhandle reservoir exhibited similar pressure depletion and gas production as reservoirs in Kansas. The FWL subsea depth is approximately 30 ft (9 m) at the east margin of the Hugoton and, moving west, remains relatively flat to the east margin of the Panoma, where it begins to rise at a rate of 15 ft/mi (2.85 m/km) to a datum of +250 ft (+80 m), where it then rises at 50 ft/mi (9.4 m/km) to a height of +1000 ft (+300 m) at the western margin of the Hugoton. The configuration closely parallels the gaswater contact described by Parhman and Campbell (1993), although our estimate places it 100 ft (30 m) lower at the east margin. Our estimated gas-water contact is +40 ft (12 m) at this position in the field (FWL + 70 ft [21 m]).

Static Model Construction The extremely large area (10,042 mi2; 26,008 km2), small XY cells (660  660 ft; 201  201 m) and relatively thin layers (169 layers, 3.3 ft [1 m] thick average; Figure 36) resulted in a 108-million-cell model that required subdividing the model into parts because of

342 / Dubois et al.

FIGURE 35. Free-water level (FWL) estimated for the Hugoton model area (A, B) is applicable for all production from the Chase but only for Council Grove production that is inside the Panoma. The entire Wolfcamp (Chase and Council Grove) is thought to have a common FWL across much of the area, but the Council Grove production outside the Panoma (in Oklahoma) may be related to a different FWL. The FWL is at near sea level at the east Hugoton margin, climbs gradually in a westerly direction to midfield, and then ascends more rapidly to a height of more than 1000 ft (300 m) above sea level at the west field margin. The zero datum of the height above FWL surface for the Chase (C) and Council Grove (D) correspond well to the edges of the Hugoton and Panoma field boundaries, respectively.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 343

FIGURE 36. Intersecting structural cross sections (near midfield) in the Hugoton model illustrate the 169 layers in the 24 zones from the combined six 3-D grids. Marine zones are more finely layered than the continental intervals (vertical exaggeration is 200).

permeability were calculated at the cell level using the petrophysical equations presented above. Water saturation was calculated at the cell level using capillary pressure curves developed for each cell based on the cell lithofacies, porosity, and height above free-water level. Finally, OGIP was calculated for a bottom-hole pressure equal to 450 psi (3.1 MPa) and compressibility index (Z) equal to 0.92.

RESULTS AND DISCUSSION

computational limitations. Because the main pay intervals are separated stratigraphically by intervals with relatively poor reservoir properties and lithofacies and property modeling is restricted by zones in the modeling process, we chose to subdivide the Chase and Council Grove model stratigraphically instead of geographically, with three multizone, multiformation models in the Chase and three in the Council Grove. Each of the six models were built with the same starting architecture and layering to facilitate cutting out selected parts of the six models and assembly into smaller models. These would have a complete vertical section of the reservoir system, but limited aerial extent, and be more easily managed for further analysis and reservoir simulation. The structural framework for the six models was based on a structural tops database for 8765 wells (Figure 20B). Our data set of tops in Oklahoma is less complete than it is in Kansas. Layering within the models used the following hierarchy: (1) division between formation and members, (2) further subdivision between continental and marine intervals, and (3) further subdivision into layers based on minimum vertical thickness of the key lithofacies in the node wells (Figure 20A). Layers averaged 2 ft (0.6 m) thick in the marine intervals and 4 ft (1.2 m) thick in the continental intervals. Horizontal and vertical

The accuracy and utility of the Hugoton geomodel can be measured by several metrics, including prediction accuracy of parameters like lithofacies, petrophysical properties, and OGIP at the well to field scale. The only direct measure for lithofacies is the comparison of predicted and core-defined lithofacies (Table 3). We can also qualitatively measure the validity of the lithofacies model by (1) comparing it with earlier work at smaller scales and (2) comparing the 3-D lithofacies patterns with depositional models that have been proposed for the area and for upper Paleozoic cyclic depositional systems in general. Measures of accuracy for any parameter (e.g., permeability) at the lease and field-scale are constrained by lack of data directly measured at these scales and the consequent need to compare other parameters, such as OGIP. OGIP requires integration of many variables, which may be inaccurate due to error in a single input variable or improper upscaling and integration of accurate, but different scale, variables. Ultimately, measures of accuracy and utility of a geomodel at the lease and field scales are commonly defined by comparison of predicted and measured production and pressure history, where the predicted pressure and production data are obtained from input of the static geomodel (the focus of this chapter) into a reservoir-flow simulator to obtain a dynamic model. The workflow involved in calibration of the geomodel with dynamic data is not discussed in

344 / Dubois et al.

this chapter or shown in Figure 12, but is an important part of testing and refining a geomodel. Dynamic modeling has been performed of 9- and 12-mi2 (23and 31-km2) areas, including histories for 28 and 37 wells, respectively, to test and refine the geomodel discussed here. These simulations are ongoing and are both generally validating the present geomodel and showing how uncertainty in some properties (e.g., free-water level) must be reduced to provide a model that is sufficiently constrained for use in accurately predicting field performance.

Model Lithofacies and Properties Lithofacies and property distribution in the 3-D Hugoton cellular geomodel presented here on a larger scale and with finer resolution are consistent with earlier work on the Hugoton (Garlough and Taylor, 1941; Hubbert, 1953; Pippin, 1970; Seimers and Ahr, 1990; Parham and Campbell, 1993; Fetkovitch et al., 1994; Oberst et al., 1994; Olson et al., 1997; Heyer; 1999; Sorenson, 2005). The full-field geomodel presented here reveals facies and property patterns that could not be identified at smaller scales. Figures 37 – 39 are a series of cross sections and map views of lithofacies and properties for the six individual models. General trends in thickness and lithofacies distribution are evident in the 3-D volume: continental rocks are thickest, marine carbonate intervals thin or pinch out at Hugoton’s western updip margin, and the relationship is nearly reciprocal basinward. The important reservoir lithofacies (grain-supported carbonate, dolomite, and marine and continental sandstone) are laterally extensive, and the marine carbonates, the primary pay zones, are separated by laterally continuous continental siltstone with poor vertical transmissibility. Large-scale sedimentation patterns, interpreted from lithofacies distribution, are striking when viewed at the scale made possible by the full field-scale model.

In cross section at the intercycle scale, backstepping of major lithofacies associated with changes in water depth and energy are evident in the Chase marine carbonate, particularly in the two dolomite lithofacies. The position on the shelf where continental siliciclastics thicken at a higher rate also backsteps in a similar manner. Marine carbonates in the middle three of the seven Council Grove cycles (Figure 37C) pinch out at the western field margin. The gradual shift in the paleoshoreline position, first to lower on the shelf and then back to a higher position, is believed to be related to a change in sea level amplitude. The symmetry demonstrated by sedimentation patterns around the middle cycle of the seven fourthorder Council Grove cycles suggests the possibility of higher order cyclicity in the latter part of the middle Paleozoic icehouse. One of the more striking aspects of the model from a reservoir perspective is the demonstration of lateral continuity in the lithofacies illustrated in Figure 38. County-scale connected volumes of the more significant reservoir lithofacies are subparallel to depositional strike and the field margin. Grainsupported packstones in the Crouse (Figure 38A, B1_LM, Council Grove) are found primarily in the eastern half of the field, whereas muddier lithofacies are dominant in the western, more sheltered shoreward part of the shelf. Continental sandstone (eolian) is limited to the northwestern updip margin in the Council Grove (Figure 38B), and shallow-marine and tidal-flat sandstone in the upper Chase is most abundant in the northwestern half of the field area (Figure 38C). The core to model lithofacies workflow was sufficiently robust to characterize smaller important lithofacies bodies in the reservoir system (Figure 38D –F). For example, dolomitized grainstone and packstone of a relatively thick (30 ft, 10 m) carbonate sand shoal system in the southern half of the field is known to be an important contributor to storage and flow

FIGURE 37. Lithofacies in stratigraphic cross sections across the Hugoton shelf (A) for the Chase (B) and Council Grove (C). Cross sections are 10 – 158 from being dip sections and are hung on the top of the Chase (B) and the Council Grove (C). Some key observations can be made: (1) In both the Chase and Council Grove, continental half-cycles (yellow-orange to red lithofacies) are thickest at the west field margin and thin basinward (southeasterly). The pattern for the marine half-cycles is the opposite and somewhat reciprocal relationship with the continental half-cycles. (2) Backstepping pattern in lithofacies distribution from one marine cycle to the next in the Chase. (3) Three Council Grove marine half-cycles pinch out near the west field margin, marking a paleoshoreline that appears to then move northwesterly (landward) upsection. (4) Trend in carbonate rock texture from mud dominated (landward) to grain dominated (basinward), especially in the Council Grove. Large-scale sedimentation patterns and distribution of resultant lithofacies (at the cycle scale) is largely a function of the position on the shelf and reflect the interaction of shelf geometry, sea level, and possibly, the proximity to siliciclastic sources. Lithofacies distribution and cycle stacking patterns at larger scales may be a function of lower order cyclicity and a shift from icehouse to greenhouse conditions (upward) during the Lower Permian.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 345

capacity in the Krider (R. Sorenson, 2004, personal communication). Modeling of this important lithofacies shows the lateral continuity of a 4  30-mi (6.5  50-km) sweet spot (porosity > 18%; Figure 38F).

Distribution of lithofacies-dependent properties (porosity, permeability, and water saturation) correlates well with lithofacies distribution (Figure 39). The best porosity and permeability coincide with

346 / Dubois et al.

FIGURE 38. Maps showing lithofacies distribution of selected lithofacies. Illustrated are map views of connected volumes TM generated in Schlumberger’s Petrel modeling application. Connected volumes are collections of touching cells in the cellular model having common properties and, for this model, help to demonstrate the remarkable lateral continuity of flow units in the Wolfcamp. (A) Thirty largest connected volumes in the Crouse limestone (B1_LM) packstone, packstone-grainstone (L7, blue), and fine-crystalline dolomite (L6, pink) having porosity greater than 8%. (B) Fifteen largest connected volumes in the Speiser shale (A1_SH) continental sandstone (L0) with porosity greater than 12%. (C) Twenty largest connected volumes of marine sandstone (L10) having porosity greater than 15%. (D) Top 20 connected volumes for Krider packstone, packstone-grainstone (L7, blue) and coarse-crystalline dolomite (L9, purple) having porosity greater than 16%. (E) Enlarged area of (D). (F) Same as (E) except for porosity greater than 18%. Grant and Stevens counties are outlined in green for reference in (D), (E), and (F).

the main reservoir lithofacies (Figure 37). Laterally extensive low-permeability intervals separate the relatively high-permeability pay zones of the layered reservoir system. Water saturations are high in the confining intervals, and the gas-water contact crosses stratigraphic boundaries at the east downdip margin as the pay intervals dip below the surface and on the

west where the free-water level is thought to rise more quickly than the rate of dip. Quantitative measures of lithofacies proportions (Table 3) in core, neural network-predicted lithofacies at node wells, upscaled at the node wells and in the 108-million-cell model, show consistency at the different scales, suggesting that the sample rate for

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 347

FIGURE 39. Property distribution in cellular Hugoton model in cross section. (A) Location of cross sections. (B) Chase Group stratigraphic cross section (datum is top of Chase). Horizontal permeability is shown in west – east section, and porosity (0 – 30%, yellow is 22%) is in the north – south section. (C) Chase Group through Easly Creek shale (B2_SH, Council Grove) water saturation. Free-water level is the base of the cross section on the west and east side and the base of the Easly Creek (B2_SH) in the middle, where the FWL is lower in the stratigraphic column (not able to display all models simultaneously). Free-water level crosses stratigraphic boundaries in both updip and downdip positions. Highest permeability (Kxy) and porosity (Phi), and lowest water saturation (Sw) is found in marine carbonates and sandstones. Continental siltstones separating the marine carbonates are the intervals with Kxy and Phi and higher Sw.

training and lithofacies prediction in each of the three steps was sufficient. Slight variations in measures are likely to be related to sample distribution in core versus the node wells and the increase in scales from 0.5 ft (0.15 m) to considerably thicker cells (layers). Several factors related to the nature, geometry, and distribution of the predicted lithofacies and the variables chosen to predict lithofacies

are believed to have enabled the success of the neural networks to predict lithofacies at the node wells and the modeling of lithofacies between nodes. Lithofacies classes were chosen to maximize differences in the signature of wire-line-log variables. Geologic constraining variables (e.g., relative position curve and marine-nonmarine curve) captured and leveraged geologic information such as the predictable vertical

348 / Dubois et al.

succession of lithofacies in the sedimentary cycles and the primary depositional environment. Extensive lateral continuity and lithofacies-body sizes being much larger than the lateral spacing of node wells helped model the lithofacies between nodes. Significantly, there was adequate core control to appropriately sample the reservoir system for the lithofacies training set. Without core, the model could not have been built. The most common misallocations of lithofacies were prediction of grainstone (L8) as wackestone or packstone (L5 and L7), predictions of carbonate mudstone (L4) as wackestone (L5), and predictions of the continental sandstone (L0) as coarse siltstone (L1). These incorrect lithofacies predictions reflect overlaps of the log characteristics of the lithofacies. Fortunately, the lithofacies involved are also sufficiently similar in their petrophysical properties that the resulting distributions of porosity, permeability, and water saturation resulting from random misallocations were not significantly different from correct property distributions. For example, a grainstone (L8) having 10% porosity and at a position 300 ft (90 m) above FWL misclassified as a packstone (L7) would be assigned a permeability of 0.35 md instead of 0.70 md (Figure 24) and a water saturation of 24% instead of 17% (Figure 28). The majority vote upscaling of lithofacies from the 0.5-ft (0.15-m) sampling interval to the thickness of the model layers intersecting each well did not significantly alter the lithofacies populations. Using this methodology, we would not expect to see a significant difference in lithofacies populations unless certain lithofacies typically occurred in thin bodies separated by fairly large vertical intervals, resulting in a systematic underrepresentation of those lithofacies in the upscaled results. Finally, the lithofacies proportions in the full 3-D model closely reflect those for the upscaled well cells. This is not surprising because the stochastic indicator simulation attempts to match the proportions observed in the conditioning data. The distribution of our node wells was sufficiently uniform such that the lithofacies proportions in the conditioning set were very similar to those in the reservoir system.

Controls on Petrophysical Properties The petrophysical analysis indicates that accurate reservoir properties prediction requires input of lithofacies, use of properties that represent reservoir conditions, and filtering of full-diameter data to avoid microfractured core. The close correspondence of DST

permeabilities and upscaled plug-scale permeabilities is interpreted to indicate that production from many wells is controlled by matrix permeability and not fractures. Comparison of petrophysical properties among lithofacies indicates that permeability increases and threshold entry height and transition zone thickness decrease with increasing energy in the depositional environment and corresponding decrease in mud and silt matrix. Within both the continental and marine siliciclastics, permeability increases with increasing mean grain size at any given porosity. The marine, very fine-grained sandstone (L10) exhibits an order of magnitude greater permeability than marine siltstone (L3). However, continental, very fine-grained sandstone (L0) exhibits only approximately 2.5 greater permeability at any given porosity than continental coarse-grained siltstone (L1) and 5 greater permeability than fine–medium-grained siltstone (L2). This difference is likely because of poorer sorting of the continental sandstone compared to the marine sandstone. The poor sorting of the continental siliciclastics would be expected to result in a wide pore-throat size distribution, which is reflected in the high capillarypressure transition zones of these rocks. Limestone permeabilities exhibit subparallel permeability-porosity trends among lithofacies (Figure 24) and increasing mean porosity (Figure 21), upper porosity range, and permeability from mudstone through packstone-grainstone. These changes result in a corresponding increase in threshold entry heights and water saturations at any given height above free water. The differences in the capillary-pressure properties between lithofacies decrease as porosity and permeability increase. Both threshold entry heights and transition zone heights for the limestone are less than for continental siliciclastics of similar porosity.

Original Gas in Place Comparison of estimated OGIP by others, production data, and OGIP from calculations in the 108-millioncell model suggests that the geomodel successfully models the Hugoton, particularly in the center of the field where control on the free-water level is greatest. The geomodel OGIP is calculated to be 21.8 tcf (0.62 trillion m3) and have a hydrocarbon pore volume of 676 bcf (19.1 billion m3, 120 billion bbl) in Grant and Stevens counties, Kansas, for the Chase (Hugoton) and Council Grove (Panoma) intervals (Table 5; Figure 40). Cumulative gas production for the area is 14.1 tcf (0.4 trillion m3) or 65% of calculated OGIP,

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 349

Table 5. Estimated original gas in place (OGIP).* Group Chase

Council Grove

Total

Zone

OGIP (bcf)

OGIP (109 m3)

HCPV (bcf)

HCPV (109 m3)

Herington Krider Odell Winfield Gage Towanda Holmesville Fort Riley Matfield Wreford A1_SH A1_LM B1_SH B1_LM B2_SH B2_LM B3_SH B3_LM B4_SH B4_LM B5_SH B5_LM C_SH C_LM

1227 2795 295 3215 807 4686 663 5212 127 1048 331 656 76 143 9 167 56 34 67 22 3 113 2 20 21,775

34.8 79.1 8.4 91.0 22.9 132.7 18.8 147.6 3.6 29.7 9.4 18.6 2.2 4.1 0.3 4.7 1.6 1.0 1.9 0.6 0.1 3.2 0.1 0.6 589.7

38.1 86.7 9.2 99.8 25.0 145.5 20.6 161.8 3.9 32.5 10.3 20.4 2.4 4.4 0.3 5.2 1.7 1.0 2.1 0.7 0.1 3.5 0.1 0.6 675.9

1.08 2.46 0.26 2.83 0.71 4.12 0.58 4.58 0.11 0.92 0.29 0.58 0.07 0.13 0.01 0.15 0.05 0.03 0.06 0.02 0.00 0.10 0.00 0.02 19.1

*By zone for part of the Hugoton model covered by Grant and Stevens counties, Kansas, for 450 psi (3103 kPa) initial bottom-hole pressure.

slightly low compared to earlier work by others. For the Chase in Kansas, Oberst et al. (1994) estimated OGIP volumetrically at 31.1 tcf (0.88 trillion m3), whereas Olson et al. (1997) placed it at 34.5 – 37.8 tcf (0.98 – 1.1 trillion m3). Because their estimates were for different reservoir volumes than ours (Chase in Kansas versus Chase and Council Grove in two counties in Kansas), we cannot compare directly, but assuming similar reservoir performance, we can compare the estimates on the basis of production efficiency. The ratio of Chase cumulative production to date (24.8 tcf; 0.7 trillion m3) to OGIP is 79.7% by Oberst et al. (1994) and 65.6– 71.9% by Olson et al. (1997). Our estimate for the entire Wolfcamp reservoir volume (65%) is closer to the Olson et al. (1997) estimate. In Kansas, 89.2% of the Hugoton – Panoma production is attributed to the Chase and 10.8% to the Council Grove, although we believe the two behave as one reservoir system.

It is important to note that estimation of OGIP using the matrix capillary-pressure method employed in this study is influenced by natural variance in the capillary-pressure curves and the determination of free-water level. Natural variance in capillary-pressure curves can result in a one standard deviation confidence interval for predicted water saturation of greater than 10% of the saturation value (e.g., Sw = 10% results in 9% < Sw < 11% or, for S w = 80%, 72% < S w < 88%), which results in a one standard deviation confidence interval for predicted OGIP of approximately 3%. Change in the free-water level results in little change in water saturation for intervals greater than 300 ft (91 m) above FWL, but can have significant influence on intervals that are within their transition zone and have rocks that exhibit transition zones that are only tens of feet high. For these intervals, shift in the FWL by a few tens of feet can result in significant

350 / Dubois et al.

FIGURE 40. Map views of original gas in place (OGIP) for Grant and Stevens counties, Kansas (A), are the two most prolific counties in Hugoton. Hydrocarbon height (HCH) at surface conditions for Chase Group (B), Council Grove Group (C), and combined, Wolfcamp (D). Wolfcamp volume map varies little from the Chase because the Council Grove volume is small relative to the Chase (note the smaller scale range and contour interval for the Council Grove). Areas of high Chase OGIP are coincident with Krider ooid shoal complex illustrated in Figure 38D – F.

water saturation change and, consequently, OGIP changes. This was important for the Council Grove OGIP estimate and is still being refined.

CONCLUSIONS The more than 100-million-cell, 10,000-mi 2 (26,000-km2), 3-D geologic and petrophysical property geomodel of the Hugoton presented in this study demonstrates the application of a detailed reservoir

characterization and modeling workflow for a giant field. Core-based calibration of neural net prediction of lithofacies using wire-line-log signatures, coupled with geologicconstraining variables, provided accurate lithofacies models at well to field scales. Differences in petrophysical properties among lithofacies and within a lithofacies among different porosities illustrate the importance of integrated lithologic-petrophysical modeling and of the need for closely defining these properties and their relationships. Lithofacies models, coupled with lithofacies-dependent petrophysical properties, allowed the construction of a 3-D geomodel for the Hugoton that has been effective at the well, section (1 mi2; 2.6 km2) and multisection scales. The model provided a tool to predict lithofacies and petrophysical properties distribution, water saturations, and OGIP. It will likely provide a quantitative basis for evaluating remaining gas in place, particularly in low-permeability intervals, and help direct field management and production practices that will potentially enhance ultimate recovery. The reservoir characterization and modeling from pore to field scale discussed provides a comprehensive lithological and petrophysical view of a mature giant Permian gas system. Both the knowledge gained and the techniques and workflow employed have implications for understanding and modeling similar reservoir systems worldwide.

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 351

ACKNOWLEDGMENTS We are thankful for the support from partners in the Hugoton Asset Management Project, including Anadarko Petroleum Corporation, BP America Production Company, ConocoPhillips Company, Cimarex Energy Company, EOG Resources Inc., ExxonMobil Production Company, Medicine Bow Energy Corporation, Osborn Heirs Company, OXY U.S.A., Inc., and Pioneer Natural Resources U.S.A., Inc., and from the Kansas Geological Survey. We thank Nathan Winters for his assistance on core descriptions, Raymond Sorenson for insightful discussions, Shane Seals in work on earlier models, and geoPlus Corporation and Schlumberger for providing the software. Reviewers Eugene Rankey, Sean Guidry, and Krishnan Srinivasan provided helpful comments and suggestions that improved this manuscript.

REFERENCES CITED Asquith, G., and D. Krygowski, 2004, Basic well log analysis, 2d ed.: AAPG Methods in Exploration, v. 16, 244 p. Baars, D. L., ed., 1994, Revision of stratigraphic nomenclature in Kansas: Kansas Geological Survey Bulletin, v. 230, 80 p. Berg, R. R., 1975, Capillary pressures in stratigraphic traps: AAPG Bulletin, v. 59, p. 939 – 956. Boardman II, D. R., and M. K. Nestell, 2000, Outcrop-based sequence stratigraphy of the Council Grove Group of the midcontinent, in K. S. Johnson, ed., Platform carbonates in the southern midcontinent, 1996 Symposium: Oklahoma Geological Survey Circular 101, p. 275– 306. Bohling, G. C., and J. H. Doveton, 2000, Kipling.xla: An Excel add-in for nonparametric regression and classification, Kansas Geological Survey: http://www.kgs .ku.edu/software/Kipling/Kipling1.html (accessed December 31, 2005). Byrnes, A. P., 2003, Aspects of permeability, capillary pressure, and relative permeability properties and distribution in low-permeability rocks important to evaluation, damage, and stimulation: Proceedings of the Rocky Mountain Association of Geologists Petroleum Systems and Reservoirs of Southwest Wyoming Symposium, Denver, Colorado, September 19, 2003, 12 p. Byrnes, A. P., M. K. Dubois, M. Magnuson, 2001, Western tight gas carbonates: Comparison of Council Grove Group, Panoma field, southwest Kansas and western low permeability sandstones (abs.): AAPG Annual Meeting Program, v. 10, p. A31. Caldwell, C. D., 1991, Cyclic deposition of the Lower Permian, Wolfcampian, Chase Group, western Guymon–

Hugoton field, Texas County, Oklahoma, in W. L. Watney, A. W. Walton, C. G. Caldwell, and M. K. Dubois, organizers, Midcontinent Core Workshop on Integrated Studies of Petroleum Reservoirs in the Midcontinent: Midcontinent AAPG Section Meeting, Wichita, Kansas, p. 57 – 75. Deutsch, C. V., and A. G. Journel, 1998, Geostatistical software library and user’s guide: Oxford, Oxford University Press, 369 p. Dubois, M. K., and R. H. Goldstein, 2005, Accommodation model for Wolfcamp (Permian) redbeds at the updip margin of North America’s largest onshore gas field (abs.): Proceedings AAPG 2005 Annual Convention, June 19– 21, Calgary, Alberta, Canada, and Kansas Geological Survey Open-file Report 2005-25: http://www .kgs.ku.edu/PRS/AAPG2005/2005-25/index.html (accessed December 31, 2005). Dubois, M. K., A. P. Byrnes, G. C. Bohling, S. C. Seals, and J. H. Doveton, 2003, Statistically-based lithofacies predictions for 3-D reservoir modeling: Examples from the Panoma (Council Grove) field, Hugoton embayment, southwest Kansas (abs.): AAPG Annual Meeting Program, v. 12, p. A44, and Kansas Geological Survey Open-file Report 2003-30, 3 panels: http://www .kgs.ku.edu/PRS/publication/2003/ofr2003-30/index .html (accessed October 10, 2005). Dubois, M. K., A. P. Byrnes, and G. C. Bohling, 2005, Geologic model for the giant Hugoton and Panoma fields (abs): AAPG Midcontinent Section Meeting, Oklahoma City, Oklahoma: http://www.kgs.ku.edu /PRS/Poster/2005/MidcontAAPG/index.html (accessed October 10, 2005). Dunham, R. J., 1962, Classification of carbonate rocks according to depositional texture, in W. E. Ham, ed., Classification of carbonate rocks: AAPG Memoir 1, p. 108 –121. Dutton, S. P., E. K. Kim, C. L. Broadhead, W. D. Raatz, S. C. Ruppel, and C. Kerans, 2005, Play analysis and digital portfolio of major oil reservoirs in the Permian Basin: Bureau of Economic Geology Reports of Investigations, RI0271, 302 p. Fetkovitch, M. J., D. J. J. Ebbs Jr., and J. J. Voelker, 1994, Multiwell, multilayer model to evaluate infill-drilling potential in the Oklahoma Hugoton field: 65th Society of Petroleum Engineers Annual Technical Conference and Exhibition, New Orleans, Louisiana, SPE Paper 20778, p. 162 – 168. Folk, R. L., 1954, The distinction between grain size and mineral composition in sedimentary rock nomenclature: Journal of Geology, v. 62, p. 344 – 359. Garlough, J. L., and G. L. Taylor, 1941, Hugoton gas field, Grant, Haskell, Morton, Stevens, and Seward counties, Kansas, and Texas county, Oklahoma, in A. I. Levorsen, ed., Monograph Stratigraphic type oil fields: AAPG, p. 78 – 104. Gaymard, R., and A. Poupon, 1968, Response of neutron and formation density logs in hydrocarbon bearing formations: The Log Analyst, v. 9, no. 5, p. 3 – 12. George, B. K., C. Torres-Verdin, M. Delshad, R. Sigal, F.

352 / Dubois et al. Zouioueche, and B. Anderson, 2004, Assessment of insitu hydrocarbon saturation in the presence of deep invasion and highly saline connate water: Petrophysics, v. 45, no. 2, p. 141– 156. Grammer, G. M., G. P. Eberli, F. S. P. van Buchem, G. M. Stevenson, and P. Homewood, 1996, Application of high-resolution sequence stratigraphy to evaluate lateral variability in outcrop and subsurface; Desert Creek and Ismay intervals, Paradox basin, in M. W. Longman and M. D. Sonnenfeld, eds., Paleozoic systems of the Rocky Mountain region: Rocky Mountain Section Society for Sedimentary Geology, p. 235 – 266. Hastie, T., R. Tibshirani, and J. Friedman, 2001, The elements of statistical learning: Data mining, inference, and prediction: New York, Springer, 533 p. Hecker, M. T., M. E. Houston, and J. D. Dumas, 1995, Improved completion designs in the Hugoton field utilizing multiple gamma emitting tracers: Society of Petroleum Engineers Annual Technical Conference and Exhibition, Dallas, Texas, SPE Paper 30651, p. 223 –235. Hemsell, C. C., 1939, Geology of Hugoton gas field of southwestern Kansas: AAPG Bulletin, v. 23, no. 7, p. 1054– 1067. Heyer, J. F., 1999, Reservoir characterization of the Council Grove Group, Texas County, Oklahoma, in D. F. Merriam, ed., AAPG Midcontinent Section Meeting Transactions, Geosciences for the 21st Century, p. 71 – 82. Hough, E. W., M. J. Rzasa, and B. B. Wood, 1951, Interfacial tensions at reservoir pressures and temperatures; apparatus and the water-methane system: American Institute of Mechanical Engineering Petroleum Transactions, Technical Publication 3019, v. 192, p. 57 – 60. Hubbert, M. K., 1953, Entrapment of petroleum under hydrodynamic conditions: AAPG Bulletin, v. 37, p. 1954 – 2026. Hubbert, M. K., 1967, Application of hydrodynamics to oil exploration: 7th World Petroleum Congress Proceedings, Mexico City: Amsterdam, Elsevier Publishing Co. Ltd, v. 1B, p. 59 – 75. Jennings Jr., H. Y., and G. H. Newman, 1971, The effect of temperature and pressure on the interfacial tension of water against methane-normal decane mixtures: Transactions of the American Institute of Mining Metallurgical and Petroleum Engineers, v. 251, p. 171 – 175. Kluth, C. F., 1986, Plate tectonics of the ancestral Rocky Mountains, in J. A. Peterson, ed., Paleotectonics and sedimentation of the Rocky Mountains, United States: AAPG Memoir 41, p. 353 – 369. Kluth, C. F., and P. J. Coney, 1981, Plate tectonics of the ancestral Rocky Mountains: Geology, v. 9, no. 1, p. 10 – 15. Konnert, G., A. M. Afifi, S. A. Al-Hajri, K. de Groot, A. A. Al Naim, and H. J. Droste, 2001, Paleozoic stratigraphy and hydrocarbon habitat of the Arabian plate, in M. W. Downey, J. C. Threet, and W. A. Morgan, eds., Petroleum provinces of the twenty-first century: AAPG Memoir 74, p. 483 – 515. Luczaj, J. A., and R. H. Goldstein, 2000, Diagenesis of the

Lower Permian Krider Member, southwest Kansas, U.S.A.: Fluid-inclusion, U-Pb, and fission-track evidences for reflux dolomitization during latest Permian time: Journal of Sedimentary Research, v. 70, no. 3, p. 762 – 773. Mason, J. W., 1968, Hugoton and Panhandle field, Kansas, Oklahoma and Texas, in W. B. Beebe and B. F. Curtis, eds., Natural gases of North America, v. 2: AAPG Memoir 9, p. 1539 – 1547. Mazzullo, S. J., C. S. Teal, and C. A. Burtnett, 1995, Facies and stratigraphic analysis of cyclothemic strata in the Chase Group (Permian Wolfcampian), south-central Kansas, in N. J. Hyne, ed., Sequence stratigraphy of the mid-continent: Tulsa Geological Society Special Publication 4, p. 217 – 248. McGillivray, J. G., and M. I. Husseini, 1992, The Paleozoic petroleum geology of central Arabia: AAPG Bulletin, v. 76, p. 1491 – 1506. Oberst, R. J., P. P. Bansal, and M. F. Cohen, 1994, 3-D reservoir simulation results of a 25-square mile study area in Kansas Hugoton gas field: Society of Petroleum Engineers Mid-Continent Gas Symposium, SPE Paper 27931, p. 137 – 147. Olson, T. M., J. A. Babcock, and P. D. Wagner, 1996, Geologic controls on reservoir complexity, Hugoton giant gas field, Kansas, in D. L. Swindler and C. P. Williams, compilers, AAPG Mid-continent Section Meeting Transactions: Mid-continent Section, p. 189–198. Olson, T. M., J. A. Babcock, K. V. K. Prasad, S. D. Boughton, P. D. Wagner, M. K. Franklin, and K. A. Thompson, 1997, Reservoir characterization of the giant Hugoton gas field, Kansas: AAPG Bulletin, v. 81, p. 1785 – 1803. Olszewski, T. D., and M. E. Patzkowsky, 2003, From cyclothems to sequences: The record of eustacy and climate on an icehouse eperic platform (Pennsylvanian – Permian), North American Midcontinent: Journal of Sedimentary Research, v. 73, no. 1, p. 15 – 30. Parham, K. D., and J. A. Campbell, 1993, PM-8. Wolfcampian shallow shelf carbonate — Hugoton embayment, Kansas and Oklahoma, in D. G. Bebout, ed., Atlas of major midcontinent gas reservoirs: Gas Research Institute, p. 9 – 12. Parrish, J. T., 1995, Geologic evidence of Permian climate, in P. A. Scholle, T. M. Peryt, and D. S. Ulmer-Scholle, eds., The Permian of northern Pangea: v. I. Paleogeography, paleoclimate, stratigraphy: Berlin, Germany, Springer-Verlag, p. 53 – 61. Parrish, J. T., and E. Peterson, 1988, Wind directions predicted from global circulation models and wind directions determined from eolian sandstones of the western United States — A comparison: Sedimentary Geology, v. 56, p. 261 – 282. Perry, W. J., 1989, Tectonic evolution of the Anadarko basin region, Oklahoma: U.S. Geological Survey Bulletin, v. 1866-A, p. A1 – 16. Pippin, L., 1970, Panhandle–Hugoton field, Texas–Oklahoma–Kansas— The first fifty years, in M. T. Halbouty,

Geologic and Petrophysical Modeling of the Hugoton Gas Field, Kansas and Oklahoma / 353 ed., Geology of giant petroleum fields: AAPG Memoir 14, p. 204–222. Puckette, G. R., D. R. Boardman II, and Z. Al-Shaieb, 1995, Evidence for sea-level fluctuation and stratigraphic sequences in the Council Grove Group (Lower Permian) Hugoton embayment, southern mid-continent, in N. J. Hyne, ed., Sequence stratigraphy of the mid-continent: Tulsa Geological Society Special Publication 4, p. 269– 290. Purcell, W. R., 1949, Capillary pressure — Their measurements using mercury and the calculation of permeability therefrom: American Institute of Mechanical Engineers Petroleum Transactions, v. 186, p. 39 – 48. R Development Core Team, 2005, R: A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, Austria: http://www .R-project.org (accessed December 31, 2005). Rankey, E. C., 1997, Relations between relative changes in sea level and climate shifts; Pennsylvanian – Permian mixed carbonate-siliciclastic strata, western United States: Geological Society of America Bulletin, v. 109, no. 9, p. 1089 – 1100. Rascoe Jr., B., 1968, Permian system in western midcontinent: Mountain Geologist, v. 5, p. 127 – 138. Rascoe Jr., B., and F. J. Adler, 1983, Permo-Carboniferous hydrocarbon accumulations, midcontinent, U.S.A.: AAPG Bulletin, v. 67, p. 979 – 1001. Ryan, T. C., M. J. Sweeney, and W. H. Jamieson Jr., 1994, Individual layer transient tests in low-pressure, multilayered reservoirs: Society of Petroleum Engineers Mid-Continent Gas Symposium, Amarillo, Texas, SPE Paper 27928, p. 99 – 113. Scotese, C. R., 2004, A continental drift flipbook: Journal of Geology, v. 112, p. 729 – 741. Siemers, W. T., and W. M. Ahr, 1990, Reservoir facies, pore characteristics, and flow units: Lower Permian Chase

Group, Guymon – Hugoton field, Oklahoma, Society of Petroleum Engineers Proceedings 65th Annual Technical Conference and Exhibition, New Orleans, Louisiana, September 23 – 26, 1990, SPE Paper 20757, p. 417 – 428. Soreghan, G. S., 2002, Sedimentologic-magnetic record of western Pangean climate in upper Paleozoic loessites (lower Cutler beds, Utah): Geological Society of America Bulletin, v. 114, no. 8, p. 1019 – 1035. Sorenson, R. P., 2005, A dynamic model for the Permian Panhandle and Hugoton fields, western Anadarko basin: AAPG Bulletin, v. 89, no. 7, p. 921 – 938. Venables, W. N., and B. D. Ripley, 2002, Modern applied statistics with S, 4th ed.: New York, Springer, 512 p. Washburn, E. W., 1921, A method of determining the distribution of pore sizes in a porous material: Proceedings of the National Academy of Sciences, v. 7, p. 115 – 116. Weber, L. J., F. M. Wright, J. F. Sarg, E. Shaw, L. P. Harman, J. B. Vanderhill, and D. A. Best, 1994, Reservoir delineation and performance; application of sequence stratigraphy and integration of petrophysics and engineering data, Aneth field, southeast Utah, U.S.A., in E. L. Stout and P. M. Harris, eds., Hydrocarbon reservoir characterization; geologic framework and flow unit modeling: Society of Sedimentary Geology, p. 1 – 29. Winters, N. D., M. K. Dubois, and T. R. Carr, 2005, Depositional model and distribution of marginal marine sands in the Chase Group, Hugoton gas field, southwest Kansas and Oklahoma Panhandle (abs.): AAPG Midcontinent Section Meeting, Oklahoma City, Oklahoma: http://www.kgs.ku.edu/PRS/Poster/2005/MidcontAAPG /index.html (accessed October 10, 2005). Zeller, D. E., ed., 1968, The stratigraphic succession in Kansas: Kansas Geological Survey Bulletin, v. 169, 81 p.

10

Ruppel, S. C., and R. H. Jones, 2006, Key role of outcrops and cores in carbonate reservoir characterization and modeling, Lower Permian Fullerton field, Permian basin, United States, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 355 – 394.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling, Lower Permian Fullerton Field, Permian Basin, United States Stephen C. Ruppel Bureau of Economic Geology, Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, U.S.A.

Rebecca H. Jones Bureau of Economic Geology, Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, U.S.A.

ABSTRACT

T

he analysis of reservoir sequence and cycle stratigraphy, of depositional and diagenetic facies, and of the interrelationships between these attributes and reservoir properties is key to the construction of an accurate reservoir framework needed for reservoir modeling and improved imaging of remaining hydrocarbons. Fundamental steps in such a rock-based process of model construction applied at Fullerton Clear Fork field, a shallow-water carbonate platform reservoir of middle Permian age, included (1) creating and applying an analogous outcrop depositional model; (2) describing and interpreting subsurface core and log data in terms of this initial model; (3) defining the sequence-stratigraphic architecture of the reservoir section; (4) developing a cyclebased reservoir framework; and (5) defining controls, interrelationships, and distribution of porosity and permeability. Data used in this analysis included cores, thin sections, three- and two-dimensional seismic data, borehole image logs, and outcrop models. Key geological elements addressed and incorporated into the models include stratal architecture, differential dolomitization, karst fill, mineralogical variations, and rock-fabric distribution. These components were used to constrain interpretation and definition of flow units, permeability distribution, and saturation.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215882M88698

355

356 / Ruppel and Jones

In addition to resulting in improved and more geologically realistic reservoir models, the rock-based methods used in this study provide key insights into the formation, characterization, and interpretation of carbonate platform reservoirs; these insights have widespread application worldwide.

INTRODUCTION The Clear Fork Group (Figure 1) in the Permian basin comprises a thick (as much as 800 m [2500 ft]) succession of dominantly shallow-water platform carbonates that were deposited across west Texas and New Mexico during the Early Permian (Leonardian). Reservoirs developed in these carbonates (Figure 2) have accounted for more than 3.2 billion bbl of oil production (Dutton et al., 2005), more than 10% of the total recovered from the Permian basin to date. Despite this substantial production, estimates of original oil in place (OOIP) indicate that, overall, Leonardian reservoirs contained more than 14.5 billion bbl of oil at discovery. Recovery efficiency is thus only about 22%, considerably below the 32% average for carbonate reservoirs in the Permian basin (Tyler and Banta, 1989; Holtz and Garrett, 1990). Recovery from the shallow-water platform reservoirs of the Clear Fork Group has been even less efficient. Holtz et al. (1992) estimated a recovery efficiency of only 18% of OOIP for these reservoirs. To recover the remaining oil in Clear Fork reservoirs, operators must turn to increasingly sophisticated recovery technologies, e.g., water flooding, gas injection, horizontal wells, etc. To effectively deploy these technologies, however, it is critical that an accurate reservoir framework first be constructed to form

the basis for modeling and interpreting past, present, and future recovery operations. This report details approaches used to develop such a framework in the Fullerton Clear Fork field in west Texas (Figures 2, 3). By any standard, Fullerton is a giant. As of 2002, cumulative production stood at about 310 million bbl, about 18% of the approximately 1.69 billion bbl of OOIP (Wang and Lucia, 2004). Like many mature Clear Fork reservoirs in the Permian basin, Fullerton field has undergone a major decline in oil production rate and a major increase in water production rates for several years. An understanding of the depositional and diagenetic facies, the cycle and sequence stratigraphy, and the architecture of these stratigraphic elements is the crucial first step to determining the probable oil distribution at discovery and defining the best strategies for recovering the sizeable remaining oil volume.

METHODS Data and interpretations presented in this report are based on investigation of 29 cores totaling 4384 m (14,383 ft), the examination of more than 1700 rock thin sections, and the correlation of approximately 45,000 stratigraphic tops in more than 800 wells. The basic procedure followed in this study to develop a full-field sequence-stratigraphic model and reservoir framework is as follows: (1) identify depositional facies and vertical facies-stacking patterns and cycles in cores; (2) calibrate facies and cycle patterns to wire-line logs; (3) construct two-dimensional (2-D) cross sections of core/ log data sections; (4) define

FIGURE 1. Leonardian stratigraphic section in the Permian basin including the Clear Fork Group and analogous units in New Mexico and in outcrop. Productive interval at Fullerton field is shown in color.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 357

FIGURE 2. Regional map of the Permian basin showing the location of the Fullerton field and major reservoirs producing from Leonardian rocks. 2-D cycle and sequence stratigraphy from core and log sections; and (5) extrapolate 2-D cycle and sequence correlations into three-dimensional (3-D) space by correlating well logs. Stratigraphic sequences were tied to available 3-D seismic data to check their accuracy and geometry. Conventional core analysis data were available for all cores; these data were used to determine relationships between facies, cyclicity, and porosity and permeability development. Figure 3 shows the distribution of cores and 3-D seismic data in Fullerton field.

PREVIOUS WORK Mazzullo (1982) and Mazzullo and Reid (1989) presented overviews of lower Leonardian stratigraphy and depositional systems in the Midland basin.

Presley and McGillis (1982; see also Presley, 1987) documented the highly cyclic, predominantly evaporitic facies of the upper Leonardian Glorieta and Upper Clear Fork units in the Texas Panhandle. Ruppel (1992, 2002) described the facies, cyclicity, and diagenesis in the Glorieta and Upper Clear Fork at Monahans Clear Fork field on the Central Basin platform and postulated that reservoir development was caused by cyclic deposition and diagenesis driven by episodic sea level rise and fall. Atchley et al. (1999) described Clear Fork facies at Robertson field at the north end of the Central Basin platform and proposed a similar model for structural control over facies deposition and reservoir development. Ruppel et al. (2000) described outcrop equivalents of the producing subsurface Clear Fork reservoirs from outcrops in the Sierra Diablo Mountains of west Texas. Kerans et al. (2000)

358 / Ruppel and Jones

FIGURE 3. Structure of the Fullerton Clear Fork field. Datum is a prominent subtidal flooding event near the base of the Tubb Formation (see Figures 4, 5).

documented the depositional setting, facies, and architecture of the Abo from outcrops in the Sierra Diablo (Hudspeth and Culberson counties, Texas) and showed that karsting has had a major effect on both the Abo and the overlying Lower Clear Fork succession.

GEOLOGIC SETTING The Fullerton Clear Fork field is the largest of a large number of fields developed in the Leonardian series on the Central Basin platform of the Permian basin (Figure 2). The field covers an area of about 14,000 ha (35,000 ac); the reservoir column averages

about 210 m (700 ft) in thickness at a depth of about 2040 m (6700 ft) (Figures 2, 3). In some parts of the Permian basin, oil production is obtained from the entire Leonardian section (Figure 1). The productive reservoir section at Fullerton, however, is essentially restricted to the Lower Clear Fork, Wichita, and Abo stratigraphic units (Figure 4), although very minor production has been reported locally from the Upper Clear Fork section. By far, the bulk of the oil production has come from the lower two-thirds of the Lower Clear Fork and the Wichita sections of the reservoir. Even where oil saturated, the Abo has proven difficult to exploit in part because of its active water drive (as opposed to the pressure depletion drive that characterizes the overlying parts of the reservoir) and the fact that it is at or near the oil-water contact throughout most of the field. Regionally, the Leonardian is dominated by shallowwater platform carbonates. Each of the component stratigraphic units contains updip peritidal tidal-flat carbonates and downdip subtidal carbonates. Mineralogy in each is dominated by dolomite and anhydrite. Limestone is relatively uncommon; however, where present, it is most common in the lower part of the Leonardian section and in distal, downdip sections (Ruppel, 2002). The reservoir section at Fullerton is generally consistent with this regional pattern, but it does contain a higher volume of limestone than most other Leonardian platform reservoir successions in the Permian basin. Structurally, Fullerton field is developed over a large, compound structural high (Figure 3) that reflects deep-seated faulting and differential uplift of the area that began in the Pennsylvanian (Jones and Ruppel, 2004). Deeper oil production comes from block-faulted Silurian (Wristen Group) carbonates in the southern and central parts of the fields. The Clear Fork reservoir seal is provided by evaporite-rich carbonates of the Upper Clear Fork and Glorieta. The reservoir is currently drilled to well spacings of 16 to 4 ha (40 to 10 ac) and is under active water flood. The greatest well density is in the northern part of the field (mostly 8- and 4-ha (20- and 10-ac) well spacings); poorer well control exists in the southern half of the field. In general, these closely spaced wells provide good control for definition of stratigraphic horizons and reservoir attributes. In many parts of the field, however, especially along the western edge of the field and in the southern half of the field, wells are represented only by poor-quality (old gamma-ray and neutron) logs and cannot be correlated or interpreted with any precision.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 359

FIGURE 4. Type log of producing reservoir section at Fullerton Clear Fork field showing cyclicity and general facies distribution. Shaded areas of the PE log indicate the local presence of limestone in the dolomite-dominated section. Zones of elevated corrected gamma-ray (CGR) log response indicate the presence of silt and clay associated with tidal-flat facies. SGR = spectral gamma ray; PE = photoelectric; Neu = neutron; Den = density.

karsted, platform-margin subtidal carbonates of the Abo (Fitchen et al., 1995; Kerans et al., 2000). Styles of depositional architecture and facies development are very representative of subsurface succession and, thus, form excellent, if not essential, models for interpreting sparser subsurface data sets. Key observations established from these outcrops are described below.

Abo

OUTCROP ANALOGS Studies of analogous reservoir outcrops in the Sierra Diablo in Hudspeth and Culberson counties, Texas (Figure 2), provide important insights into the geological controls on reservoir development in the Fullerton reservoir and on the reservoir architecture. The Leonardian of the Sierra Diablo contains direct analogs of all major reservoir intervals at Fullerton field, including the shallow-water platform carbonates of the Lower Clear Fork Group (Fitchen et al., 1995; Ruppel et al., 2000) and Wichita units and the

Integrated studies of outcrops and subsurface data sets indicate that the Abo represents the basal depositional sequence (sequence L1) of the Leonardian (Fitchen et al., 1995). Studies of Abo outcrops in the Sierra Diablo (Fitchen et al., 1995; Kerans et al., 2000) have demonstrated three important aspects of this succession in the Permian basin: (1) it consists of dominantly open-marine, outerplatform facies; (2) it displays clinoformal architecture; and (3) it is overprinted by karst features (sinkholes, caves, cave fill, and collapse features). The dominance of clinoformal, outer-ramp, fusulinidcrinoid packstones and wackestones and less common ramp-crest, ooid-peloid, grain-rich packstones and grainstones in the Abo contrasts with the flat-lying, alternating tidal-flat and shallow subtidal wackestonepackstone successions of the Wichita and Lower Clear Fork. The top-lapping clinoforms of the Abo document rapid basinward progradation, a forced regression probably caused by a rapid fall in sea level. Karsting in the Abo, although initiated at the exposed top of the Abo during sea level fall, is manifested downsection in the Abo by caves and sinkholes and upsection in the

360 / Ruppel and Jones

overlying Lower Clear Fork as collapse features. Where karsting and associated thickness variations are developed, largely in platform-marginal settings, the contact between the Abo and Lower Clear Fork is a relatively sharp and undulating, unconformable surface. Updip, karsting is less apparent, and the contact is less pronounced.

Lower Clear Fork The outcropping Lower Clear Fork succession in the Sierra Diablo represents a single depositional sequence (L2). Key elements of this succession are (1) basal backstepping tidal-flat deposits that locally fill relief on the underlying, karsted Abo surface; (2) an updip succession of amalgamated tidal-flat facies; (3) a cyclic, downdip succession of alternating tidal-flat and midramp, subtidal facies; and (4) an overall backstepping (upward-deepening) trend (Fitchen et al., 1995; Kerans et al., 2000; Ruppel et al., 2000). Lower Clear Fork deposits are characterized by alternating peritidal, tidal-flat deposits and subtidal, skeletal wackestones and packstones that document cyclic rise and fall of sea level at the high-frequency sequence and cycle scale. These cycles, which average about 6 m (20 ft) in thickness, display consistent patterns of facies stacking (cycle-base skeletal wackestones and overlying peloidal grain-rich packstones) and appear to be widely continuous (Ruppel et al., 2000). These midramp, shallow subtidal platform rocks pass downdip into clinoformal fusulinid-crinoid wackestones and packstones of the outer ramp-slope in less than 3 km (2 mi) basinward. Updip (landward), the Lower Clear Fork is characterized by amalgamated, inner-platform, peritidal, tidal-flat deposits. These updip tidal flats are analogous to Wichita facies in the subsurface. The absence of shallow-water highstand deposits at the top of the Lower Clear Fork suggests a rapid, perhaps forced, regression, followed by exposure and possible erosion at the top of the L2 sequence.

these names are best considered rock-stratigraphic terms (essentially facies) at the formation level of nomenclature. To place these units in their proper perspective, however, it is best to consider both facies and time interrelationships. In this report, we treat these named stratigraphic units (i.e., formations) as facies or rock-stratigraphic units but place them in a sequence-stratigraphic (i.e., time-stratigraphic) framework that has been developed from previous studies (e.g., Fitchen et al.,1995; Kerans et al., 2000; Ruppel et al., 2000). On the basis of these studies and data from Fullerton field, we interpret the Abo and the lower part of the Wichita Formation to be time-equivalent facies (or systems tracts) of the earliest composite (third-order) depositional sequence of the Leonardian (sequence L1). The overlying sequence (L2) comprises updip peritidal deposits of the upper Wichita Formation and downdip, dominantly subtidal deposits of the Lower Clear Fork Formation (Figure 5). Sequence L3 contains basal transgressive systems tract (TST) siliciclastics of the Tubb Formation and carbonates of part of the overlying Upper Clear Fork (Figure 4). Correlations of usable wire-line-log suites (about 850) show that across most of the area of the Fullerton Clear Fork unit (Figure 3), stratigraphic units are relatively isopachous. This is consistent with the depositional setting of the field area on the Central Basin platform, a broad, flat carbonate platform not unlike the modern-day Bahamas platform. The generally isopachous nature of the Wichita and Lower Clear Fork is confirmed by 3-D and 2-D seismic data. These data, however, show that the Wichita and Abo vary considerably in thickness across the field. Wichita deposits thin, and Abo deposits thicken to the east and southeast (Figure 6). Synthesis of core, wire-line, and seismic data suggests that this reciprocal thickness relationship is the result of facies change; i.e., the Wichita represents the updip, shallower water (inner-platform) equivalent of the distal, deeper water (outer platform to slope) Abo. Support for this conclusion is presented in subsequent sections.

FULLERTON RESERVOIR FACIES AND STRATIGRAPHY

Facies and Depositional Setting

The productive reservoir section at Fullerton field includes parts of the Abo, Wichita, and Lower Clear Fork formations (Figures 4, 5). Like many stratigraphic names used in the subsurface, each of these units is commonly considered to display both regional time equivalency and facies constancy across the region. In other words, each is considered to be both a rockstratigraphic and a time-stratigraphic unit. In fact,

Reservoir facies encountered at Fullerton field are typical of those observed throughout most carbonate platform successions of Leonardian and early Guadalupian age in the Permian basin. The characteristics of these and similar facies have been documented by many authors, including Bebout et al. (1987), Ruppel and Cander (1988a, b), Garber and Harris (1990), Longacre (1990), Kerans et al. (1994; see also Kerans and Fitchen, 1995; Kerans and Kempter, 2002),

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 361

FIGURE 5. Northwest – southeast cross section (AA0 ) across the Fullerton field area showing sequence architecture and general facies development based on cored well control. Line of section shown in Figure 3. Depth in feet.

FIGURE 6. Three-dimensional seismic section (in time) across the southern part of Fullerton field showing the seismic definition of the Clear Fork reservoir section. Yellow lines are timeline boundaries; dotted line defines the top of the Abo Formation. Note that whereas the upper Wichita and Lower Clear Fork intervals are essentially isopachous and continuous across the field, the lower Wichita and Abo display reciprocal thickness relationships. Dashed lines are top-lapping Abo clinoforms. UCF = Upper Clear Fork; LCF = Lower Clear Fork.

362 / Ruppel and Jones

FIGURE 7. Core and thin-section photographs of typical Wichita tidal-flat facies. (A) Slab photo of peritidal mudstonewackestone showing weak laminations and local burrowing. Core is 10 cm (4 in.) wide. Exxon FCU 5927, depth: 2140 m (7021 ft). (B) Photomicrograph of peritidal mudstone-wackestone facies showing abundant intercrystalline porosity. Exxon FCU 6122, depth: 2162 m (7092 ft). Porosity: 13.8%; permeability: 1.6 md. (C) Slab photo of clay-rich carbonate mudstone facies in the Wichita. Core is 10 cm (4 in.) wide. Pan American FM-1, depth: 2234 m (7329 ft). Note underlying fenestral mudstone.

on differences in grain preservation, diagenesis, grain size, etc. Most can be found in any part of the reservoir succession.

Peritidal Mudstone-wackestone

and Ruppel and Bebout (2001). Leonardian facies successions have been documented both in outcrop (Ruppel et al., 2000) and in the subsurface (Atchley et al., 1999; Ruppel, 2002). Facies groupings defined in this study and presented below were defined with three goals in mind: (1) to document the characteristics of distinct depositional settings; (2) to document widespread and potentially correlatable sediment packages; and (3) to create a sound geological basis for assessing relationships between rock textures and fabrics and petrophysics. In many ways, it is the last of these goals that is most important because it provides the foundation for reservoir modeling. However, as a result of this, facies characteristics used for subdivision are dominantly small-scale matrix properties, e.g., grain size and shape, grain type. Larger scale features, such as fractures, anhydrite nodules, and evidence of burrowing, although noted, are not used for facies subdivision. Twelve facies can be at least locally identified in the Fullerton reservoir succession. Most of these are intergradational with one or more others. In many cases, their distinction is somewhat subjective, based

These rocks, which are most abundant in the Wichita but also locally common in the Lower Clear Fork, are massive to parallel-laminated mudrich rocks that only sparsely contain grains other than a few peloids (Figure 7A, B). They are most typically associated with the exposed tidal-flat facies and the clay-rich carbonate mudstone facies and are interpreted to represent peritidal deposition on a very low-energy tidal flat. They are dominantly dolomitized, and where dolomite, they may display very high porosities (as much as 15%). Because this porosity is caused by intercrystalline pores in fine-crystalline dolomite, however, the permeability is generally low. In limestone, the facies invariably exhibits very low porosity (typically less than 2%) and permeability. Gamma-ray response is typically medium to high and highly variable in these rocks because of the local presence of clay minerals.

Clay-rich Carbonate Mudstone These rocks are found intimately associated with rocks of the peritidal mudstone-wackestone facies in the Wichita. They are typically dark gray to nearly black and are found in beds 1– 8 cm (0.4 –3.1 in.) in thickness (Figure 7C). Where cyclic upward-shallowing successions can be identified, these deposits are found at or near cycle tops in both the Lower Clear Fork

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 363

FIGURE 8. Core, thin-section, and borehole image log photographs of typical tidal-flat facies in the Wichita and Lower Clear Fork. (A) Slab photo of exposed tidal-flat facies showing typical parallel laminations and fenestral pores. Core is 10 cm (4 in.) wide. Lower Clear Fork Exxon FCU 7322, depth: 2075 m (6808 ft). (B) Image log of laminated tidal-flat facies. Lower Clear Fork, HFS L2.1. Exxon FCU 2564, depth: 2103–2104 m (6900–6904 ft). Width of image is 63.5 cm (25 in.). (C) Photomicrograph of exposed tidal-flat facies showing fenestral vuggy porosity. Lower Clear Fork Exxon FCU 6122, depth: 2104 m (6903 ft). (D) Slab photo of cycle top showing fenestral exposed tidal-flat facies overlain by subtidal, burrowed peloid wackestone. Note small lithoclasts above cycle top. Lower Clear Fork, top HFS L2.2. Core is 10 cm (4 in.) wide. Exxon FCU 7322, depth: 2074 m (6804 ft).

and the Wichita. Although they are probably contributors to the high gamma-ray response observed in the Wichita and in Lower Clear Fork peritidal intervals, because of their thinness, they cannot be discretely defined by logs and, therefore, are not correlatable. Their color and clay content and lack of apparent lateral continuity suggest that they were formed as local organic-rich ponds or stagnant pools on the tidal flat. They may act as local baffles to reservoir fluid flow but are probably very discontinuous laterally.

Exposed Tidal Flat These deposits show obvious evidence of exposure such as fenestral pores, pisolites, mud cracks, insect burrows, sheet cracks, tepee structures, and cyanobacterial or microbial laminations (Figure 8). They are locally common in both the Lower Clear Fork and the Wichita. In the Lower Clear Fork, they define cycle tops (Figure 8D); in the Wichita, their eustatic significance is less apparent. Their sedimentary structures

demonstrate, however, that they were formed during at least local sea level fall and exposure. Porosity is locally very high (at least 30%) and is associated with a combination of fenestral pores and intercrystalline and interparticle pores. Permeability is lower than in grain-dominated subtidal rocks but can be significant where high porosities are present. However, this study and others (e.g., Ruppel, 2002) show that effective permeabilities (0.1 md) are associated only with porosities above approximately 12%. Like the peritidal mudstonewackestone facies, these rocks typically display somewhat elevated and variable gamma-ray response. This can sometimes be used to distinguish these rocks from overlying and underlying subtidal facies.

Peloid Wackestone This grain-poor facies differs from the rocks of the peritidal mudstone-wackestone facies in its association and sedimentary structures. The rocks are invariably associated with other, demonstrably subtidal, facies; they are burrowed; and they locally contain skeletal debris (Figure 9). These features suggest that they were deposited in low-energy subtidal settings. Dominant grains are 80 –150-mm-diameter peloids that are probably fecal pellets created by infaunal burrowers. The absence of skeletal allochems supports a low-energy, perhaps restricted, setting.

364 / Ruppel and Jones

FIGURE 9. Core, thin-section, and image log photo images of nodular peloid wackestone typical of the Lower Clear Fork HFS L2.2. (A) Core photo showing anhydrite nodules surrounded by lightcolored halos of higher porosity. Exxon FCU 6739, depth: 2130 m (6989 ft). Width of core is 10 cm (4 in.). (B) Image log depicting anhydrite burrows and rimming porosity as white (high-resistivity) patches surrounded by black (lowresistivity) rims. Exxon FCU 2564, depth: 2097 – 2098 m (6879 – 6883 ft). Width of image is 63.5 cm (25 in.). (C) Thinsection photomicrograph of low-porosity, burrowed, peloid wackestone. Exxon FCU 6429, depth: 2078 m (6817 ft). (D) Thin-section photomicrograph of peloid wackestone with moldic pores. Exxon FCU 6229, depth: 2085 m (6841 ft).

burg Formation (Ruppel and Bebout, 2001), suggesting that these features are an expected and perhaps predictable feature of late highstand sedimentation on Permian shallow-water carbonate platforms. Burrowed, nodular fabrics are readily defined on borehole imaging logs (Figure 9B). Porosity, except where enhanced around burrows, is generally moldic and low (Figure 9C, D).

Peloid Packstone

Anhydrite nodules are locally common in these rocks, probably reflecting the postdepositional entry of sulfate-bearing diagenetic fluids into permeability pathways created by burrowers. In many cases, nodules occupy solution-widened vertical burrow pathways that are surrounded by alteration halos (Figure 9A). These halos (which have been reported from many Permian platform carbonate successions) commonly display a more grain-rich texture, higher porosity and permeability, and a depleted oxygen isotope signature (Major et al., 1990; Ruppel and Bebout, 2001). At Fullerton, they are most commonly developed in highfrequency sequence (HFS) 2.2. This stratigraphic position (in the late highstand of the L2 composite sequence) is analogous in terms of accommodation to similar features documented from the younger Gray-

The distinction between these rocks and those assigned to the peloid wackestone facies (Figure 10) is commonly based on apparent peloid abundance and can be subjective. As with the peloid wackestone facies, peloids are mostly fecal pellets created by burrowing infauna, although skeletal debris (chiefly mollusk fragments) are locally common. It should be noted that the preservation of these pellets is largely a function of diagenesis, both early and late. Nearly all of the low-energy mud-dominated facies in the Permian contain obvious pellets, indicating that burrowers were ubiquitous in these deposits. A complete gradation in texture exists between pelleted mudstones and peloid packstones. Differences in texture are probably most commonly caused by differences in early diagenesis. Pellets that were early lithified and stabilized are more likely to be preserved. Thus, peloid (i.e., pellet) packstones are sediments that underwent significant amounts of early diagenesis, whereas peloid wackestones and

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 365

FIGURE 10. Core, thin-section, and image log photographs of mud-rich peloid packstone-wackestone. (A) Slab photo of burrowed Lower Clear Fork peloid packstone (dolostone). Exxon FCU 6229, depth: 2133 m (6997 ft). Core is 10 cm (4 in.) wide. (B) Image log of peloid wackestone-packstone facies typical of Lower Clear Fork HFS L2.2. Exxon FCU 2564, depth: 2105 – 2106 m (6905–6909 ft). Width of image is 63.5 cm (25 in.). (C) Photomicrograph of Lower Clear Fork peloid packstone (limestone) with moldic porosity. Exxon FCU 5927, depth: 2104 m (6903 ft). (D) Photomicrograph of Lower Clear Fork peloid packstone (limestone) with moldic porosity. Exxon FCU 6229, depth: 2078 m (6817 ft).

pelleted mudstones underwent relatively little. The textural significance of these pelleted facies is thus more diagenetic than depositional. Accordingly, apparent variations in peloid abundance in these muddominated facies packstones, wackestones, and mudstones are not necessarily an indication of differing depositional environment or fluctuations in wave energy. Instead, they may be predominantly caused by local changes in rates and effects of early diagenesis. Reflecting these subtleties, borehole image logs cannot differentiate between peloid wackestones and packstones, imaging them all as granular, relatively homogeneous sediment (Figure 10B). Porosity in these rocks is commonly associated with intercrystalline pores and uncommon skeletal moldic pores (Figure 10C, D).

Peloid Grain-dominated Packstone Unlike mud-rich peloid facies discussed above, grain-dominated (or grain-rich) peloid packstones commonly display evidence of possible wave-related

transport. These rocks, which contain pellets (90–120 mm), less common ooids (150–250 mm), and unidentifiable spherical grains, are typically well sorted and contain interparticle pores that are either open or filled with cements (Figure 11). The interparticle pores indicate that these peloids acted as true grains instead of pelleted mud, in contrast to the intercrystalline and moldic pores that typify mud-dominated facies. As discussed above, in some instances, the grain-dominated texture is associated with vertical burrows and probably owes its origin to burrow-related diagenesis. For the most part, this facies is restricted to the subtidal legs of Lower Clear Fork depositional sequences. These rocks are among the highest quality reservoir facies in the field. Highest porosity (as much as 20%) and permeability are encountered in peloid grain-dominated packstone limestones (Figure 11C), although dolostones also exhibit good porosity and permeability (Figure 11B, D). Where burrowed, porosity and permeability can be greatly enhanced in and around burrows (Figure 12A). These high-porosity burrow fills commonly account for apparent moldic fabric on image logs (Figure 12B).

Ooid-peloid Grain-dominated Packstone-grainstone These rocks differ from the peloid grain-dominated packstone facies, with which they are commonly closely associated, in having recognizable ooids (as much as 250–300 mm in diameter) in addition to pervasive

366 / Ruppel and Jones

FIGURE 11. Core and thin-section photomicrographs of peloid, grain-rich packstones. (A) Slab photo of Abo peloid grain-dominated packstone. University Consolidated IV-25, depth: 2210 m (7252 ft). Core is 10 cm (4 in.) wide. (B) Photomicrograph of Lower Clear Fork peloid dolostone graindominated packstone showing interparticle porosity. White areas are poikilotopic anhydrite. Exxon FCU 6122, depth: 2126 m (6976 ft). (C) Photomicrograph of Lower Clear Fork peloid (ooid?) dolostone grain-dominated packstone-grainstone with interparticle porosity. Exxon FCU 6946, depth: 2083 m (6835 ft). (D) Photomicrograph of Lower Clear Fork peloid grain-dominated packstone dolostone showing interparticle porosity. Exxon FCU 6229, depth: 2078 m (6816 ft). (E) Slab photo of Lower Clear Fork (L 2.2) peloid graindominated packstone (dolostone). Exxon FCU 6229, depth: 2079 m (6822 ft). Width of photo is 10 cm (4 in.).

Fusulinid Wackestone-packstone

pellets. Skeletal grains in the form of fusulinids, mollusks, and crinoids are also common. In most cases, these deposits, which are restricted to subtidal sections of the Lower Clear Fork, are also well sorted, probably because of wave action. Grainstones, although relatively uncommon, display excellent size sorting and, in some cases, possess inclined or cross-laminations (Figure 13B). These rocks, which occur as both dolostone and limestone, represent the highest energy facies in the reservoir succession and, in many cases, display the best porosity and permeability. Pores are dominantly interparticle, although moldic pores are also very abundant, especially in limestone-dominated intervals (Figure 13D).

Fusulinid-bearing rocks (Figure 14) are found in all three formations (Abo, Wichita, and Lower Clear Fork), although they are very limited in the Wichita. In all occurrences, they are most commonly dolomitized. Fusulinids, which constitute as much as 40% of the rock are preserved either as open (Figure 14C) or anhydritefilled (Figure 14A) molds or as wellpreserved fossil tests (Figure 14D). In the Lower Clear Fork and Wichita, they are associated with abundant peloids (probably pellets). In the Abo, they commonly co-occur with crinoids and less commonly with brachiopods. Fusulinids are thought to have occupied water depths of 30 m (100 ft) or more and here represent the deepest water facies observed at Fullerton field. Accordingly, their presence is an indicator of platform flooding and relative sea level rise, making them key indicator facies of cycle and sequence boundaries.

Skeletal Wackestone-packstone Typically, these rocks contain small volumes of skeletal debris (most commonly mollusk fragments, but also including crinoids and less common ostracods) and the ubiquitous peloids. They are gradational into

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 367

FIGURE 12. Core photo (A) and image log (B) images of burrowed peloid packstone. Although low-resistivity (black) features are commonly interpreted to be open vugs on image logs, core photo reveals that vugs are actually burrow fills (light colored) of higher porosity. Lower Clear Fork. Core: Exxon FCU 6739, depth: 2123 m (6966 ft). Image log: Exxon FCU 2564, depth: 2095 – 2096 m (6873 – 6877 ft). Core is 10 cm (4 in.) wide. Image log is 63.5 cm (25 in.) wide.

a marker facies indicative of transgression or platform deepening (i.e., sea level rise).

Siltstone-sandstone

peloid wackestones and packstones. Evidence of burrowing is common. The dominance of mollusks and the essential absence of more normal-marine organisms in these rocks suggest that they were deposited in an inner-platform setting. Typically, they exhibit low porosity; pore space is created by skeletal molds and intercrystalline pores.

Oncoid Wackestone-packstone Oncoids (or oncolites) are large microbially coated grains (Figure 15) formed under conditions of frequent wave agitation in shallow water. They are abundant at the base of the Lower Clear Fork throughout the entire field area. Invariably, they are associated with fusulinids and other faunas indicative of openmarine deposition. This association and their stratigraphic position immediately above the top of the tidalflat–dominated Wichita indicate that they represent marine flooding of the platform during sea level rise (transgression). In some downdip wells (e.g., Pan American FM-1; Figure 3), they are common at Lower Clear Fork cycle bases. Their distribution suggests that they document relatively high-energy conditions developed during platform flooding. Like the fusulinid facies, the oncoid wackestone-packstone facies is

Quartz silt- and sand-bearing rocks are common only at the top of the Lower Clear Fork and in the overlying Tubb Formation. Because the quartz is associated with potassiumand thorium-rich clay minerals, its presence in the Lower Clear Fork and Tubb is well defined by high corrected gamma-ray (CGR) wire-linelog response (Figures 4, 5). Their maturity, size (fine sand to coarse silt), sorting, and shape (generally subangular) suggest that these sediments were originally wind blown (Fischer and Sarntheirn, 1988). This indicates that most of the quartz in the section was delivered to the area during sea level lowstands when the platform was emergent. Quartz silt-sand occurs in two scenarios: (1) in peritidal tidal-flat facies and (2) in reworked subtidal facies. The former, which are probably formed as small volumes of silt, sand, and clay, are blown onto intermittently exposed, slowly accumulating tidal flats and admixed with peritidal carbonate sediment. The presence of potassium and thorium in these clastics produces the increased gamma-ray response that is associated with many tidal-flat deposits in the Clear Fork and facilitates recognition of these cycle-capping deposits in these sediments (Ruppel, 2002). Most of the occurrences of silt, sand, and clay in the Lower Clear Fork at Fullerton are of this type. Subtidal silt, sand, and clay deposits are most common in the Tubb. These rocks were formed first by large-volume eolian deposition during extended sea level lowstand and carbonate nondeposition, then reworked during the ensuing sea level rise and marine

368 / Ruppel and Jones

FIGURE 13. Core, thin-section, and image log photographs of Lower Clear Fork and Abo grainstones. (A) Slab photo of Lower Clear Fork ooid grainstone showing cross-laminations. Exxon FCU 6122, depth: 2109 m (6920 ft). Core is 10 cm (4 in.) wide. (B) Image log of cross-bedded grainstone. Lower Clear Fork. Exxon FCU 2564, depth: 2134–2135 m (7002–7005 ft). Width of image is 63.5 cm (25 in.). (C) Photomicrograph of Lower Clear Fork ooid limestone-grainstone showing oomoldic pores. Exxon FCU 5927, depth: 2107 m (6913 ft). (D) Photomicrograph of Lower Clear Fork ooid dolostone grainstone wackestone showing interparticle and minor moldic pores. Exxon FCU 4828, depth: 2175 m (7137 ft).

Depositional Model

flooding of the platform. These deposits are generally richer in clastic content, are intermixed with carbonate mud, and show evidence of subtidal conditions (e.g., burrows, stratification). Although these rocks may locally exhibit some minor porosity, they do not appear to display any significant reservoir permeability.

Lithoclast Wackestone Thin intervals (commonly less than 0.3 m [1 ft]) containing scattered lithoclasts are locally encountered at the contacts between transgressive marine facies and underlying tidal-flat deposits or other facies showing evidence of subaerial exposure (Figure 8D). Clasts are variable in composition but most commonly consist of fragments or intraclasts derived from the underlying bed. Clasts typically are 1 cm (0.4 in.) in width or less. Although volumetrically minor, these rocks are an important indicator facies of lithification because of exposure or nondeposition and subsequent sea level rise and, thus, of cycle boundaries.

The relative distribution of the facies described above can best be understood when considered in light of a conceptual geological model. Figure 16 portrays idealized 3-D relationships among major facies types and depositional environments typical of most middle Permian (Leonardian and Guadalupian) platform carbonate successions in the Permian basin. It should be understood that this model reflects only the relative interrelationships among facies tracts. The actual position of individual facies tracts at a given point in time is a function of many factors, including accommodation, rates and magnitude of sea level rise, rising versus falling sea level trends, climate, tectonics, etc. However, during normal relative rises in sea level, facies tracts normally step landward, whereas during falls, they step basinward. Exceptions to this general pattern are common. For example, ramp-crest facies may be much better developed during sea level highstand and fall than during transgression and rise. The basal Leonardian (L1) depositional sequence at Fullerton, for example, appears to be dominated by an updip, inner-ramp succession (Wichita) and a downdip outer-platform succession (Abo) with a very poorly developed middle-ramp to ramp-crest facies tract. Nevertheless, the vertical successions (facies-stacking patterns) formed by sea level-driven migrations of these facies tracts are the key to identifying depositional cycles in cycles from cores and logs.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 369

FIGURE 14. Core, thin-section, and image log photographs of the fusulinid wackestone-packstone facies. (A) Slab photo of fusulinid wackestone with open and anhydrite-filled fusumoldic pores. Exxon FCU 7322, depth: 2132 m (6996 ft). Core is 10 cm (4 in.) wide. (B) Image log of fusulinid wackestone-packstone facies. Exxon FCU 2564, depth: 2131 – 2132 m (6990–6994 ft). Width of image is 63.5 cm (25 in.). (C) Photomicrograph of fusulinid wackestone showing open fusumolds. Exxon FCU 7630, depth: 2066 m (6778 ft). (D) Photomicrograph of fusulinid wackestone showing wellpreserved fusulinids but little porosity. Exxon FCU 4828, depth: 2139 m (7018 ft).

Among the identified facies at Fullerton, the peritidal mudstone-wackestone facies, the clay-rich carbonate mudstone facies, and the exposed tidal-flat facies, for the most part, represent deposition on the inner ramp in a sabkha-tidal-flat setting as low-exposure index (cf. Hardie and Garrett, 1977) peritidal deposits, tidal-flat ponds, and high-exposure tidal-flat deposits, respectively (Figure 16). It should be noted, however, that exposure fabrics, like those developed in the inner platform, can also form on the ramp crest following complete aggradation and exposure at any time. The peloid- and pellet-rich facies at Fullerton (peloid wackestone facies, peloid packstone facies, peloid graindominated packstone facies, and skeletal wackestonepackstone facies) are also dominantly associated with the inner ramp but occupy a somewhat more distal, lagoonal to restricted subtidal setting, where sediment formation is dominated by infaunal and epifaunal burrowing activities (Figure 16).

The ooid-peloid grain-dominated packstone-grainstone facies in the Lower Clear Fork at Fullerton is typical of facies deposited in a platform ramp crest at the platform margin where relatively high wave energies are developed (Figure 16). However, none of the cored wells in the field reveal the kind of vertical facies stacking and amalgamation that typically defines a well-developed ramp crest (see, for example, Kerans et al., 1994; Ruppel et al., 2000). This may reflect insufficient core control or the absence of a well-developed ramp crest on the Lower Clear Fork platform. As previously stated, the Abo also appears to lack a rampcrest facies succession. As in most middle Permian carbonate-platform successions, the fusulinid wackestone-packstone facies documents outer-platform deposition. Such rocks are well developed in both the Lower Clear Fork and the Abo. In the Abo, these rocks display classic clinoformal bedding and the general lack of cyclicity that typifies a distal outer-platform setting (Figure 16). By contrast, fusulinid wackestone-packstone facies rocks in the Lower Clear Fork are horizontally bedded and interbedded with middle- and inner-platform facies. This suggests that these fusulinid-rich rocks were deposited in a much more proximal outer-ramp position. The oncoid wackestone-packstone facies is closely associated with the fusulinid wackestone-packstone facies in the Lower Clear Fork at Fullerton, indicating that the former was deposited during platform deepening. This facies has not previously been reported

370 / Ruppel and Jones

FIGURE 15. Core photographs of oncoid, wackestone-packstone facies. (A) Small, algally coated fusulinids. Lower Clear Fork, Exxon FCU 5927, depth: 2128 m (6980 ft). (B) Large oncoids. Lower Clear Fork, Pan American FM-1, depth: 2182 m (7160 ft). Cores are 10 cm (4 in.) wide.

Sequence Stratigraphy Figure 17 illustrates diagrammatically the sequence architecture, basic facies tracts interrelationships, and rock and time terminology of the lower Leonardian at Fullerton field. In this section, we describe the sequence-stratigraphic characteristics of each of the major Leonardian units at Fullerton.

Abo Formation

from middle Permian rocks in the Permian basin. Its occurrence at Fullerton field may indicate the development of relatively higher energy conditions on the outer platform in the Fullerton area than is generally developed in other regions in the Permian basin.

As already discussed, the Abo represents the distal facies or systems tract of the earliest (oldest) Leonardian sequence in the Permian basin: the L1 sequence (Figures 4, 5, 17). Regionally, the Abo consists of outer ramp to slope, skeletal crinoid-fusulinid-dominated subtidal facies and, less commonly, ramp-crest ooidpeloidal grainstones (Kerans et al., 2000). In both

FIGURE 16. Depositional model for Permian shallow-water carbonate platforms in the Permian basin. This model is applicable to most Leonardian and Guadalupian carbonate platform successions, including the reservoir succession at Fullerton field. Modified from Kerans and Ruppel (1994).

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 371

FIGURE 17. Generalized sequence-stratigraphic model of the lower Leonardian succession at Fullerton field showing sequence boundaries, primary facies tracts, and stratigraphic nomenclature.

outcrop and the subsurface, the Abo is dominantly characterized by clinoformal, top-lapping geometries (Kerans et al., 2000; Zeng and Kerans, 2003). This typical Abo architecture is very apparent in 3-D and 2-D seismic data from Fullerton field, where the Abo is readily defined by its clinoformal reflectors, and its top is defined by a prominent toplap surface that is apparent throughout much of the field (Figure 6). Although the Abo at Fullerton field is penetrated by relatively few wells and has been cored in even fewer wells, existing cores display facies and bedding characteristics typical of sediments deposited in an outer ramp-slope setting (such as open-marine fauna, inclined bedding, graded beds, slump features, conglomerates, and lack of shallow-water sediments). Facies consists of alternating beds of fusulinid-crinoid packstones and wackestones and peloidal packstones. No cyclicity is apparent in the Abo, although alternations between skeletal-rich and skeletal-poor intervals are evident. Inclined beds are locally common. Additionally, correlations in the Abo are not

apparent either from core sections or from wire-line logs. In part, this is the result of limited core and welllog control through the Abo section. But poor correlatability is typical of outer-platform depositional successions that display clinoformal architectures because of the associated dipping bedding surfaces, discontinuous facies packages, and poor vertical facies contrasts. Abo rocks are dominantly dolostones at Fullerton field. Porosity is locally very high, reaching values of as much as 25%. Pore types vary among fusumolds, intercrystalline pores in coarse-crystalline dolostones, and interparticle pores in skeletal-peloidal packstones. The thickness of the Abo is indeterminate largely for two reasons. First, few wells and no cores penetrate the complete Abo to Wolfcamp section. Second, outcrop studies reveal that the Abo – Wolfcamp contact may be lithologically indistinct in many areas: in outcrop, both the uppermost Wolfcamp and the Abo are composed of clinoformal, outer-platform fusulinid-crinoid wackestones (Kerans et al., 2000).

372 / Ruppel and Jones

A minimum thickness of 90 m (300 ft) is established by the Pan American FM-1 cored well in the southeastern corner of the field (Figures 3, 5). The contact between the Abo and the overlying Wichita varies from sharp to gradational. In some cores, the contact is marked by an interval (as much as several feet thick) of intermixed clasts and fragments of Wichita tidal-flat deposits and Abo fusulinidbearing subtidal deposits. The brecciated nature of this contact interval suggests that it is the result of karst-related dissolution and collapse. These breccias and their distribution are discussed more fully in a later section of this chapter.

karsted contact between the outer-ramp subtidal deposits of the Abo and peritidal tidal-flat rocks of the overlying Wichita.

Lower Clear Fork Formation Rocks assigned to the Lower Clear Fork Formation represent the subtidal systems tract of the Leonardian 2 (L2) sequence. Core studies show that the Lower Clear Fork can be subdivided into three HFSs throughout most of the Fullerton field area: L2.1, L2.2, L2.3 (Figures 4, 5). Each of these HFSs can be identified as having a transgressive, dominantly subtidal lower leg and an overlying, highstand upper leg.

Wichita Formation

High-frequency Sequence L2.0

The Wichita at Fullerton field consists of a diverse assemblage of peritidal to supratidal tidal-flat deposits. Integration of outcrop and subsurface data suggests that the Wichita Formation is the updip, proximal facies equivalent of both the Abo and the Lower Clear Fork subtidal successions. These data also imply that the Wichita actually comprises parts of two depositional sequences: the highstand leg of sequence L1 and the transgressive leg of sequence L2 (Figure 17). The lower Wichita represents the updip, tidal-flat facies tract equivalent of the downdip, outer-platform facies tract of the Abo in Leonardian sequence L1, whereas the upper Wichita represents the updip tidalflat facies equivalent of the basal Lower Clear Fork subtidal facies in sequence L2 (HFS 2.0; Figures 5, 17). The Wichita is dominantly composed of dolostone; however, intervals of limestone are common in the upper Wichita in the northern part of the field. The limestone intervals are relatively persistent laterally and are generally subparallel to stratigraphic markers in the field (Figure 18a). The thickness of the Wichita is relatively consistent throughout the northern two thirds of the field, ranging from about 75 to 105 m (280 to 350 ft), but thickness decreases markedly to as little as 34 m (110 ft) in the southeastern part of the field (Figure 19). This decrease in thickness, which occurs relatively abruptly along a generally northeast-trending belt, marks the platform-margin boundary between updip tidal-flat Wichita facies and downdip outer-ramp Abo fusulinid facies (Figure 5). The position of the L1-L2 boundary is difficult to precisely place within the thick succession of Wichita peritidal deposits. The contact shown in Figure 5 is extrapolated from downdip core and well control, where the L1-L2 contact is well defined by a sharp,

High-frequency sequence L2.0 documents the initial flooding of the platform following sea level fall and exposure at the end of L1 deposition. In most of the field area, HFS 2.0 consists of amalgamated tidal-flat deposits of the upper Wichita. Subtidal Lower Clear Fork deposits of L2.0 are only present at the margins of the field (Figures 4, 5).

High-frequency Sequence L2.1 In most of the Fullerton field area, HFS L2.1 forms the base of the Lower Clear Fork. As defined for this study, L2.1 consists of a basal section of transgressive to early highstand subtidal platform facies and an upper section of highstand tidal-flat facies (Figures 4, 5, 17). The basal transgressive subtidal facies of the Lower Clear Fork represent the first marine flooding of the platform and a sharp change in depositional style from the tidal-flat deposition of the Wichita to subtidal deposition of the Lower Clear Fork. The basal transgressive leg of L2.1 consists dominantly of fusulinid wackestone-packstone and oncoid wackestone-packstone facies that document landward backstepping of outer platform facies across the platform. These rocks are generally overlain by a succession of peloid packstones and grain-dominated packstones that represent late transgression and early highstand. Locally, within this succession, there are exposed tidal-flat facies indicating periodic exposure. L2.1 is capped throughout most of the field by a succession of one to three tidal-flat-capped cycles that represent exposure during late highstand (Figures 5, 17). Tidal-flat facies in some cases exhibit an elevated gamma-ray-log response that aids in their recognition (Figures 4, 5). Like most of the Lower Clear Fork in the Permian basin, HFS 2.1 is dominantly dolostone. However,

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 373

FIGURE 18. Distribution of limestone and dolostone in Wichita and Lower Clear Fork based on cores and wire-line logs. (a) Cross section (BB0 ) across the northern part of the field showing cycle stratigraphy and mineralogy. (b) Maps showing limestone in Wichita HFS L2.0, Lower Clear Fork HFS L2.1, and Lower Clear Fork HFS L2.2. limestone is locally common at Fullerton field, especially in the transgressive leg of 2.1. Areally, limestone is most abundant in an arcuate belt through the field (Figure 18b). The thickness of L2.1 is relatively constant across the Fullerton field area, ranging from about 43 to 46 m (140 to 150 ft) across most of the area. Good evidence shows that local thickness changes are a function of

differential subsidence along deep-seated faults (Jones and Ruppel, 2004). This is especially apparent along the north-trending fault in the northeastern part of the field, where the thickness changes from less than 43 m (140 ft) on the western, upthrown side of the fault to more than 49 m (160 ft) on the eastern, downthrown side of the fault. Examination of modern structure (Figure 3) shows that considerable movement

374 / Ruppel and Jones

High-frequency Sequence L2.2

FIGURE 19. Thickness of the Wichita Formation in the Fullerton field area. Marked thinning in southeast part of the field corresponds to facies change from Wichita to Abo.

on this fault occurred even after Lower Clear Fork deposition. Porosity is most abundant in the subtidal (transgressive and early highstand) parts of the sequence. Pore space is dominated by intercrystalline and moldic pores, the latter being especially abundant in limestone sections. Overall, these rocks are most porous in areas where limestone is present. Tidal-flat rocks are also locally porous but generally contain low permeabilities. The subtidal rocks of L2.1 probably constitute the most productive reservoir interval in the field.

High-frequency sequence L2.2 is similar to HFS L2.1 in consisting of a basal transgressive leg composed of backstepping tidal-flat facies, a middle (late transgressive to early highstand) leg composed dominantly of subtidal facies, and an uppermost (late highstand) leg composed of tidal-flat facies (Figures 5, 17). As is the case with L2.1, there is good evidence that the marine flooding of the platform following the postL2.1 lowstand was progressive. Basal L2.2 tidal-flat deposits are thickest in the center of the field area but generally absent along the margins, reflecting greater accommodation and early flooding of downdip areas. The lower abundance of outer-ramp fusulinid-rich facies in the TST of L2.2 relative to L2.1 suggests that the overall accommodation during the L2.2 sea level rise was somewhat less than during L2.1. Lower accommodation and attendant lower wave energies are also suggested by the near absence of the oncoid wackestone-packstone facies in L2.2. These rocks are found only in the most downdip core in the field. The fusulinid wackestone-packstone facies is also largely restricted to the eastern and southeastern parts of the area (Figures 5, 17). Throughout most of the field area, L2.2 is dominated by peloid wackestones and packstones typical of middle-platform deposition. The sequence is capped by a thin succession of tidalflat cycles (Figures 5, 17). High-frequency sequence L2.2 is composed dominantly of dolostone, with two important exceptions. Like L2.1, limestone is dominant in the southern part of the field (Figure 18b). Dolomite in this area is largely restricted to mud-rich fusulinid wackestones; virtually all grain-rich facies in this area are limestone. Limestone is also abundant in a small area in the north-central part of the field. The thickness of L2.2 is generally significantly less than that of L2.1. L2.2 ranges from about 26 to 27 m (85 to 90 ft) across most of the central part of the area to about 100 ft (30 m) along the northern and southern field margins. Porosity development in L2.2 is similar to that in L2.1. Porosity is greatest in the subtidal (transgressive and early highstand) parts of the sequence and dominated by intercrystalline and moldic pores. In addition, like L2.1, porosity is greatest in sections containing significant limestone.

High-frequency Sequence L2.3 The uppermost Lower Clear Fork HFS (L2.3) is composed of tidal-flat-capped restricted subtidal cycles throughout most of the field area (Figures 5, 17). These tidal-flat caps typically contain fenestral and

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 375

fine intercrystalline pore space that is definable on porosity logs. Log correlations suggest that these cyclecapping tidal-flat facies are relatively continuous across significant areas of the field. The underlying cyclebase subtidal rocks are dominantly mud-rich packstones and wackestones. Porosity, as indicated, is mostly restricted to tidal-flat caps. Because these rocks are dominated by fine intercrystalline and moldic pores, they contain little, if any, reservoir permeability. Accordingly, locally they contain oil stain and rarely contribute to oil production.

ships among facies to define cycles, (3) the nonuniform response of wire-line logs to these facies, and (4) the overprinting effects of diagenesis. Internal, cycle-scale correlations are possible, however, in the upper Wichita in the northern part of the field, where limestone beds are relatively correlative. Sedimentary structures suggest that these limestones represent undolomitized bases of cycles whose tops have been dolomitized by early diagenesis. The Wichita cycles defined by these limestone-dolostone couplets average 6 m (20 ft) in thickness.

Tubb Formation

Lower Clear Fork Cyclicity: HFS L2.1

The Tubb is characterized by fine-grained siliciclastics (coarse siltstone and fine sandstone), which are relatively easily defined by high-gamma-ray-log response (Figures 4, 5). These clastics are interpreted to represent eolian deposits that were deposited during the post-L2 lowstand and then reworked during the ensuing L3 sea level rise (Kerans et al., 2000; Ruppel et al., 2000; Ruppel, 2002). Most intervals are varying mixtures of siltstone-sandstone and carbonate (typically mud-rich, shallow-water facies). The Tubb locally contributes to oil production in the Permian Basin but is not part of the reservoir at Fullerton field.

Facies-stacking patterns in HFS 2.1 define numerous apparent cycle tops at thicknesses ranging from 0.6 to 4.6 m (2 to 15 ft). Cycles are typically characterized by mud-rich facies at their bases and grainrich facies at their tops. Basal TST cycles in HFS L2.1 contain abundant oncoids at cycle bases, along with accompanying fusulinids (Figure 20). These cycles are capped by better sorted, peloid-rich facies. Highfrequency cycles (typically 1.5–3 m [5–10 ft] in thickness) stack into cycle sets that average 7.5 – 12 m (25 – 40 ft) in thickness (Figure 20). Cycle sets display similar facies-stacking patterns to cycles consisting of fusulinid-rich bases and peloid-rich caps. Locally, tidalflat facies cap these cycle sets. Porosity is generally

Cycle-scale Stratigraphy The fundamental goal of cycle stratigraphy is to develop a correlation framework based on timeequivalent surfaces. The basic underlying premise for this approach is the assumption that widely correlative depositional cycles are formed by punctuated, allocyclic processes (e.g., sea level rise and fall) that affected sedimentation over broad areas. The methodology used to develop a cycle-scale stratigraphic framework in the Leonardian section at Fullerton field consists of the following: (1) characterization of facies-stacking patterns and cycle development in analogous outcrops; (2) description, interpretation, and logging of facies, stacking patterns, and possible cycle tops in cores; (3) integrated log- and core-based correlation of tentative cycle tops; and (4) definition of cycle architecture. At Fullerton field, nearly 4570 m (15,000 ft) of core (from 29 cored wells) was described in detail to provide the basic data for cycle definition.

Wichita Cyclicity Correlations in the thick Wichita succession of relatively similar tidal-flat deposits are difficult because of (1) the discontinuous nature of tidal-flat facies, (2) the lack of systematic vertical stacking relation-

FIGURE 20. Facies stacking and cycle development in the fusulinid- and oncoid-rich, transgressive systems tract of Lower Clear Fork HFS L2.1. Note that porosity is typically developed at or near cycle tops. Note also the lack of any systematic relationship between gammaray log and facies or cyclicity.

376 / Ruppel and Jones

facies. Cycles typically average 1.5 – 3 m (5 – 10 ft) in thickness; cycle sets are commonly 6 – 9 m (20 – 30 ft) thick (Figure 21).

Lower Clear Fork Cyclicity: HFS L 2.2

FIGURE 21. Facies stacking and cycle development in the grain-rich, highstand systems tract of Lower Clear Fork HFS L2.1. Note that porosity is typically developed at or near cycle tops. As with other Lower Clear Fork successions, gamma-ray logs do not show a systematic response to facies or cyclicity. highest at both cycle tops and cycle-set tops relative to bases. However, because cycles in the lower parts of cycle sets are generally more mud and fusulinid rich, overall porosity in these basal TST cycles is commonly relatively low. Cores and outcrop studies suggest that Clear Fork platform cycles and cycle sets are correlative across significant distances. However, where cores are not available, these correlations can be difficult to establish. Gamma-ray logs display virtually no systematic response to subtidal facies and cycles (Figures 20, 21) and, thus, cannot be reliably used for cycle-scale correlation throughout most of the Lower Clear Fork. In the absence of gamma-ray logs, the most effective way of establishing cycle-scale correlations is through the use of porosity logs. This approach is based on the observation from outcrops and cores that cycle tops consistently contain the most grain-rich facies, and that these rocks are most likely to contain high porosity. Accordingly, porosity logs can be used to correlate both the high-porosity facies in the upper part of the cycle and the overlying cycle top. L2.1 highstand cycles are dominated by mud-rich peloidal facies at their bases and grain-rich peloidor ooid-bearing facies at their tops (Figure 21). Fusulinids are commonly sparse in these highstand cycles, reflecting the basinward shift in facies tracts. In general, these cycles are dominated by peloid packstones and grain-rich packstones. Porosity is commonly highest at both cycle tops and cycle-set tops because of the abundance of these grain-rich

Facies stacking and cycle development in HFS L2.2 are very similar to those in L2.1. Cycles average 1.5– 3 m (5 – 10 ft) in thickness, and porosity is best developed in cycle tops (Figure 22). L2.1 cycles differ, however, in the lack of oncoid facies and the relative scarcity of fusulinid facies. This presumably reflects decreasing overall accommodation in L2.2 because of platform aggradation and slowing rates of long-term sea level rise. Additionally, cycle sets are not generally definable in L2.2. This is largely caused by the general absence of fusulinid facies that, in L2.1, define these intermediate-scale sea level-rise events.

Lower Clear Fork Cyclicity: HFS L2.3 High-frequency cyclicity in the Leonardian at Fullerton field is most readily definable in L2.3. These rocks, which are characterized by tidal-flat-capped shallow subtidal cycles (Figure 23), appear to be much more widely correlative than cycles in L2.1 or L2.2. Although neither facies nor cyclicity is defined by gamma-ray logs, both are distinguishable on porosity logs because of the typically well-developed porosity associated with cycle-capping tidal-flat deposits. The porosity in these rocks is generally caused by the presence of fenestral pores and can be relatively high, especially relative to the generally low porosity exhibited by cycle-base mud-rich wackestones and

FIGURE 22. Facies stacking and cycle development in the grain-rich, late transgressive systems tract and early highstand systems tract of Lower Clear Fork HFS L2.2. Note that porosity is typically developed at or near cycle tops. Note here again that gamma-ray logs do not display any relationship to facies or cycle development.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 377

FIGURE 23. Facies stacking and cycle development in the nonreservoir Lower Clear Fork HFS L2.3. Porosity is typically developed in cycle-capping tidal-flat facies, but little or no permeability is associated with the fenestral pores that dominate these caps.

packstones. However, because of the separate-vug fenestral pores, permeability is generally low, and L2.3 sparsely contributes to hydrocarbon production in the field. Porosity-based wire-line correlations suggest that cycle and facies continuity is high. This is somewhat unexpected, considering outcrop observations that suggest that tidal-flat facies are highly discontinuous. The laterally continuous porosity development at these cycle tops probably reflects the effects of early diagenesis associated with sea level fall and subsequent rise at these surfaces.

MINERALOGY AND DIAGENESIS The reservoir section at Fullerton field is similar to most other platform Leonardian successions in the Permian basin in showing evidence of significant postdepositional diagenesis. Principal products of this diagenesis are matrix-replacive and pore-filling dolomite and anhydrite. However, limestone is locally present in the Leonardian section at Fullerton, including the Abo, the Wichita, and the Lower Clear Fork.

Dolomite and Limestone Distribution Dolomite is, by far, the dominant mineral in the reservoir section. The Abo consists entirely of dolomite except in the most downdip wells. In the FM-1 well, perhaps the most downdip well in the field, the Abo contains alternating zones of limestone and dolostone. Dolostone is more commonly associated with mud-rich facies (peloid wackestone and fusulinid wackestone) in the Abo clinoformal succession, whereas limestone intervals are more commonly grain-rich, skeletal facies. Limestone is locally common in the Wichita, especially in the upper half of the formation in the north-

ern part of the field (Figure 18). In all cases, calciterich rocks in the Wichita are peritidal mudstones or wackestones (the peritidal mudstone-wackestone facies). These calcite-dominated facies characteristically exhibit very low porosity (<2%) and essentially no permeability. Log correlations suggest that limestone intervals are locally correlative (Figure 18a). Facies-stacking patterns suggest a systematic relationship between mineralogy and cyclicity, and it seems probable that these limestones are the result of cyclepunctuated diagenesis: limestones representing undolomitized cycle bases. Limestone is virtually absent from the lower Wichita (L1 highstand deposits) but very common in the upper Wichita (L2 TST) (Figure 18a). Limestone abundance in the upper (L2) Wichita displays a systematic trend across the field. The highest abundance of limestone is in the northwestern part of the field; essentially, no limestone is present in the southern part of the field (Figure 18b). Because of the low porosity associated with limestones in the Wichita, porosity logs can be reliably used to define and correlate limestone even where cores, dual porosity, or photoelectric effect (PE) logs are unavailable. Lower Clear Fork rocks at Fullerton field also contain locally abundant limestone (Figure 18a). In HFS L2.1, limestone is especially common in a sinuous belt that generally follows the margin of the outer platform (Figure 18b). Essentially, all grain-rich facies in this belt are dominantly calcitic. Limestone also dominates L2.2 in the southern part of the field. However, outside of this area, limestone is uncommon, except in a small area in the northwestern part of the field (Figure 18b). Porosity is generally high in limestone-rich intervals in the subtidal Lower Clear Fork. However, porosity is also well developed in dolostones, so except where cores, dual-porosity log suites, or PE logs are available, it is not possible to differentiate limestone from dolostone in the Lower Clear Fork.

Stable Isotope Chemistry Stable isotopes can provide insights into the timing and spatial distribution of diagenetic events. Because diagenesis is a key factor in porosity preservation and loss, such data can commonly provide insights into causes and distribution of porosity trends. Stable isotope data collected during this study and those previously gathered by J. Kaufman (1991, personal communication) reveal similarities to other Leonardian successions in the Permian basin but also appear to define spatially and temporally distinct patterns of diagenesis (Figure 24).

378 / Ruppel and Jones

FIGURE 24. Stable isotope data for the Tubb, Wichita, Lower Clear Fork, and Abo at Fullerton field. Solid red circles and blue triangles are dolomite and calcite samples, respectively, from J. Kaufman (1991, personal communication). Open green circles and green triangles are dolomite and calcite samples, respectively, collected during this study. Dolostone samples from the Tubb, Lower Clear Fork, Wichita, and Abo at Fullerton field exhibit d13C values ranging from 1.41 to 5.54% Peedee belemnite (PDB) (n = 50). These values are generally lighter than

data reported from the Guadalupian of the Permian basin (Ruppel and Cander, 1988a, b; Leary and Vogt, 1990; Saller and Henderson, 1998) but similar to most Leonardian values (Saller and Henderson, 1998; Ruppel, 2002). The d18O data obtained from Lower Clear Fork dolostones range from 2.96 to +2.97% PDB (Figure 24). With the exception of the lightest data, most of these values are similar to Leonardian data reported by Ye and Mazzullo (1993), Saller and Henderson (1998), and Ruppel (2002), but very different from data reported for Guadalupian rocks. Typical d18O values for Guadalupian (San Andres and Grayburg formations) platform dolomites are 3 – 6% (Vogt, 1986; Ruppel and Cander, 1988a, b; Saller and Henderson, 1998; Ruppel and Bebout, 2001; Ruppel, 2002). The relatively enriched isotopic signatures of Guadalupian dolomites have invariably been interpreted to have been produced during dolomitization by evaporatively concentrated seawater brines (Bein and Land, 1982; Bebout et al., 1987; Ruppel and Cander, 1988a, b). Taken collectively, the relatively depleted values for the Leonardian dolostones at Fullerton and in other Leonardian fields suggest that either (1) dolomitization was caused by brines that were less evaporatively concentrated than those that caused Guadalupian dolomitization; (2) dolomitization was caused by fluids of a mixed water origin; or (3) a combination of cases 1 and 2 above. Systematic spatial variations in d18O and d13C values of dolomites in the Wichita and Lower Clear Fork suggest an even more complex history of dolomitization and diagenesis. Stable isotope values from upper Wichita peritidal dolomites, for example, define two spatially distinct trends (Figure 25a). d18O and d13C values from high-porosity dolomites along the other margins of the Wichita tidal flat display relatively light values (average: 0.0% d18O, +2.6% FIGURE 25. Maps showing trends of facies, porosity, and stable isotope data in the upper Wichita L1-L2 tidal-flat succession (a) and the Lower Clear Fork L2 subtidal succession (b).

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 379

d13C). By contrast, values from low-porosity Wichita peritidal dolomites elsewhere in the field are distinctly heavier, especially with respect to d18O (average: +2.8% d18O, +3.0% d13C). These marked differences suggest that two mechanisms of dolomitization were involved in the diagenesis of Wichita dolomites. Isotopically light d18O values along the arcuate belt of higher porosity tidal-flat facies are consistent with early, seawater-dominated dolomitization. The heavier d18O values exhibited by updip dolomites are typical of dolomitization by refluxing brines. Similar heavy d18O values are encountered in overlying Lower Clear Fork dolomites (average: +2.5% d18O). This isotopic similarity suggests that dolomitization of nonporous Wichita tidal-flat facies may have been caused by evaporatively concentrated brines generated during Lower Clear Fork time. Limestone isotope data from Fullerton come entirely from the Lower Clear Fork. Values of d13C range from +1.61 to +5.46% PDB (n = 30), whereas d18O data range from 3.89 to 1.55% PDB (Figure 24). These data are very different from previously reported values for other calcite-bearing samples in the Leonardian and Guadalupian. Most reported d18O values for Guadalupian calcite have ranged from about 7.6 to 10.4 (Leary, 1985; Vogt, 1986; S. Ruppel, 1996, unpublished data from the Grayburg Formation at South Cowden field). These values, however, were recorded from replacement calcite interpreted to have precipitated from meteoric water possibly sourced from deep basin fluids. The d13C values from these replacement calcites are also very depleted (19 to 30% PDB), suggesting precipitation from bacterially mediated sulfate reduction (Leary, 1985; Vogt, 1986; Ruppel and Cander, 1988b). The more normal d13C values for the Lower Clear Fork limestones at Fullerton indicate a very different origin for these calcites. Lower Clear Fork d18O values are very similar to the current best estimate for marine calcite precipitates from seawater during the middle Permian (2.8% PDB; Given and Lohmann, 1985; Lohman and Walker, 1989). This suggests that Lower Clear Fork limestones contain a preserved record of original seawater chemistry and by extension that these rocks have undergone relatively little chemical alteration.

Karst Development Karst Fabrics Karst diagenesis can be a major factor in reservoir heterogeneity and compartmentalization in carbonate successions (Loucks, 1999). Evidence of karst-

related diagenesis and dissolution is common in the lower part of the reservoir section at Fullerton field; karst features occur within the Wichita and in the top of the Abo. These rocks are variable in fabric but include four basic types: (1) polymict conglomerates, (2) monomict breccias, (3) fractured and tilted beds, and (4) void-filling cement. The predominant style of polymict conglomerate typically consists of rounded clasts of multiple peritidal lithologies (Figure 26). Clasts range in size from a few millimeters to several centimeters in maximum dimension and are commonly subequant and rounded. Clasts are commonly enclosed in mudstone or abut one another at stylolitic contacts. Polymict fabrics are most common in the middle of the Wichita (Figure 26). Intervals of polymict conglomerate of at least 7.5 to as much as 18 m (25 to as much as 60 ft) thick are present in the Exxon FCU (Fullerton Clear Fork Unit) 6122 core. The multiple facies character of these clasts and their rounded character indicate that they were formed by sediment transport. Their discontinuous nature and their association with other features indicative of karst processes suggest that they originated as cavefill deposits. Although probably not true polymict conglomerates, superficially similar deposits that also indicate karst processes are present at the contact of the Wichita and the Abo at the downdip edge of the field area. These rocks are commonly characterized by a mixture of two or more facies types, including Wichita tidal-flat facies, green silty carbonate, dark-gray silty carbonate, and Abo subtidal facies. Instead of being interbedded with one another, the first three surround the Abo facies in many cores. This suggests that these polymict conglomerates actually represent dissolution and/or erosion of the top of the Abo and subsequent infilling of the irregular surface or differential compaction of the Abo (L1) and overlying transgressive Wichita sediments (L2) at the Abo–Wichita contact. In rare instances, there are also features suggestive of collapse brecciation. All of these features have been observed in analogous outcrops of the Abo –Wichita contact in Apache Canyon in the Sierra Diablo (Kerans et al., 2000). The outcrop succession reveals that karst features (sinkholes and caves) were formed during post-L1 sea level fall and then filled with transgressive L2 tidal-flat deposits (i.e., Wichita facies). These tidal-flat facies, in some instances, were brecciated and intermixed with the underlying Abo in sinkholes and collapsed caves. Monomict breccias or conglomerates consist of broken and rotated clasts of constant lithology and

380 / Ruppel and Jones

FIGURE 26. Core slab and image log photographs of polymict conglomerates in the Wichita Formation of probable karst origin. (A) Exxon FCU 6122, depth: 2195 m (7201 ft). (B) Amoco University Consolidated IV-25, depth: 2199 m (7213 ft). Cores 10 cm (4 in.) wide. (C) Image log of conglomerate in the Wichita. Exxon FCU 2564, depth: 2186– 2187 m (7171– 7174 ft). Image is 63.5 cm (25 in.) wide.

facies (Figure 27). These rocks are restricted to the Wichita, commonly the middle of the section. Commonly associated with these deposits are fractures, sediment infill (cracks and fissures), void-filling cement (chiefly anhydrite), and other evidences of dissolution. Most of these features can also be formed by nonkarst processes in tidal-flat intervals not exposed to true karst (e.g., tepee formation is commonly accompanied by broken and rotated blocks, cracks and fissures, and cement and sediment infill). Thus, it is possible that some of these deposits may not be the result of true karst processes. However, their thickness, abundance, and association with other features of karst formation suggest that many are also karst related.

Fractured and tilted beds are also observed in some cores, especially in the middle of the Wichita section. Tilted beds have clearly been formed by postdepositional collapse. In the Exxon FCU 5927 core, they form part of a succession that very much resembles the classic cave fill-cave roof succession described by Loucks (1999). The succession in the Exxon FCU 5927 well consists of 6 m (20 ft) of polymict cave-fill conglomerate composed of mixed tidal-flat facies overlain by 20 ft (6 m) of fractured and locally tilted but apparently generally in-situ beds of tidal-flat and subtidal facies probably representative of a cave roof. Large zones of void-filling anhydrite cement are also strong indicators of karst-related dissolution. Zones of massive anhydrite as much as 0.5 m (1.5 ft) thick are present in cores in the Wichita at Fullerton field (e.g., Exxon FCU 6122, 5927). Smaller anhydrite-filled voids and fractures are ubiquitous in the Wichita and at the Wichita–Abo contact and also point to late cementation of dissolution voids by diagenetic fluids associated with reflux dolomitization of the succession.

Causes and Timing of Karst Outcrop studies demonstrate that major karsting of the Leonardian sequence occurred at the sea level fall and rise event that is defined by the L1-L2 sequence boundary (Kerans et al., 2000). In downdip areas of Fullerton field, cores demonstrate this same relationship. Here, the top surface of the Abo outer platform facies succession (L1) is karsted and is infilled and overlain by brecciated Wichita tidal-flat facies (L2). In updip areas, however, establishing a

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 381

FIGURE 27. Core slab photos of monomict breccias from the Wichita. (A) Rotated clasts of tidal-flat facies. Exxon FCU 5927, depth: 2203 m (7230 ft). (B) Tilted blocks of laminated peritidal facies. Exxon FCU 5927, depth: 2199 m (7215 ft). (C) Brecciated clasts of peloidal wackestone. Amoco University Consolidated V 15, depth: 2103 m (6899 ft). (D) Clasts of peritidal mudstone-wackestone. Exxon FCU 5927, depth: 2204 m (7233 ft). Cores are 10 cm (4 in.) wide. top L1 surface (consisting of Abo subtidal sediments downdip and lower Wichita tidal-flat sediments updip) developed local caves, sinkholes, and an irregular topography. During the subsequent L2 sea level rise, transgressive peritidal deposits of the Wichita filled the irregular karsted surface, including sinkholes and caves. With continued sedimentation and compaction, parts of the overlying L2 (upper Wichita tidal flat) succession underwent local brecciation and collapse probably because of stress differences set up over underlying karst features. This scenario is consistent with outcrops (Kerans et al., 2000) and with the distribution of karst features seen at Fullerton field.

Impact of Karsting on Reservoir Quality

spatial and temporal relationship between karst formation and the L1-L2 sequence boundary is more problematic. Most karst features in the Fullerton field area are found in an interval of about 45 m (150 ft) in the middle of the Wichita section. In general, this interval correlates approximately to the interpreted L1-L2 boundary (Figure 5). However, karst features are developed both below and above this horizon, indicating that karst-related diagenesis was not limited to the L1 sequence. Instead, it appears that two types of karst-related processes occurred as outlined below. Primary karsting and dissolution probably occurred during the post-L1 lowstand. At this time, the exposed

Some karst-related deposits at Fullerton most certainly exhibit at least local differences in petrophysical properties (i.e., porosity, permeability, and saturation) from surrounding undisturbed and unaltered deposits. Polymict conglomerates and anhydrite voids are two obvious examples. However, two factors make quantification of the importance of these differences difficult. First, most karst deposits do not record significantly different porosity or permeability from that of surrounding nonkarsted deposits on both wire-line and core data. This is probably because of the fact that most karst fills are composed of the same facies as nonkarsted intervals. Anhydrite void fills are an obvious exception to this because they contain no porosity or permeability; but they are generally very small and probably of little impact on reservoir flow.

382 / Ruppel and Jones

Second, with the exception of areas in which there are cores, the distribution of karst features in the reservoir section is not definable. Efforts to identify karst fill using logs and 3-D seismic data are thwarted by the similar lithological and petrophysical properties these features share with surrounding rocks. It is tempting to conclude from these observations that karst features have no impact of reservoir heterogeneity or fluid flow. However, there are anomalies in water production and flow rates in the Wichita that cannot be readily explained by matrix petrophysical properties (T. Anthony, 2003, personal communication). These phenomena may be the result of karst development.

RESERVOIR IMAGING Accurate definition of reservoir architecture and the distribution of rock fabrics in this architecture is the key to defining improved methods for recovery of hydrocarbons remaining in these systems. We used several methods to better image the reservoir at Fullerton field. Especially important in defining the geologic architecture of the reservoir are (1) the calibration and use of borehole image logs to aid in the identification and mapping of facies, cyclicity, and rock fabrics; and (2) the use of 3-D seismic data to constrain the geologic framework.

Identifying Facies and Cyclicity from Borehole Image Logs Image logs are highly underused in the characterization of carbonate reservoirs. To most, the principal use of such logs is in the identification of fractures. However, image logs also have the potential to accurately image many matrix properties that are key to the proper characterization, modeling, and exploitation of carbonate reservoirs. Traditionally, cores have been obtained to provide the key data needed for constraining the distribution of reservoir facies, cyclicity, and rock fabrics. However, because of cost, the number of cores available is generally far smaller than needed to accurately constrain these elements. If properly calibrated with core observations, borehole image logs can provide most of the required data to construct an accurate reservoir model at a fraction of the cost. Examination, correlation, and comparison of image log character with cores show that seven facies can reliably be identified from the image log in the Abo–Wichita–Clear Fork succession at Fullerton. This is fewer than the 12 facies defined from core studies

but is sufficient to provide necessary information to define major facies successions and cyclicity and to provide a strong basis for accurate correlation and interpretation of facies and cyclicity to nearby wells. Facies recognizable on image logs include (1) tidal-flat facies (Figure 8B), (2) peloid wackestone-packstone (Figure 10B), (3) nodular wackestone-packstone (Figure 9B), (4) cross-bedded grainstone (Figure 13B), (5) fusulinid wackestone-packstone (Figure 14B), (6) karst breccia (Figure 26C), and (7) clay-rich mudstone. Additionally, siltstone-sandstone facies can be identified from many wire-line logs. The identification of tidal-flat facies is especially crucial for defining both reservoir architecture and reservoir quality. Because they generally occupy cycle tops, their definition makes it possible to define cycle boundaries and thereby facilitates cycle-scale correlation. Equally important is the identification of tidal-flat facies for their petrophysical significance. As discussed previously, tidal-flat facies commonly display relatively high porosities but low permeability. Because of this, it is critical for accurate reservoirquality mapping to distinguish tidal-flat rocks from subtidal rocks that are typically higher in permeability. Definition of the fusulinid wackestone facies is also of great importance because these rocks generally represent the deepest water facies in the Leonardian succession and typically are found at cycle bases. They are thus important indicators of sea level rise and guides to cycle definition and correlation. Using calibrated image log responses, it is possible to create a very detailed record of facies in wells having good-quality image logs. Vertical resolution is potentially even better than that attainable from cores (because of uncertainties of core and well depth ties). Cycles defined from image logs range in thickness from 1 to 6 m (3 to 20 ft). Highest facies and cycle resolution was obtained in intervals characterized by alternating subtidal and tidal-flat facies in the Lower Clear Fork because of the marked contrast in image log character of these two facies types.

Imaging Stratigraphic Architecture and Reservoir Development from 3-D Seismic Data Like image logs, 3-D seismic data are remarkably underused in the characterization of carbonate reservoirs. A critical need in developing approaches that lead to improving recovery from reservoirs containing significant volumes of remaining hydrocarbons is a better understanding of the 3-D distribution of reservoir attributes. The approaches outlined in this report for assembling and interpreting well data are

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 383

FIGURE 28. Fullerton field 3-D seismic section (in time) showing general continuity and isopachous nature of Lower Clear Fork and Wichita reservoir intervals and contrasting clinoformal nature of the Abo. Yellow lines are time lines; dotted line defines the top of Abo Formation.

wire-line-log data are inappropriate for the Abo. In most cases, it is not possible to resolve cycle-scale correlations of either facies or time surfaces in such settings. the most critical part of this effort. However, because they are limited to well control, they leave important gaps in our understanding. Three-dimensional seismic data offer valuable data on interwell and extrawell areas (meaning areas of the field where usable well data are absent) and when properly interpreted and applied can greatly improve and, thus, constrain reservoir attribute models. Here, we provide some brief insights into how 3-D data can be used to better image both reservoir framework and porosity distribution. A far more detailed and rigorous application of these 3-D data to the construction of the geological model at Fullerton is described by Zeng (2004).

Constraining Reservoir Architecture Two- and three-dimensional seismic data at Fullerton field provide important guides to the stratal architecture of the reservoir succession. Seismic amplitude sections through the entire reservoir interval reveal that much of the section is characterized by generally parallel seismic reflectors (Figure 28). This is not unexpected, considering the shallow-waterplatform depositional setting indicated by the cores and the apparently subhorizontal correlations suggested by wire-line logs. However, 3-D and some 2-D data suggest a very different architecture for the basal Leonardian, Abo Formation. Dip-oriented seismic lines reveal sets of clinoformal reflections in the Abo that dip generally basinward (toward the east). This apparent clinoformal architecture of the Abo is consistent with observations of clinoformal fusulinid wackestones and packstones in outcropping Abo-equivalent sections in the Sierra Diablo. These clinoforms demonstrate that conventional horizontal correlations of

Defining Reservoir Quality Seismic data can also be robust indicators of porosity distribution in carbonate reservoirs. Zeng (2004) demonstrated the strong agreement between 3-D seismic impedance data and reservoir porosity at Fullerton field. Because of this robust relationship, even simple amplitude extractions reveal field-scale and fieldwide changes in porosity development that are important for understanding controls on porosity development and for defining interwell and extrawell distribution of porosity. Figure 29 is an amplitude extraction map for HFS 2.1 in a small area of the field created to examine possible infill drilling locations. Note that whereas the northern half of the map displays high negative amplitudes indicative of high porosity, the southern half of the area is characterized by markedly lower amplitude. The porosity distribution revealed by this map indicates that only those wells in the northern half of the target area are likely to encounter good porosity in this zone. It should be pointed out that this information on porosity distribution is only obtainable from the 3-D data volume; the quality of the well logs in the area is too poor to determine porosity. Areas of poor well control like these (caused by either the lack of wells or poor-quality logs) are common through the field, and they greatly compromise efforts to develop a fieldwide strategy to identify and target oil resources. This is readily apparent from a comparison of a porosity map based on well logs and a seismic amplitude map for HFS L2.0 (Figure 30). The value of the seismic data volume is especially apparent in the southern end of the field where the limited well control does not accurately image the

384 / Ruppel and Jones

FIGURE 29. Map of negative amplitude data extracted from Fullerton 3-D data showing porosity distribution in the L2.1 sequence. Data show that proposed wells in the northern half of the area will encounter reservoir porosity, whereas those in the southern half will not. east – west-trending area of high porosity shown by the amplitude map (compare Figure 30a with b). It is apparent that basic amplitude data like these provide a powerful supplement to well control in defining and predicting porosity distribution in the field.

RESERVOIR FRAMEWORK A critical component of a robust reservoir model is a geologically constrained reservoir framework. Geologically accurate models must be based on correlations of time-stratigraphic units, the most readily correlative of which are cycle and sequence boundaries. At Fullerton field, we used correlations of cycle tops to construct the reservoir model for the Clear Fork. Although these correlations are ultimately based on correlations of wire-line logs, the underlying basis for the interpretation and correlation of these logs is a knowledge of 1-D and 2-D facies and cycle-stacking relationships developed from integrated studies of cores and outcrops.

Cycle-based Flow-unit Definition Reservoir flow units are most appropriately based on the definition and mapping of depositional cycle boundaries. This is because when properly defined

and correlated, depositional cycles represent the best available indicator of original depositional surfaces or time lines. The procedure for identifying and using cycle boundaries for flow-unit definition has been well described by Ruppel and Ariza (2002) and Lucia and Jennings (2002) for the South Wasson Clear Fork reservoir. Ideally, the correlation of depositional cycle tops should be based on a log that can be directly tied to facies and one that is independent of diagenesis or porosity development. The gamma-ray log in certain ideal settings serves this function. However, throughout most of the Leonardian carbonate succession (in fact, throughout most of the Permian carbonate section in the Permian basin), the gamma-ray log is not accurate for detailed correlation because of variable volumes of uranium, potassium, and thorium. The spectral gamma-ray log can help to distinguish variations in these elements and, thus, is useful in separating clastic-rich sections (that contain high levels of potassium and thorium). Because small volumes of clastics are commonly associated with tidal-flat facies, the spectral gamma ray is locally helpful in distinguishing tidal-flat facies from subtidal facies (Ruppel, 1992, 2002). However, variations in uranium content are not always systematic, and because of this, gammaray response commonly varies independently of facies. Resistivity logs can also locally be used to separate tidal-flat facies from subtidal facies on the basis of differences in saturation (Ruppel, 2002). However, even under ideal circumstances, neither gamma-ray logs nor resistivity logs can accurately depict facies in Leonardian rocks. Accordingly, we used porosity logs to define and correlate facies and cycle tops in the Lower Clear Fork section at Fullerton field. The basis for this approach comes from integrated studies of cores and wire-line logs at Fullerton that demonstrate two important attributes of Leonardian cycles. First, cycle-capping facies are either grain-rich subtidal or tidal-flat facies. Second, porosity is most commonly associated with these facies and, therefore, is most typically developed at cycle tops. These relationships are key to the use of wire-line logs for correlation of cycles and of flow units. It is important to understand that there are limitations to the accuracy of the porosity-log correlation method. For example, because of diagenesis, all carbonate facies display some variations in porosity. Cycletop facies in the Leonardian, whether subtidal grainstone or tidal flat, do exhibit lateral changes in porosity. So locally, cycle tops defined from porosity may be slightly mispositioned (e.g., because of a local decrease

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 385

FIGURE 30. Porosity development in HFS 2.0 (upper Wichita) across entire field based on wire-line logs (a) and 3-D seismic data (b). Note that 3-D amplitude patterns suggest that well-based porosity mapping is inaccurate in some areas.

in porosity in the cycle-top facies or a local increase in porosity in the overlying cycle-base facies). From a strict chronostratigraphic point of view, this means that some cycle tops are incorrect, and that cycle correlations locally cross time lines. From a reservoir modeling point of view, however, this result is actually probably preferable. That is because these Leonardian rocks uncommonly contain cycle-base flow barriers, and flow probably does locally cross cycle boundaries. In any case, the porosity-log correlation method, if properly constrained by core and outcrop calibration of facies, porosity, and cyclicity, is the most geologically sound basis for constructing cycle correlations and establishing the basis for true flow-unit correlations. It is also important, however, to correlate geologically defined cycle tops at the highest resolution possible. For example, outcrop studies demonstrate that cycles less than 3 m (10 ft) thick can be correlatable across large areas of the platform (Ruppel et al., 2000). Where wire-line data permit, an effort should be made to correlate these thin cycles through the reservoir as well, even if later upscaling is planned for

reservoir modeling. This is important for two reasons. First, coarser scale correlations are much more likely to be in error. At Fullerton, we found that early correlations made at the sequence or cycle set scale were later proven to be off by a cycle or two after we recorrelated the succession at the cycle scale. From a reservoir modeling point of view, this means that flow units defined by coarse-scale correlations are more likely to cross-connect flow layers than those defined by finer scale correlations. Second, upscaling (grouping of cycles into thicker flow-unit packages for modeling) is more likely to retain the original geological architecture if based on fine-scale (i.e., cyclescale) correlations. Accordingly, we attempted to correlate the reservoir succession at the highest possible level of detail supported by outcrop and core observations and wire-line resolution.

Lower Clear Fork Reservoir Architecture The robustness of the use of porosity logs for defining facies and cyclicity is apparent from core and log relationships in HFS L2.2. A comparison of

386 / Ruppel and Jones

FIGURE 31. Comparison of coredefined facies and cyclicity with porosity logs in subtidal cycles of HFS L2.2. Exxon FCU 6122. The systematic relationship between cycle top facies and higher porosity permits porosity logs to be used for cycle and flow unit correlation. lateral changes in facies stacking in L2.1 cycles suggest local variations in sediment accumulation patterns that may have been caused by topographic relief on the platform during L2.1 flooding. The resultant complex vertical and lateral facies distribution patterns make accurate definition of cycles difficult. As a result, cycles defined for L2.1 may actually reflect combinations of cycles or cycle sets.

Wichita Reservoir Architecture

core data and porosity logs in the Exxon FCU 6122 well (Figure 31) shows that porosity is nearly entirely associated with cycle-top, grain-rich subtidal facies. Thus, porosity logs can be used to define both facies and cyclicity. Other cored wells in the field exhibit the same core and log relationship. Using this relationship, we defined and correlated 15 cycles in the Lower Clear Fork. The average thickness of these cycles is about 5 m (17 ft) for L2.1 and L2.2 throughout the entire interval. However, cycle thickness varies systematically by sequence in this interval. High-frequency sequence L2.2 cycles average about 3.3 m (11 ft) in thickness, whereas L2.1 cycles, with the exception of low-accommodation tidal-flat cycles at the base and top of the sequence, are nearly twice as thick. This is probably caused by two factors. First, L2.1 deposits record the maximum flooding of the platform and probably the development of maximum accommodation. The overall upward thinning of cycles from L2.1 to L2.2 is consistent with an overall upward decrease in accommodation. Second,

Rigorous cycle definition is not possible for the Wichita because of the preponderance of very lowaccommodation, tidal-flat facies. Both core and outcrop studies show that rocks deposited in such settings uncommonly display systematic trends in vertical facies stacking and generally exhibit very low lateral facies continuity. Accordingly, patterns of depositional facies do not define extrinsic controls (e.g., sea level rise and fall) but, instead, local controls on sediment accumulation (e.g., paleotopography and climate). Thus, for the Wichita, it is necessary to use diagenetic features to define the reservoir framework. Key diagenetic features used to define Wichita cycles are mineralogy and porosity. In the northern part of the field, the upper Wichita contains multiple intervals of low-porosity limestone that we have interpreted to be cycle bases (Figure 18a). Although interpreted largely from diagenetic relationships, cycles are closely tied to depositional surfaces and, thus, represent a good approximation of time surfaces. These surfaces are also especially useful from a reservoir point of view because they define layers of low and high porosity and permeability. The top four cycles in the Wichita are defined on the basis of these cyclic dolostone (high-porosity)-limestone (low-porosity) couplets.

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 387

However, no limestone is present in the upper Wichita in the southern part of the field area nor in the lower part of the Wichita in any part of the field (Figure 18a). In these areas, it was necessary to use porosity trends alone to construct the reservoir framework. Although it is still likely that porosity variations in these areas are closely tied to diagenesis associated with depositional surfaces (i.e., cycle tops), we have little independent evidence (i.e., mineralogical variations) with which to demonstrate this. Accordingly, the framework established for these parts of the Wichita is far less geologically robust than that defined for other parts of the reservoir. For this part of the reservoir, we defined and correlated 12 surfaces on the basis of porosity. We have confidence that these surfaces are subparallel to time surfaces on the basis of their parallelism to overlying Lower Clear Fork cycle-top surfaces and to a middle Wichita marine flooding surface. However, we cannot tie them rigorously to cyclicity.

Abo Reservoir Architecture The Abo is dominated by clinoformal bedding typical of outer-ramp carbonate deposits; this is demonstrated both by outcrop studies (Kerans et al., 2000) and by 3-D seismic data at Fullerton (Figure 28). It is therefore certain that the Abo architecture differs significantly from the generally subparallel character of the depositional surfaces of the platform-top Wichita and Lower Clear Fork successions. Outcrop studies and seismic studies at Fullerton and elsewhere (Kerans et al., 2000; Ruppel et al., 2000) also demonstrate that clinoformal outer platform successions like the Abo do not contain readily correlatable cyclic successions. This is caused by changes in sources and distributional patterns and to extreme variations in lateral textures and fabrics of these transported sediments. Accordingly, it is not possible to establish an accurate internal architecture for these deposits. For purposes of reservoir modeling, we created a series of conceptual clinoform surfaces to constrain the reservoir architecture. Although these surfaces do not accurately describe the architecture of the Abo, they do illustrate the nonparallel and nonhorizontal nature of these deposits and provide more realistic constraints for modeling of reservoir attributes.

Reservoir Model The architecture of the reservoir framework developed for the Clear Fork reservoir at Fullerton is depicted by Figure 32. Key aspects of this model are (1) the subhorizontal and subparallel nature of Clear

Fork and upper Wichita (L2) surfaces; (2) the effective pinch-out of lower Wichita (proximal L1) surfaces at the facies boundary from Wichita tidal flats to Abo subtidal; and (3) the clinoformal nature of Abo (distal L1).

Fieldwide Patterns of Porosity Distribution Core and log data from wells that penetrate the Abo indicate that the Abo locally contains high porosity. However, well control is too sparse and incomplete to accurately map porosity distribution. Wang and Lucia (2004) presented a realization of the 3-D distribution of Abo porosity using a full-field model based on available porosity data and a conceptual geological framework. However, the accuracy of this porosity distribution must be considered relatively low. Mapping of phih (porosity  thickness) from wireline logs demonstrates that the Wichita Formation contains higher total phih (and higher average porosity) than either the Abo or the Lower Clear Fork. In general, maps show that the highest phih lies along the field structural crest, suggesting that porosity development is related to structure. However, a more detailed examination of phih by stratigraphic horizon shows that phih is related more to platform position than structure. The upper Wichita (HFS L2.0 TST) displays a well-defined, arcuate belt of higher phih (Figure 30) that generally parallels but is slightly landward of the inner-outer platform boundary (the Wichita– Lower Clear Fork facies boundary). Decrease in phih to the east is a function of the decreasing thickness of the Wichita near the facies transition (Figure 19). The decrease in phih to the northwest, however, is not associated with any change in depositional facies; Wichita rocks comprise similar peritidal tidal-flat facies across the entire area. Accordingly, the cause of porosity development must be more a function of diagenesis than deposition. A model for this diagenesis is presented in the following section. Lower Clear Fork HFS L2.1 also displays an arcuate trend in porosity development that is very similar to that seen in the upper Wichita (Figure 33; see also Figure 25b). In contrast with the Wichita, there is evidence that this porosity trend is, at least in part, associated with facies. Areas of highest porosity are associated with middle-ramp, grain-rich packstones. Lower porosity areas are associated with outer-ramp fusulinid-rich wackestones downdip and mud-rich packstones updip. It is probable that the high-phih belt represents a low-energy ramp crest. However, the

388 / Ruppel and Jones

FIGURE 32. Diagrammatic cross section showing framework used for reservoir model construction and general distribution of major rock-fabric classes at Fullerton Clear Fork field. Class 1 rock fabrics display the highest permeability relative to porosity, whereas class 3 fabrics display the lowest. It should be noted that the Lower Clear Fork displays far more complex lateral variations in rock fabrics than depicted here. Datum is sea level.

change in phih appears also to be significantly related to diagenesis. Areas of high porosity are also areas of high limestone abundance. An almost 1:1 relationship exists between the presence of limestone and high-porosity areas (compare Figures 18b, 33). No similar trend of high phih is apparent in HFS L2.2. However, like L2.1, areas of highest porosity are associated with limestone. For L2.2, this is dominantly in the southern end of the field; however, a small area of limestone also exists in the northwest corner of the field (Figure 18b).

Models for Porosity Development Porosity is developed in all parts of the Abo – Wichita– Lower Clear Fork reservoir section. However, there are obvious areal variations in porosity that reflect a combination of depositional, diagenetic, and structural controls on reservoir development. Most reservoir intervals exhibit spatial variations in porosity development that parallel platform paleotopography. This is particularly apparent in the Wichita

HFS L2.0 and in the Lower Clear Fork HFS L2.1 (Figures 30, 33). In each of these cases, belts of high porosity are situated immediately up depositional dip (landward) from the position of the underlying inner-outer platform margin (Wichita – Abo facies boundary). However, the mechanism for porosity development must differ for the Wichita, which comprises peritidal facies that display no systematic variations across the area, and the Lower Clear Fork, which consists of subtidal rocks of three facies tracts. The absence of depositional facies variations in the Wichita indicates that porosity formation is the result of diagenetic processes. Stable isotope data suggest that high-porosity dolomites in the Wichita are the result of seawater-dominated diagenesis (Figures 25a, 34a). These rocks display light d18O values (average: 0.0% PDB) that are consistent with seawater dolomitization (based on an approximate 3% fractionation from penecontemporaneous seawater calcite; Land, 1980). The restriction of these high-porosity rocks to the outer margin of the Wichita

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 389

FIGURE 33. Map of wire-line phih showing distribution of porosity in the Lower Clear Fork HFS 2.1. Areas of high porosity in HFS L2.1 correlate with areas of abundant limestone. Compare with Figure 17. tidal flat suggests that porosity development may be the result of early seawater-dominated dolomitization and stabilization along the seaward margin of the tidal flat. Dolomitization may have been favored along this seaward margin by the development of somewhat shallower water conditions, perhaps like those seen in tidal-flat island complexes (Figure 34a). The Lower Clear Fork HFS 2.1 also displays a trend in porosity development that closely parallels the underlying Wichita– Abo and Wichita– Clear Fork facies transitions (Figures 25b, 33). In this case, however, both mineralogy and porosity development are related to facies. Core descriptions imply that the trend of high porosity is coincident with a platform ramp crest, an area characterized by more common ooid-bearing, grain-rich packstones and grainstones (e.g., Figure 13C) and by locally more abundant tidalflat caps. Ramp-crest development was probably controlled by inherited paleotopography over the L1 (Wichita –Abo) inner to outer ramp margin and/or differential subsidence over deep-seated faults in the same way that the position of the Wichita– Abo facies transition is most likely controlled by such deep

FIGURE 34. Models depicting three major stages of diagenesis and porosity formation in the upper Wichita and the Lower Clear Fork (L2 sequence). (a) Early seawater dolomitization: Wichita, L1-L2. Wichita porosity development is associated with early seawater dolomitization along the outer margins of the tidal-flat complex. (b) Early seawater cementation and stabilization: Lower Clear Fork, L 2.1. Porosity is related to preservation of primary porosity in Lower Clear Fork ramp-crest limestones. (c) Reflux dolomitization. Most of the Wichita and Lower Clear Fork was dolomitized by evaporatively concentrated, reflux brines. See text for discussion.

390 / Ruppel and Jones

structures. The ramp-crest trend is an area of abundant calcite; lower porosity areas to the east (inner ramp) and west (outer map) are nearly entirely dolomite (see Figure 18). Strontium and oxygen isotope data ( J. Kaufman, 1991, personal communication; this study) from the calcites indicate that these rocks are essentially unaltered original marine precipitates. Oxygen isotopes (d18O = 3.0%) are virtually identical to best estimates of marine precipitates of middle Permian age (d18O = 2.8%; Given and Lohmann, 1985). Collectively, these data suggest the HFS L2.1 porosity trend is the result of early marine calcite cementation in well-agitated and well-oxygenated conditions along the ramp crest (Figure 34b). Lower porosity in updip dolomites is probably the result of predolomitization compaction of these muddier sediments and possibly of porosity occlusion by dolomite and anhydrite cementation. Compaction and dolomitization, and, thus, porosity loss, were probably limited in the ramp crest because of early calcite cementation and stabilization and the grain-rich textures that are dominant there. Most Wichita and Lower Clear Fork rocks outside these two high-porosity corridors exhibit lower porosities and a very different stable isotopic character. These rocks contain much heavier oxygen isotopes (d18O average 3.0% PDB) that are indicative of dolomitization by evaporatively concentrated refluxing brines (Saller and Henderson, 1998; Ruppel, 2002). Such brines may have been generated at multiple times during the Leonardian. Conditions necessary for reflux were probably developed at each of the documented sequence boundaries (Figure 34c). However, it seems likely that the greatest potential for major reflux was at or soon after the L1-L2 (Wichita) or L2-L3 (Lower Clear Fork– Tubb) sea level lowstands (Figure 34c). The similarity in isotopic character throughout both Wichita and Lower Clear Fork reflux dolomites suggests that all were dolomitized by a single massive reflux event, presumably at about the L2-L3 sea level fall and rise. Although most reflux dolomites at Fullerton have low porosities, grain-rich, Lower Clear Fork dolomite facies typically exhibit good porosity and permeability (e.g., Figure 13D). This demonstrates the key function of depositional texture in porosity preservation in dolomites. The lack of porosity in updip Wichita reflux dolomites, as well as in Lower Clear Fork mud-rich dolomite facies, probably reflects predolomitization compaction and/or porosity reduction by dolomitization and sulfate emplacement during reflux.

Rock Fabrics and Permeability Distribution In addition to a geologically defined stratigraphic framework and an accurate distribution of porosity, a robust reservoir model must contain a realistic distribution of permeability. To calculate and distribute permeability at Fullerton, we used the rock-fabric approach developed by Lucia (1995). This approach is based on defining and mapping key rock fabrics. Figure 32 displays the stratigraphic distribution of rock fabrics and associated petrophysical classes (groupings of rock fabrics having similar relationships between porosity and permeability) in the Fullerton reservoir. Abo outer platform skeletal packstones are dominated by rock fabrics with interparticle pores representative of petrophysical class 1 (which exhibits the highest permeability for a given porosity). Wichita tidal-flat facies, by contrast, comprise petrophysical class 2 rock fabrics (which have the lowest permeability for a given porosity). Rocks of the Lower Clear Fork are more complex than either the Abo or Wichita in terms of type and distribution of rock fabrics. As suggested by Figure 32, subtidal Lower Clear Fork deposits include both petrophysical class 1 and 2 rocks (and locally contain mappable class 3 tidal-flat facies). The distribution of class 1 and 2 rock fabrics is dominantly a function of diagenesis (principally mineralogy and pore type) and secondarily related to depositional facies. For example, most rampcrest grain-rich limestone packstones are class 2 rocks; thus, their distribution can be mapped with wire-line logs that define mineralogy (e.g., dual-porosity log suites, PE log). Other variations in rock fabrics and petrophysical class in the Lower Clear Fork are a function of dolomite crystal size, anhydrite cementation, development of moldic pores, and depositional texture. We examined 950 thin sections and obtained 705 new measurements of core porosity and permeability and 30 special core analysis measurements to define these properties across the field. A detailed discussion of the methodology and results of the characterization and mapping of these rock fabrics and calculation of permeability at Fullerton is available in Jones and Lucia (2004).

SUMMARY AND CONCLUSIONS Both the procedures used to develop the reservoir framework at Fullerton field and many of the attributes of this framework offer important guidelines for the characterization of shallow platform carbonate reservoirs and for the development of predictive

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 391

models of the distribution of reservoir properties. Key geological findings from this study include the following. The Leonardian reservoir succession in the Permian basin consists of three formations or facies successions, each of which is characterized by distinctive facies, cyclicity, depositional architecture, porosity development, and petrophysical relationships. The Abo consists largely of porous outer-ramp, open-marine, fusulinid and crinoid facies whose clinoformal architecture is clearly expressed on seismic data. Lateral facies continuity is poor, and as a result, porosity distribution is complex and probably highly discontinuous. The relationship between permeability and porosity, however, is among the best in the reservoir. The Wichita tidal-flat rocks comprise both highstand systems tract updip equivalents of the Abo (Leonardian sequence L1) and TST updip equivalents of the basal Lower Clear Fork (Leonardian sequence L2). Cyclicity is poorly developed; reservoir architecture is more controlled by diagenesis than by depositional facies. Although these deposits locally contain high porosity, they exhibit relatively low permeability and display poor, small-scale continuity. Areas of high-porosity development are a result of early dolomitization and stabilization along the outer margins of the inner-platform, tidal-flat complex. Cycle-base limestones display very low porosity and permeability and act as local fluid-flow baffles between higher porosity cycle-top dolostones. The Lower Clear Fork consists of a dominantly subtidal succession of three high-frequency sequences, each of which records sea level rise (transgression) and fall (regression). Reservoir development is largely restricted to subtidal facies (late transgression and early highstand). Highest porosity and permeability are associated with undolomitized grain-rich, subtidal packstones and grainstones of the ramp crest that have been relatively unaffected by major diagenesis. Karst modification of the reservoir (including inclined beds and monomict and polymict cave-fill breccias) is widespread at and around the L1-L2 sequence boundary. However, karst features are resolvable only with cores or image logs, and neither wire-line log nor core analysis data reveal definitive differences in petrophysical properties. Nevertheless, well-performance data suggest that karst development may affect fluid flow. Wire-line logs must be used for constructing the reservoir framework. However, gamma-ray logs are

unreliable for this purpose because of the presence of highly variable volumes of uranium that show no consistent relationship to facies. We found porosity logs best suited for high-resolution, cycle-scale correlations, but only if they have been properly calibrated to facies-stacking patterns and facies. Wire-line logs do not always provide the best realization of porosity distribution. Where good-quality well logs are lacking (e.g., between wells and in areas with older logs) simple amplitude extractions from 3-D seismic data provide superior resolution of both reservoir architecture and the distribution of reservoir porosity. All of the forgoing findings have significant impact on reservoir heterogeneity, architecture, and fluid flow. It is thus critical that all be incorporated into reservoir models for accurate imaging of reservoir properties and meaningful simulation of fluid flow.

ACKNOWLEDGMENTS The results presented in this chapter are part of continuing research into the styles and causes of heterogeneity in shallow-water platform carbonate reservoirs in the Permian basin by the Bureau of Economic Geology. Frequent, sometimes vigorous, discussions with colleagues at the Bureau, including Charlie Kerans, Robert Loucks, Jerry Lucia, and Jim Jennings, have been especially helpful in formulating the interpretations presented herein. Principal funding for the study was provided by the U.S. Department of Energy, ExxonMobil Corporation, and the University of Texas System. Additional support was provided by member sponsors of the Bureau’s Carbonate Reservoir Characterization Research Laboratory, including Anadarko, Aramco, BP, Chevron, ExxonMobil, Great Western Drilling, Kinder Morgan, Marathon, Occidental Petroleum, Petroleum Development Oman, Shell International, Statoil, and TotalFinaElf. Special thanks are extended to David Smith, Terry Anthony, Steve Krohn, and Amy Powell of ExxonMobil and Jeff Simmons, Craig Kemp, and John Stout of Oxy for their contributions to organizing the project and providing data. David Smith and Terry Anthony have been especially helpful in proving insights into the field geology and engineering issues. We also express our appreciation to Stephen Hartman and Tim Hunt of the University of Texas System West Texas Operations Office for providing both collaborative funding and data.

392 / Ruppel and Jones

REFERENCES CITED Atchley, S. C., M. G. Kozar, and L. A. Yose, 1999, A predictive model for reservoir characterization in the Permian (Leonardian) Clear Fork and Glorieta formations, Robertson field area, west Texas: AAPG Bulletin, v. 83, p. 1031 – 1056. Bebout, D. G., F. J. Lucia, C. R. Hocott, G. E. Fogg, and G. W. Vander Stoep, 1987, Characterization of the Grayburg reservoir, University Lands Dune field, Crane County, Texas: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 168, 98 p. Bein, A., and L. S. Land, 1982, The San Andres carbonates in the Texas Panhandle: Sedimentation and diagenesis associated with Mg-Cl brines: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 121, 48 p. Dutton, S. P., E. M. Kim, R. F. Broadhead, C. L. Breton, W. D. Raatz, S. C. Ruppel, and C. Kerans, 2005, Play analysis and digital portfolio of major oil reservoirs in the Permian basin: University of Texas at Austin, Bureau of Economic Geology Report of Investigations No. 271, 287 p., CD-ROM. Fischer, A. G., and M. Sarntheirn, 1988, Airborne silts and dune-derived sands in the Permian of the Delaware basin: Journal of Sedimentary Petrology, v. 58, p. 637 – 643. Fitchen, W. M., M. A. Starcher, R. T. Buffler, and G. L. Wilde, 1995, Sequence stratigraphic framework and facies models of the Early Permian platform margins, Sierra Diablo, west Texas, in R. A. Garber and R. F. Lindsay, eds., Wolfcampian – Leonardian shelf margin facies of the Sierra Diablo — Seismic scale models for subsurface exploration: West Texas Geological Society Publication 95-97, p. 23 – 66. Garber, R. A., and P. M. Harris, 1990, Depositional facies of the Grayburg/San Andres dolomite reservoirs; Central Basin platform, Permian basin, in D. G. Bebout and P. M. Harris, eds., Geologic and engineering approaches in evaluation of San Andres/Grayburg hydrocarbon reservoirs — Permian basin: University of Texas at Austin, Bureau of Economic Geology, p. 1 – 20. Given, R. K., and K. C. Lohmann, 1985, Derivation of the original isotopic composition of Permian marine carbonates: Journal of Sedimentary Petrology, v. 55, p. 430– 439. Hardie, L. A., and P. Garrett, 1977, General environmental setting, in L. A. Hardie, ed., Sedimentation on the modern carbonate tidal flats of the northwest Andros Island, Bahamas: Baltimore, Johns Hopkins University Press, p. 12 – 50. Holtz, M. H., and C. M. Garrett, 1990, Geologic and engineering characterization of Leonardian carbonate oil reservoirs: A framework for strategic recovery practices in four oil plays (abs.), in J. E. Flis and R. C. Price, eds., Permian basin oil and gas fields: Innovative ideas in exploration and development: West Texas Geological Society Publication 90-87, p. 76.

Holtz, M. H., S. C. Ruppel, and C. R. Hocott, 1992, Integrated geologic and engineering determination of oil-reserve-growth potential in carbonate reservoirs: Journal of Petroleum Technology, v. 44, no. 11, p. 1250– 1258. Jones, R. H., and F. J. Lucia, 2004, Integration of rock fabric, petrophysical class, and stratigraphy for petrophysical quantification of sequence-stratigraphic framework, Fullerton field, Texas, in S. C. Ruppel, ed., principal investigator, Multidisciplinary imaging of rock properties in carbonate reservoirs for flow unit targeting: University of Texas at Austin, Bureau of Economic Geology, final contract report to Department of Energy, Contract DE-FC26-01BC1535 1, p. 125 – 162. Jones, R. H., and S. C. Ruppel, 2004, Evidence of postWolfcampian fault movement and its impact on Clear Fork reservoir quality: Fullerton field, west Texas, in R. C. Trentham, ed., Banking on the Permian basin: Plays, field studies, and techniques: West Texas Geological Society Fall Symposium: West Texas Geological Society Publication 04-112, p. 207. Kerans, C., and W. M. Fitchen, 1995, Sequence hierarchy and facies architecture of a carbonate ramp system: San Andres Formation of Algerita escarpment and western Guadalupe Mountains, west Texas and New Mexico: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 235, 86 p. Kerans, C., and K. Kempter, 2002, Hierarchical stratigraphic analysis of a carbonate platform, Permian of the Guadalupe Mountains: University of Texas at Austin, Bureau of Economic Geology (AAPG Datapages Discovery Series 5), CD-ROM. Kerans, C., and S. C. Ruppel, 1994, San Andres sequence framework, Guadalupe Mountains: Implications for San Andres type section and subsurface reservoirs, in R. A. Garber and D. R. Keller, eds., Field guide to the Paleozoic section of the San Andres Mountains: Permian Basin Section SEPM Publication 94-35, p. 105 – 115. Kerans, C., F. J. Lucia, and R. K. Senger, 1994, Integrated characterization of carbonate ramp reservoirs using Permian San Andres Formation outcrop analogs: AAPG Bulletin, v. 78, no. 2, p. 181 – 216. Kerans, C., K. Kempter, J. Rush, and W. L. Fisher, 2000, Facies and stratigraphic controls on a coastal paleokarst: Lower Permian, Apache Canyon, west Texas, in R. Lindsay, R. Trentham, R. F. Ward, and A. H. Smith, eds., Classic Permian geology of west Texas and southeastern New Mexico, 75 years of Permian basin oil and gas exploration and development: West Texas Geological Society Publication 00-108, p. 55 – 82. Land, L. S., 1980, The isotopic and trace element geochemistry of dolomite: The state of the art, in D. H. Zenger, J. B. Dunham, and R. L. Ethington, eds., Concepts and models of dolomitization: SEPM Special Publication 28, p. 87– 110. Leary, D. A., 1985, Diagenesis of the Permian (Guadalupian) San Andres and Grayburg formations, Central Basin platform, Permian basin, west Texas: Master’s

Key Role of Outcrops and Cores in Carbonate Reservoir Characterization and Modeling / 393 thesis, University of Texas at Austin, Austin, Texas, 125 p. Leary, D. A., and J. N. Vogt, 1990, Diagenesis of the San Andres Formation (Guadalupian) reservoirs, University Lands, Central Basin platform, in D. G. Bebout and P. M. Harris, eds., Geologic and engineering approaches in evaluation of San Andres/Grayburg hydrocarbon reservoirs— Permian basin: University of Texas at Austin, Bureau of Economic Geology, p. 21 – 28. Lohman, K. C., and J. C. G. Walker, 1989, The d18O record of Phanerozoic abiotic cements: Geophysical Research Letters, v. 16, p. 319 – 322. Longacre, S. A., 1990, The Grayburg reservoir, north McElroy unit, Crane County, Texas, in D. G. Bebout and P. M. Harris, eds., Geologic and engineering approaches in evaluation of San Andres/Grayburg hydrocarbon reservoirs — Permian basin: University of Texas at Austin, Bureau of Economic Geology, p. 239 – 273. Loucks, R. G., 1999, Paleocave carbonate reservoirs: Origins, burial-depth modifications, spatial complexity, and reservoir implications: AAPG Bulletin, v. 83, p. 1795 – 1834. Lucia, F. J., 1995, Rock-fabric/petrophysical classification of carbonate pore space for reservoir characterization: AAPG Bulletin, v. 79, no. 9, p. 1275 – 1300. Lucia, F. J., and J. W. Jennings Jr., 2002, Calculation and distribution of petrophysical properties in the South Wasson Clear Fork field, in F. J. Lucia, ed., Integrated outcrop and subsurface studies of the interwell environment of carbonate reservoirs: Clear Fork (Leonardian-age) reservoirs, west Texas and New Mexico: University of Texas at Austin, Bureau of Economic Geology, final technical report to the Department of Energy, Contract DE-AC26-98BC15105, p. 95 – 142. Major, R. P., G. W. Vander Stoep, and M. H. Holtz, 1990, Delineation of unrecovered mobile oil in a mature dolomite reservoir: East Penwell San Andres unit, University Lands, west Texas: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 194, 52 p. Mazzullo, S. J., 1982, Stratigraphy and depositional mosaics of lower Clear Fork and Wichita groups (Permian), northern Midland basin, Texas: AAPG Bulletin, v. 66, p. 210 – 227. Mazzullo, S. J., and A. Reid, 1989, Lower Permian platform and basin depositional systems, northern Midland basin, Texas, in P. D. Crevello, J. J. Wilson, J. F. Sarg, and J. F. Read, eds., Controls on carbonate platform and basin development: SEPM Special Publication 44, p. 305 – 320. Presley, M. W., 1987, Evolution of Permian evaporite basin in Texas Panhandle: AAPG Bulletin, v. 71, p. 167 – 190. Presley, M. W., and K. A. McGillis, 1982, Coastal evaporite and tidal-flat sediments of the upper Clear Fork and Glorieta formations, Texas Panhandle: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 115, 50 p.

Ruppel, S. C., 1992, Styles of deposition and diagenesis in Leonardian carbonate reservoirs in west Texas: Implications for improved reservoir characterization: Society of Petroleum Engineers Annual Exhibition and Technical Conference, SPE Paper 24691, p. 313 – 320. Ruppel, S. C., 2002, Geological controls on reservoir development in a Leonardian (Lower Permian) carbonate platform reservoir, Monahans field, west Texas: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 266, 58 p. Ruppel, S. C., and E. E. Ariza, 2002, Cycle and sequence stratigraphy of the clear fork reservoir at South Wasson field: Gaines County, Texas, in F. J. Lucia, ed., Integrated outcrop and subsurface studies of the interwell environment of carbonate reservoirs: Clear Fork (Leonardian-age) reservoirs, west Texas and New Mexico: University of Texas at Austin, Bureau of Economic Geology, final technical report to the Department of Energy, Contract DE-AC26-98BC15105, p. 59 – 94. Ruppel, S. C., and D. G. Bebout, 2001, Competing effects of depositional architecture and diagenesis on carbonate reservoir development: Grayburg Formation, South Cowden field, west Texas: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 263, 62 p. Ruppel, S. C., and H. S. Cander, 1988a, Effects of facies and diagenesis on reservoir heterogeneity: Emma San Andres field, west Texas: University of Texas at Austin, Bureau of Economic Geology, Report of Investigations 178, 67 p. Ruppel, S. C., and H. S. Cander, 1988b, Dolomitization of shallow-water platform carbonates by sea water and seawater-derived brines: San Andres Formation (Guadalupian), west Texas, in Sedimentology and geochemistry of dolostones: SEPM Special Publication 43, p. 245 – 262. Ruppel, S. C., W. B. Ward, E. E. Ariza, and J. W. Jennings Jr., 2000, Cycle and sequence stratigraphy of Clear Fork reservoir-equivalent outcrops: Victorio Peak Formation, Sierra Diablo, Texas, in R. Lindsay, R. Trentham, R. F. Ward, and A. H. Smith, eds., Classic Permian geology of west Texas and southeastern New Mexico, 75 years of Permian basin oil and gas exploration and development: West Texas Geological Society Publication 00-108, p. 109 – 130. Saller, A. H., and N. Henderson, 1998, Distribution of porosity and permeability in platform dolomites: Insights from the Permian of west Texas: AAPG Bulletin, v. 82, no. 8, p. 1528 – 1550. Tyler, N., and N. J. Banta, 1989, Oil and gas resources remaining in the Permian basin: Targets for additional hydrocarbon recovery: University of Texas at Austin, Bureau of Economic Geology, Geological Circular 89-4, 20 p. Vogt, J. N., 1986, Dolomitization and anhydrite diagenesis of the San Andres (Permian) Formation, Gaines County, Texas: Master’s thesis, University of Texas at Austin, Austin, Texas, 202 p.

394 / Ruppel and Jones Wang, F., and F. J. Lucia, 2004, Reservoir modeling and simulation of Fullerton Clear Fork field, Andrews County, Texas, in S. C. Ruppel, ed., principal investigator, Multidisciplinary imaging of rock properties in carbonate reservoirs for flow unit targeting: University of Texas at Austin, Bureau of Economic Geology, final contract report to Department of Energy, Contract DEFC26-01BC1535 1, p. 219 – 304. Ye, Q., and S. J. Mazzullo, 1993, Dolomitization of lower Permian platform facies, Wichita Formation, north platform, Midland basin, Texas: Carbonates and Evaporites v. 8, no. 1, p. 55 – 70.

Zeng, H., 2004, Construction and analysis of 3-D seismic porosity inversion models, in S. C. Ruppel, ed., principal investigator, Multidisciplinary imaging of rock properties in carbonate reservoirs for flow unit targeting: University of Texas at Austin, Bureau of Economic Geology, final contract report to Department of Energy, Contract DE-FC26-01BC1535 1, p. 305 – 342. Zeng, H., and C. Kerans, 2003, Seismic frequency control on carbonate seismic stratigraphy: A case study of the Kingdom Abo sequence, west Texas: AAPG Bulletin, v. 87, no. 2, p. 273 – 293.

11

Weissenberger, J. A. W., R. A. Wierzbicki, and N. J. Harland, 2006, Carbonate sequence stratigraphy and petroleum geology of the Jurassic deep Panuke field, offshore Nova Scotia, Canada, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 395 – 431.

Carbonate Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field, Offshore Nova Scotia, Canada John A. W. Weissenberger Husky Energy Inc., Calgary, Alberta, Canada

Richard A. Wierzbicki EnCana Corporation, Calgary, Alberta, Canada

Nancy J. Harland EnCana Corporation, Calgary, Alberta, Canada

ABSTRACT

P

anCanadian Petroleum (now EnCana Corporation) discovered the deep Panuke field in 1998 with the drilling of the PP-3C well. The well was drilled in 90 m (295 ft) of water, 250 km (155 mi) southeast of Halifax, Canada. Subsequent delineation and development drilling has proven a significant gas accumulation. The gas is trapped, by a combined structural-stratigraphic configuration, in the Upper Jurassic reefal and oolitic limestones and dolomites of the Abenaki Formation. The Jurassic carbonate platform on the Scotian Shelf was attached to a metamorphic hinterland, so that the sediments contain varying amounts of siliciclastics. Abundant secondary porosity was encountered, ranging from leached matrix and intercrystalline to vuggy and/or cavernous. Textural, petrographic, and isotopic evidences suggest that deep burial and hydrothermal diagenetic processes caused the porosity. The gas is believed to have been sourced from adjacent Verrill Canyon Formation shales, whereas small amounts of hydrogen sulfide have been isotopically linked to synrift evaporites underlying the Abenaki. The Abenaki is divided into seven third-order depositional sequences, the Abenaki V being the primary gas zone. These sequences have been regionally correlated using geology and a grid of two-dimensional seismic data. Three-dimensional seismic data was used in delineation drilling and reservoir characterization. Deep Panuke is the first, and remains at writing, the only significant hydrocarbon discovery in the Mesozoic carbonates of the continental shelf of eastern North America.

Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215883M883275

395

396 / Weissenberger et al.

INTRODUCTION A recent round of exploration drilling for hydrocarbons on the Scotian Shelf has provided abundant new data on the Jurassic carbonates and associated strata of the region. Beginning in 1998, EnCana Corporation, then PanCanadian Petroleum Ltd., initiated drilling for deep targets below existing production of the Panuke field, discovered in 1986. The prospective reservoir was the Jurassic carbonates of the Abenaki Formation, below oil-bearing siliciclastic sands of the Cretaceous Panuke Formation. The drilling program led to the discovery and delineation of the giant deep Panuke gas field. This chapter proposes new stratigraphic, depositional, and paleogeographic interpretations of the Nova Scotia Jurassic based on the new seismic and well data.

Regional Geology The structural and stratigraphic development of the Scotian Shelf Jurassic is strongly influenced by its location on the western edge of the North Atlantic (Figure 1). The Nova Scotia Jurassic was at the northern end of the Bahama –Grand Bank gigaplatform as described by Poag (1991). This refers to the carbonate depositional system that dominated the eastern margin of North America. Extensional and transtensional tectonics, associated with the opening of the Atlantic Ocean, were dominant during the progressive widening of the Atlantic, from north to south, beginning in the Late Triassic (Jansa and Pe-Piper, 1988). The tectonic setting was essentially a Late Triassic to Early Jurassic rift basin, followed by a passive margin from the Middle Jurassic through today. The basement rocks of the Scotian Shelf are of continental origin, consisting of Devonian granites and late Precambrian to Ordovician metasediments (Welsink et al., 1989), which were then intruded by Devonian-aged granites. Figure 2 shows some of the main structural elements of the Scotian Shelf, including major, fault-bounded structural highs and intervening basinal areas. Three-dimensional (3-D) seismic data from the deep Panuke area suggest that these larger structural elements are broken into smaller horsts and grabens. Jurassic strata are present only in the outer part of the Scotian Shelf in the form of an onlapping wedge of sediments (Figure 3). The gross stratigraphy of this wedge reflects the tectonic history, from initial rift to essentially passivemargin sedimentation. The early, basin-fill sediments

are Late Triassic to Early Jurassic (Sinemurian–Pliensbachian) in age. These comprise synrift siliciclastics and evaporites of the Eurydice and Argo formations (McIver, 1972; Jansa and Wade, 1975). Overlying these are microcrystalline, restricted-marine, occasionally anhydritic dolomites of the Iroquois Formation (McIver, 1972). The Iroquois is overlain by a siliciclastic succession, comprising red beds, immature sands, and associated strata. These are assigned to the Mohican Formation (Given, 1977). Normal-marine carbonate deposition began with the Abenaki Formation (McIver, 1972; revised by Eliuk, 1978). It has been lithostratigraphically subdivided into several members: the Scatarie, Misaine, Baccaro, and Artimon (and the Roseway unit; Wade, 1977). For further discussion, see Eliuk (1978). However, for the Misaine Member, which is a marine shale, these comprise the bulk of the Jurassic carbonate strata of the Scotian Shelf. Regionally, these are time equivalent to shallow-marine and deltaic sediments of the Mic Mac Formation. The lateral relationships between these lithofacies belts will be discussed below. The corresponding Jurassic deep-water shale succession is termed the Verrill Canyon Formation, thought to be time equivalent to sediments from the Mohican Formation to Early Cretaceous strata. This lithostratigraphy (Figure 4) will be discussed from a sequencestratigraphic perspective below. The most comprehensive early work on the Abenaki was presented by Eliuk (1978), with later updates (Eliuk et al., 1986; Eliuk and Levesque, 1989). Other interpretations of the older wells included Jansa and Wade (1975), Jansa and Wiedmann (1982), and Ellis (1984). We and our colleagues have presented preliminary aspects of this project (e.g., Weissenberger et al., 2000; Harland et al., 2002; Hogg and Enachescu, 2003), which is augmented and amended herein.

Background and Methodology Prior to the drilling of the PP3C discovery well in 1998, 53 wells had penetrated the Abenaki Formation on the Scotian Shelf. Only 17 of these wells were cored. The initial stratigraphic interpretation and exploration program was therefore based on these data; wireline logs, cores, and drill cuttings from the aforementioned wells and a regional grid of two-dimensional seismic data. A 3-D survey, originally shot to image shallower, oil-prone Cretaceous strata, covered part of the Panuke area itself. This was used to pick the drilling location of the discovery well (PP-3C).

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 397

FIGURE 1. Paleogeography of the Late Jurassic, approximately 150 m.y. Modified from Scotese (1997). Drilling of the appraisal wells, which followed PP-3C, provided the additional data presented herein. Some additional conventional cores were cut (see Appendix 1). The distribution of all these data is shown in Figure 5. Because of the prohibitive costs of cutting conventional core and the additional risk of encountering lost circulation while coring, a system of rotary sidewall coring combined with formation image logs (and other wire-line logs), was devised to interpret lithofacies and stratigraphy. Sufficient core data are now available to confidently construct facies models and correlate stratigraphy, as described in detail below. Special analyses to determine diagenetic history, reservoir, and source rock quality were also undertaken. These include detailed petrographic studies of drill cuttings and conventional and sidewall core samples; fluid-inclusion and stable isotope analyses,

isotopic analyses of gas samples, and geochemical analyses of hydrocarbon and potential source intervals, from wells that are no longer confidential, will be described in detail below.

LITHOFACIES AND DEPOSITIONAL ENVIRONMENTS Deposition on the Jurassic continental margin of Nova Scotia comprised a mixed carbonate-siliciclastic depositional system. The climate was tropical to subtropical (Golonka et al., 1994; Scotese, 1997). Depositional environments and their component lithofacies (belts) reflect this overall paleogeographic setting. A generalized facies model for the Abenaki Formation is shown in Figure 6. The depositional environments and component lithofacies are described

398 / Weissenberger et al.

FIGURE 2. Tectonic elements of the Scotian Shelf (modified from Welsink et al., 1989).

FIGURE 3. Regional seismic dip line across the Scotian Shelf. Note the wedge-shaped geometry of the Jurassic succession. Colored lines indicate Abenaki third-order depositional sequences from Abenaki (AB) I at base (red), to AB II (purple), AB III (brown), AB IV (orange), AB V (yellow), and AB VI/VII (black) at top. Depth is travel time in milliseconds.

related strata (right). Chronostratigraphy is from Haq et al. (1987); Abenaki biostratigraphy from van Helden (see Appendix 2).

FIGURE 4. Stratigraphy of the Scotian Shelf (left) (modified from CNSOPB, 1997), with detailed sequence and chronostratigraphy of the Abenaki Formation and

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 399

400 / Weissenberger et al.

FIGURE 5. Location map of the deep Panuke pool showing Abenaki penetrations and core control.

below, from the most landward to the most basinward. More detailed facies analysis in the literature includes Eliuk (1978), Ellis (1984), Eliuk and Levesque (1989), and Pratt and Jansa (1989). The most recent summary, with examples from deep Panuke, was presented by Wierzbicki et al. (2002). Because a correct facies model was imperative in building an accurate sequence-stratigraphic reservoir model, the exploration and development teams reviewed numerous potential analogs for deep Panuke. Outcrop studies from similar basins (e.g., Leinfelder,

1994; Wenzel and Strasser, 2001; Blomeier and Reijmer, 2002, respectively), which were most helpful, were supplemented with our own field studies in Portugal, the Swiss Jura, and Morocco. The team was also able to effectively tie formation image logs to specific lithofacies. This was done by cutting numerous rotary sidewall cores in the penetrated formations. Two complimentary methods were undertaken: cutting the cores before or after logging. The former method would pick core points from sample descriptions. This did not allow precise picking of sample points, but sample points were clearly visible on the image logs, giving absolute certainty as to which log facies was sampled. The latter method allowed examination of the entire log suite and subsequent sidewall coring of specific formation image log (Formation MicroImager; FMI) facies. Overall, more than 200 sidewall core samples were taken in wells with FMI log suites. Figure 7 shows the interpreted paleogeography of the Scotian Shelf during the Late Jurassic. The distribution of the facies belts is by definition generalized, because, as discussed below, facies distribution changed with changing stands of sea level. That said, a carbonate shelf margin to shelf interior coexisted with siliciclastic inner shelf deposits. Siliciclastics were also passed into the basin at points along the platform margin, notably the large Sable delta (Mic Mac Formation) in the northeast part of the study area. The shelf was generally carbonate dominated to the southwest and more siliciclastic rich to the northeast of the delta. A brief description of the depositional environments and the constituent lithofacies is given below. Estimates for the relative bathymetries of the

FIGURE 6. Facies model for the Abenaki Formation with basic depositional environments (modified from Weissenberger et al., 2000).

FIGURE 7. Paleogeography of the upper Abenaki Formation, modified from Wierzbicki and Harland (2004).

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 401

402 / Weissenberger et al.

FIGURE 8. Representative Abenaki lithofacies; (A) burrowed, fine- to medium-grained sandstone with a micritic matrix (reworked lowstand shelf; I-100 8295.5 ft [2528.4 m]); (B) oncolite packstone (lagoon proximal to margin; K-62, 3384.2 m [11,103.01 ft]); (C) oolitic grainstone (shoal margin; B-13, 2514 m [8248 ft]; coin is 18 mm [0.8 in.] in diameter).

lithofacies are derived from their position in shoalingupward successions; facies identification in wells that are tied to seismic reflectors traceable from platform to basin; and from the literature cited above.

interior can also have a significant siliciclastic component. These are interpreted as deeper lagoon deposits, representing as much as 20 m (66 ft) bathymetry (the moat of Eliuk, 1978).

Siliciclastic Coastal Plain

Platform Interior and Lagoon

Light-gray, very fine to medium quartz sand with a micritic matrix is found in association with several other sediment types on the carbonate platform. These associations are with ooids and oncoids; with red and green shales and coal; and with oncolite-ooidmegalodont packstone to wackestone and cryptalgal mudstones. The sediments vary from pure siliciclastic to mixed carbonate-siliciclastic deposits depending on local conditions. The respective subenvironments of deposition are lowstand sandstones preserved in lows or karst features on the platform; shallow lagoon; and nearshore biotopes, deltas, shoreface sands, and/or estuaries. These lithologies are all interpreted to have been deposited in less than 10 m (33 ft) of water. Figure 8A shows a bioturbated sandstone interpreted as lowstand deposits on the shelf, reworked during the ensuing transgression. Fossiliferous (gastropod-pelecypod-ostracod) wackestones and dark-gray shales found in the platform

A variety of lithofacies characterize this environment. Peloidal wackestones to packstones have a minor coral, sponge, and stromatoporoid component and are commonly burrowed and stylolitic. These are interpreted to represent shallow lagoonal deposits proximal to the margin. Oncolitic facies are also common in this setting (Figure 8B) and comprise wackestones to packstones containing varying amounts of reefal or oolitic material, depending on the proximity to the margin. Argillaceous facies are observed in the platform interior, dominantly dark-brown skeletal wackestones, shales, and sands. Oncoids, gastropods, bivalves, ostracods, and ooids are the most common clasts. These argillaceous lithologies are interpreted as deeper shelf deposits (5 – 20 m [16 – 66 ft] water depth; the moat, Eliuk, 1978, p. 447), deposited during transgressions or in parts of the shelf that were periodically not filled in a given fourth- or fifth-order depositional cycle.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 403

FIGURE 9. Representative Abenaki lithofacies (continued from Figure 8); (A) coral-stromatoporoid packstone to boundstone (reefal platform margin; G-32, 12,710 ft [3874 m]); (B) skeletal packstone with grainy matrix (proximal foreslope; G-32, 11,839 ft [3608 m]); (C) thrombolitic mudstone (open marine, basin; G-32, 14,414 ft [4393 m]).

Platform Margin This environment comprises two dominant components, the skeletal or reefal (coral-stromatoporoid) and the oolitic. Despite limited core data and relatively sparse well spacing, the platform crest may still be interpreted to be dominated by oolitic grainstones (Figure 8C), whereas coral-stromatoporoid boundstones and packstones (Figure 9A) largely occur in somewhat deeper water. The reefal component comprises coral-stromatoporoid wackestones to packstones, with common chaetetids, sponges, bryozoa, pelecypods, gastropods, and crinoids. The matrix is typically light brown and grainy. These lithofacies are interpreted as high-energy deposits found anywhere from the immediate backreef through the reef crest to the immediate forereef. These would represent water depths from 0 to 10 m (33 ft). Similar lithologies can be found as much as 50 m (164 ft) down the foreslope, representing allochthonous material shed downslope. Coral-stromatoporoid boundstone, with chaetetid and algal components, forms the other major reefal lithology. Light to medium brown in color with a grainy skeletal matrix and typically displaying massive to graded bedding, its minor components include bryozoa, pelecypods, gastropods, and crinoids. These may be associated with dark-gray skeletal wackestones with the same fauna. These lithologies represent bound reef proper, either at the

margin or as backreef patch reefs, all deposited in from 2 to 10 m (6.6 to 33 ft) of water. The darker wackestones are interpreted as shelter cavity deposits because of the dominantly grainy texture, suggesting overall high-energy conditions. Where the margin is dominantly oolitic, the ooids occur in massive to cross-bedded grainstones (Figure 8C). These are light brown in color and have minor occurrences of oncoids, solitary corals, pelecypods, and gastropods. The lithology can vary to include ooid grapestone and oncoid wackestone fabrics. These lithologies are interpreted to represent high-energy shoal deposits, deposited in less than 5 m of water, alternating with reefal boundstone lithofacies at the platform margin or near-margin shoals where waves are predominantly wind generated.

Proximal Foreslope This environment is characterized by coral-spongedemosponge wackestones to packstones with a lightgray or buff to dark-gray skeletal or micritic matrix (Figure 9B). Other components include tubiphytes, lithistid sponges, stromatoporoids, chaetetids, crinoids, gastropods, brachiopods, and pelecypods. This lithofacies may be associated with medium- to darkgray, laminated siltstones and the aforementioned, coarser reef-derived material.

404 / Weissenberger et al.

Grey to brown coral-sponge-demosponge packstones to boundstones also occur in this setting. They are typically massive to poorly bedded, with an occasionally silty to sandy, mud-dominated matrix. Fauna is deeper water scleractinian corals, sponges, chaetetids, stromatoporoids, and tubiphytes. The wackestones and packstones are interpreted as coral-dominated deposits deposited in 10 –70 m (3.3–230 ft) of water. The packstones to boundstones are coral-dominated reefs. This facies is the main reefal facies seen in core and FMI images at deep Panuke. Although the Abenaki margin is relatively steep, abundant reefal development occurred there, and the platform crest is dominated by grainstones.

Distal Foreslope This environment is dominated by sponge coral and tubiphytes wackestone, with minor amounts of stromatoporoid, chaetetid, and ooids. Microbialite textures and a cyanobacterial micritic matrix are common. Dark-gray to black laminated shales and medium- to dark-gray siltstones are associated lithologies. In addition, lithistid sponge- and coralstromatoporoid packstones to boundstones occur in this setting. These are medium gray in color, with a micritic to thrombolitic matrix. Tubiphytes, Microselena (and other deep-water scleractinian corals), bryozoans, pelecypods, brachiopods, and crinoids are associated components. Light- to medium-gray skeletal wackestone and gray to black shales are associated lithologies. The wackestone and shaly lithofacies are interpreted as background sediments, in-situ sponge, and deep-water coral meadows, deposited in 30 –70 m (100 – 229 ft) of water. The boundstones are spongedominated mounds, typically forming in oxygenated waters, 30 –100 m (100 – 330 ft) deep.

Basin Dark-gray to black shales and mudstones with laminated to thrombolitic textures (Figure 9C), laminated siltstones, and marls constitute the dominant insitu sediment. These are interpreted to represent deposition in 100 – 200 m (330 – 660 ft) of water. Sponge material, pelagic foraminifera, radiolaria, and ammonites are the most common fossils. Hexatinellid sponge-tubiphytes boundstones to packstones, with associated calcareous shales, are also observed in this setting. Stromatoporoid and coral wackestones to packstones, with associated sponges,

demosponges, crinoids oolites, and oncolites, also occur. The boundstones are interpreted as hexatinellid sponge-microbialite reefs, the shales, wackestones, and packstones as interreef material deposited in greater than 100 m (330 ft) water depth. The stromatoporoidcoral deposits are shelf derived (turbidites) occurring in depths from 10 to more than 100 m (33 to more than 330 ft).

Discussion There has been disagreement over whether the Abenaki Formation represents a dominantly reef- or ramp-style depositional system (Eliuk, 1978; versus Ellis, 1984; Ellis et al., 1985, 1990, respectively). Our new drilling, including new core data and FMI logs obtained in the deep Panuke program, supports a modified reefal model. Wells such as M-79, which essentially penetrate the entire Abenaki section, show thick aggradational packages of coral-stromatoporoid rudstone to boundstone as stacked third-order sequences evidenced on FMI logs and in sidewall cores. These boundstones and rudstones contain significant amounts of micrite in the matrix (which has been recrystallized; J. Dravis, 2001, personal communication), suggesting that they were not subjected to continuous wave action; i.e., they were likely deposited below fair-weather-wave base. By contrast, the Abenaki (AB) I (Scatarie Member) has a lithofacies composition and depositional profile observed on seismic, suggesting more of a ramp, or open platform setting, with no well-developed reef margin. Eliuk (1978) interpreted the AB I to have more of a rimmed shelf depositional style south of the G-32 well. Seismic profiles (as discussed below) support the geological evidence. Dramatic platform-to-basin relief is developed from the AB II upward, with steep depositional dips, to a maximum of approximately 508, which clearly indicates a steep reef profile. The uppermost sequence (AB VII) has a gentler depositional profile, suggesting it is less reefal. Limited sidewall core and FMI images, along with drill-cutting examination, show a deep-water sponge, mounddominated argillaceous environment. Further, our observations of facies relationships elsewhere, particularly of the Kimmeridgian Ota reef in Portugal, suggest that the ooid shoal and reef environments can be laterally and vertically adjacent, as is seen in the Abenaki Formation. This, as well as the presence of coarse winnowed matrix in some coral-stromatoporoid

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 405

FIGURE 10. Comparison of formation image log (FMI) of a large phaceloid coral (left) from M-79 (3865 – 3866 m; 12,680 – 12,683 ft) and core of the same from L-97 (3417 m [11,210 ft]; coin is 18 mm [0.8 in.] in diameter). The coral is in growth position above (and overlain by) interpreted stromatoporoid-coral packstone to boundstone.

boundstone intervals, suggests that the reefs did periodically grow above fair-weather-wave base, forming high-energy patch reefs in bathymetries similar to the ooid shoals. They may have formed an armored reef front at the platform edge but no well has yet tested this zone. Near the Abenaki platform margin, southwest of the Sable delta, siliciclastics are found only as relatively thin, isolated sands. However, core from the lower Abenaki (sequence I; Scatarie Member) from the M-79 well has well-rounded gravel- to cobblesize metasediments in a lime mud matrix, with abundant marine fauna such as crinoids and bivalves. This suggests that a much more proximal siliciclastic source existed during lower Abenaki deposition, as no such coarse siliciclastics are known from any other cores, nor cuttings examined in this study. Eliuk (1978, p. 435) interpreted the Scatarie to be absent (because of nondeposition) from several basin ridges penetrated by at least two wells. These may have been the source for the coarse siliciclastics cored in M-79. As discussed above, FMI images were successfully matched to lithofacies, particularly through the use of sidewall coring. Interpretation of some images was relatively straightforward. Figure 10 shows a cored interval with a large phaceloid coral in growth position (foreslope environment) from the L-97 well. It also

shows an FMI image from M-79 with a similar coral, roughly 1 m (3.3 ft) in height, also in growth position, in a succession of nodular foreslope mud to wackestones. Figure 11 compares an FMI image of a vuggy reefal dolomite from M-79 to core from the vuggy upper Abenaki reservoir in the H-08 well. The textural similarities of leached and dolomitized decimeter-size vuggy rock to the unaltered reefal boundstone are clear.

SEQUENCE STRATIGRAPHY The predrill sequence-stratigraphic framework for deep Panuke primarily reflects drill-cutting descriptions and geophysical log interpretations because of the paucity of conventional core in the central Scotian Shelf. Gamma-ray logs commonly showed a higher response in muddy lithologies, whereas conversely, clean gamma readings indicated higher energy, winnowed sediments. Higher gamma response was therefore indicative of either muddy foreslope or platform interior deposits. Drill cuttings or sidewall cores were typically used to distinguish between these environments. Three basic platformal environments were distinguished using older cuttings descriptions, i.e., basic descriptions of lithology and fossil content: dark limestone representing the deep-water and foreslope; light-colored limestones representing shallow-water carbonates (typically reef and shoal); and calcareous sandstones and sandy and silty carbonates representing reworked siliciclastic coastal-plain environments. Alternation of these environments allowed

406 / Weissenberger et al.

FIGURE 11. Comparison of formation image log (FMI) of a vuggy reefal boundstone from the M-79 well (left) to core of the same lithofacies from PI-1A (4029.4 m; 13,219.8 ft). the identification of the basic transgressive and regressive trends in any given well. On the dominantly carbonate part of the shelf, southwest of the Sable delta, siliciclastics were found to be concentrated in fairly thin, discrete beds. Conversely, in the region approaching Sable and farther northeast on the Banquereau Bank, the dominantly siliciclastic section was broken by only a few carbonate intervals. Based on the relative landward position of the siliciclastic and the seaward position of the carbonate environments in the depositional model, the dominance of siliciclastics or carbonates was interpreted to represent relative falling and rising sea level, respectively.

Stratigraphic Architecture of the Second-order Sequence A schematic illustration of the depositional sequence on the Scotian Shelf near deep Panuke is shown in Figure 12. The gross stratigraphic pattern shows the Abenaki platform initiated by a relative transgression above the widespread Mohican Formation siliciclastic shelf. Following relative aggradation

in the deep Panuke area, carbonate deposition became more aerially restricted and replaced by argillaceous foreslope and basinal deposits (e.g., at the G-32 well), suggesting that the Abenaki shelf was initially terminated by transgression before it was inundated by Lower Cretaceous siliciclastics. This stratigraphy is herein interpreted as a second-order depositional sequence. Jansa (1993) discussed the possible causes for the termination of the Abenaki platform, favoring contamination by siliciclastics. This cannot be discounted, but, as discussed below, the carbonate platform coexisted with siliciclastics throughout the Jurassic. This suggests that the ultimate termination of upper Abenaki deposition in the deep Panuke area likely had other causes, such as relative sea level change. Higher order sequences, herein interpreted as thirdorder depositional sequences, were initially defined by the presence of thin siliciclastic sands and silts at the sequence boundaries and corresponding deepenings in the middle of the sequences, represented by deeper water carbonates. The latter were interpreted as the transgressive maximums of the sequences. These surfaces were initially correlated in the area illustrated

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 407

FIGURE 12. Schematic block diagram of the Jurassic carbonate platform, deep Panuke area. Maximum flooding surface of the AB V indicated by an asterisk. Time-equivalent basinal (Verrill Canyon Formation) sediments are not shown. For horizontal scale, refer to Figure 7.

in Figure 12 and later extended along the entire Scotian Shelf. In the region north of the Cohasset L-97 well, the upper Abenaki is dominated by siliciclastics. There, the sequences were correlated primarily using the flooding surfaces, represented by extensive correlatable limestone beds in the siliciclastic succession. New well data largely supported the sequences defined before PP-3C was drilled. Flooding surfaces (and sequence boundaries) were better constrained by the better log data in the new wells tied to sidewall cores, as well as detailed drill-cuttings descriptions. Interpretation of these new data led to minor adjustments of the sequence picks in the regional data set and the definition of an additional third-order sequence (at the top of the Abenaki) in the deep Panuke area, the AB VII.

Third-order Sequences The nature of third-order sequences in the defined Abenaki second-order depositional sequence was influenced by their position in the larger sequence. The

AB I, which generally corresponds to the Scatarie Member, is a broad carbonate ramp. It typically has three recognizable shoaling-upward cycles, comprising outer-ramp fossil wackestones to packstones and inner-ramp oolitic and oncolitic facies; identifiable by cleaning-upward profiles on the gamma-ray log. These shoaling patterns are also apparent in the cuttings descriptions, as in the M-79 well. In the deep Panuke area, the AB I is roughly 140 m (459 ft) thick. The AB I is interpreted as the early transgressive part of the second-order sequence. The base of the AB I should likely be placed in the upper part of the Mohican siliciclastics, but lack of deep well penetrations precludes the identification of this surface. Similarly, part of the Mohican siliciclastic shelf would correspond to the lowstand of the second-order sequence, but (again) the lack of deep penetrations and reduced seismic resolution at depth hampers the interpretation of these strata. The AB II comprises two readily identifiable components, the regionally extensive Misaine Formation shale and the progradational reefal platform

408 / Weissenberger et al.

above. The base of the sequence is taken at the top of the AB I carbonate ramp. If a lowstand systems tract is present, it may consist of shelf-interior siliciclastics bypassed into the basin. These have only been provisionally identified on seismic data and not penetrated by wells. The aforementioned, recognizable components of the AB II are interpreted as the transgressive and highstand parts of the sequence, respectively. The Misaine represents a major transgression, with basinal shale mapped seismically at least 15 km (9 mi) landward of the underlying AB I shelf margin (G. Syhlonyk, 2005, personal communication). It is on the order of 100 – 150 m (330 – 492 ft) thick near the shelf margin. The furthest landward extent of the Misaine is also interpreted as the maximum flooding surface (MFS) of the Abenaki second-order sequence. This transgression is recognized along the North Atlantic continental margins (Jansa and Wiedmann, 1982) to the southeast margin of Tethys (Gradstein et al., 1999). The AB II highstand comprises approximately 220 m (721 ft) of relatively clean carbonates that prograde over the Misaine. Seismic data shows downlap of intrahighstand reflectors onto the Misaine (discussed and illustrated below in Figure 13). Cores and drill cuttings show the AB II to consist of skeletal wackestones immediately above the Misaine, with a gamma-ray reading transitional with the underlying shales. The remainder consists of oolitic packstones to grainstones or reefal boundstones. The AB II platform margin is typically set back from that of the AB I. The AB III is the thickest of the interpreted thirdorder sequences. It ranges from 400 to 450 m (1300 to 1470 ft) thick in the deep Panuke area. Formation MicroImager, sidewall core, and cuttings data indicate that it consists of coral-stromatoporoid packstones to boundstones, i.e., reefal strata. This composition, reflected also in a moderate gamma-ray signature, is fairly consistent through the sequence, suggesting an aggradational stacking pattern. The northeast part of the schematic (Figure 12) shows the first appearance of abundant siliciclastics in the lower part of the AB IV, between the L-30 and L-57 wells. This is interpreted as the lowstand and early transgressive part of the sequence. The MFS is interpreted to be represented by a distinctive clean carbonate that can be correlated from L-97 to L-30. Southwest of L-97, the AB IV is a platformal carbonate as much as 150 m (492 ft) thick with a slightly greater gamma response overall than the AB III. A higher gamma response in the middle to upper part of the se-

quence is observed in three of the four penetrations of the sequence in the pool itself. This is interpreted to reflect the same deepening observed in the siliciclastic realm to the northeast. The AB V has a moderate gamma-ray signature, similar to the AB III and IV. It is as much as 150 m (492 ft) thick in the deep Panuke area. The lithofacies at the base of the sequence are muddier and have a higher gamma response, with a forereef macrofauna. This is interpreted as the transgressive part of the sequence. A lowstand systems tract for the sequence is interpreted only from seismic data. Basinal mudstones and shales attributed to the AB V were penetrated in the M-88 well, but these are interpreted as transgressive and highstand deposits. The upper AB V consists of coral stromatoporoid packstones to boundstones, representing a reef margin environment. Figure 12 again shows a correlatable carbonate bed in the lower and middle part of the sequence extending northeast of L-97. This is interpreted as representing the MFS of the AB V, which temporarily interrupted siliciclastic deposition. The AB VI consists of 150 – 200 m (492 – 660 ft) of calcareous mudstones and shales with a minor siliciclastic component. They display a moderate to high gamma-ray response in the deep Panuke area. These are interpreted to represent foreslope and basinal deposits derived from a backstepped carbonate platform. Skeletal-peloidal packstones encountered in the upper AB VI in the F-09 well are interpreted to be part of this retreated carbonate platform. Approximately 150 m (492 ft) of basinal calcareous shale are interpreted as equivalent to the AB VI in the M-88 well. Wells penetrating the AB VI on the platform northeast of L-97 encountered an entirely siliciclastic section. The AB VII is a thin (roughly 50-m [164-ft]-thick) interval of shales and mudstones with a high gammaray response. Where carbonate-rich intervals have been encountered in this interval (e.g., G-32 and D-42) they consist of sponge-rich, argillaceous limestones. These are interpreted to represent deeper foreslope to basinal environments (the Artimon Member of Eliuk, 1978). A basinally restricted, 20-m (66-ft) interval of sandstone encountered in M-88 is interpreted as the lowstand systems tract of the AB VII. Regional correlations undertaken in this study suggest the AB VII at deep Panuke correlates with skeletal and oolitic grainstone shoals drilled in the uppermost Abenaki in the southwest part of the Scotian Shelf, e.g., in the Acadia K-62 and Bonnet P-23 wells. These have been termed the Roseway unit (Wade, 1977) and are of Lower Cretaceous Neocomian age.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 409

FIGURE 13. Two dip-oriented seismic lines extracted from 3-D data in the deep Panuke area. (A) Line 420, approximately half-way between H-08 and the older well G-32. It shows a strong peak (green) picked as the top of the Abenaki I; the trough above (blue) represents the Misaine Formation shale. It shows an offlapping Abenaki II succession, the top-lap geometry suggesting a sequence boundary at the top of the unit. (B) On trace 3239, the Abenaki III displays an offlapping geometry. A fairly strong trough-peak defines the top of the Abenaki II. Top-lap geometries again support the sequence boundary at the top of the unit.

FIGURE 14. Idealized fourth-order depositional sequence, Abenaki Formation. Constituent systems tracts and lithofacies are indicated (modified from Wierzbicki and Harland, 2004).

410 / Weissenberger et al.

The AB VII also has a shalier, deep-water character northeast of L-97, but appears less calcareous than in the south, suggesting persistent proximity to siliciclastic input. An idealized fourth-order depositional sequence of the Abenaki Formation, shown in Figure 14, illustrates the approximate thickness and facies distribution. It assumes some subaerial exposure and epikarstic dissolution at the basal sequence boundary. In this model, siliciclastics were deposited on the shelf and infilled dissolution pipes during relative lowstand of sea level. Where favorable conditions existed, a lowstand reef, limited in lateral and vertical extent, would grow on the preexisting slope. During the transgressive phase, normal-marine carbonate deposition resumed on the carbonate platform. The shelf-interior siliciclastics would be reworked during the transgression and commonly preserved only in topographic lows. A varied suite of depositional environments as discussed above would develop, including relatively thick beds of basinal mudstones and shale. The highstand would develop a similar suite of facies, shifted somewhat basinward of the underlying transgressive deposits. Figure 15 is a stratigraphic cross section correlating the fourth-order parasequences and lithofacies of the AB V in the deep Panuke area. The fourth-order boundaries (AB Va–Vg) were defined during the delineation and predevelopment phase of the project (see below) based on log signature and detailed drill-cuttings descriptions (strip logs in Figure 15). Immediately evident is the dominance of reefal lithofacies in most of the sequence, except for the two wells drilled most landward, B-90 and F-09, which are dominated by oolitic shoal and lagoonal environments. The reefal intervals display well-developed cyclicity from foreslope to reef, occasionally shoaling into oolitic facies (e.g., top AB V in M-79A). The downward dip of the fourth-order boundaries from the landward toward the margin wells (e.g., PI-1B, M-79) suggests

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 411

ongoing subsidence at the margin during AB V deposition. The margin wells also show significantly more foreslope lithologies in the lower AB VI, suggesting incipient drowning of the platform margin.

Seismic Stratigraphy Predrill geological interpretation of well data established the basic sequence-stratigraphic framework, tied to seismic data prior to drilling the deep Panuke discovery. This was followed, after completion of the PP-3C well, by a regional seismic-stratigraphic study. The objective of this work was to determine whether the interpreted geological sequences could be correlated on seismic data and, further, whether seismicstratigraphic criteria confirmed their interpretation as (third-order) depositional sequences. Figure 13 illustrates two arbitrary seismic lines from 3-D seismic data in the deep Panuke area. They show several aspects of seismic geometry that substantiate the geologically interpreted sequence boundaries. Figure 13A is a line (420) from the Panuke 3-D survey. Reflectors above the Misaine are clearly inclined, basinward dipping, suggesting progradation of AB II carbonate over and downlap onto the underlying shale. This is consistent with expected seismic geometries above an MFS; the Misaine is the interpreted MFS of the second-order depositional sequence. Reflectors higher in the sequence display flat-lying top sets on the interpreted platform, a discernable change in slope and dipping foresets at the basin margin. Finally, the most basinward-dipping reflectors of the AB II do not appear to have any time-equivalent top sets on the shelf. This offlapping seismic geometry suggests an accommodation minimum at the interpreted top AB II surface. This surface was tied to the geological sequence boundary interpreted from regional well control and regionally correlated on the seismic grid. Figure 13B is a dip line (trace 3239) extracted from the Musquodoboit 3-D survey. It shows similar seismic geometries in the AB III to those seen in the AB II on line 420. Reflectors in the lower AB III can be traced from the platform interior, across the break in slope at the margin, into the basinward-dipping foreslope. Higher reflectors in the upper AB III are inclined near the margin and can be traced upward (landward) where they appear to terminate against flat reflectors of the overlying sequence. This offlapping geometry suggests a lack of accommodation on the platform at the interpreted sequence boundary capping the AB III. This surface was also

first interpreted from well data and correlated to seismic data. The two lines also show other noteworthy stratigraphic relationships. The overall aggradational nature of the carbonate platform is evident on line 420, although the break in slope of each third-order margin shifts from the AB II to IV. The AB V margin is clearly stepped back several hundreds of meters from that of the AB IV. The lower Abenaki is more difficult to interpret in trace 3239 because continued basin subsidence causes significant basinward dips in the reflectors. The AB I margin likely is somewhat more than 2 km (1.2 mi) basinward of the younger margins. A strong, steeply dipping peak develops in the foreslope, regionally traceable to the top of the AB III. We interpret this as a submarine hardground representing a depositional hiatus in the basin during the post-AB III lowstand. Basinal reflectors onlap this surface. The upper Abenaki platform margin is slightly stepped back from the AB II margin. Similar seismic geometries were observed at the other sequence boundaries first interpreted using geological (well) data. This provided some confidence that regionally correlatable chronostratigraphic units could be identified and form a reliable stratigraphic framework for exploration and development.

Discussion Regional seismic data (Figure 3) suggests that the Abenaki is a depositional wedge on the Scotian Shelf. This line suggests increasing coastal onlap in the upper Abenaki (AB V –VII). Relative sea level data that incorporate other regions (Figure 4) suggest, overall, a relatively constant onlap. Decreasing onlap in the Mic Mac and lower Missisauga formations, along with the change from carbonate to dominantly siliciclastic deposition, is the main criteria for picking a secondorder sequence boundary above the Abenaki. It is important to note that Figure 12 depicts the dominantly aggradational nature of the shelf in the deep Panuke area. The time-equivalent shales and basinal mudstones of the Verrill Canyon Formation are not shown. When the pool was discovered, potential reservoir bodies were thought to include the high-energy platform margin, pinnacle, or possibly atoll reefs, and siliciclastic sands bypassed into the basin during relative sea level lowstand. Figure 16 shows two isochron maps of the platform margin south of deep Panuke. They display several noteworthy features of the platform margin in that

412 / Weissenberger et al.

FIGURE 15. Stratigraphic cross section of the upper Abenaki Formation, deep Panuke area (wells are indicated in Figure 5). Third-order depositional sequences and constituent fourth-order sequences (parasequence sets) are indicated. Modified from Wierzbicki and Harland (2004).

part of the shelf. The isochrons at the shelf margin of the AB I are more widely spaced than at the platform margin of the AB III, suggesting a gentler, ramp-style depositional profile. The configuration of the margin in the two different sequences is quite similar. For example, the embayments just southwest of the H-08

and G-32 are prominent in both time slices, as is the westward turn of the margin at the southern end of the survey. This partly reflects the aggradational nature of the platform in this region, but also the inheritance of paleogeography with time, possibly because of the persistent influence of underlying structural elements.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 413

We interpret the position of embayments and promontories on the platform margin to largely be controlled by the configuration of the underlying (synrift) horst and graben network. The designation of the Abenaki depositional sequences as second and third order is done in broad

adherence to the scheme of Vail et al. (1977), where second-order sequences have a duration of 10 – 100 m.y. and third-order sequences are 1–10 m.y. The Abenaki sequences also have seismic-stratigraphic characteristics (as described above), supporting their designation as depositional sequences. The eustatic

414 / Weissenberger et al.

FIGURE 16. Time structure on Abenaki Formation horizons, deep Panuke, and southwest. (A) Top Abenaki sequence I time structure. Relatively widely spaced isochrons indicate ramp profile at margin. (B) Top Abenaki III time structure. Tighter isochrons near the margin indicate a steeper depositional profile. A discernable similarity exists in the position of the platform margin from the AB I to the AB III (e.g., the embayment near G-32 and south of H-08), showing an inheritance of paleogeography through the section, likely reflecting deeper structural control.

sea level curve of Haq et al. (1987; partly reproduced in Figure 4) defines six supercycles (19 third-order sequences) throughout a similar time span. Therefore, the Abenaki sequences may more prudently have been termed ‘‘composite sequences,’’ or supercycles, but their ultimate importance to the exploration and development program at deep Panuke was their utility in defining correlatable time lines across the Scotian Shelf and within the pool. The sequence-stratigraphic model postulates exposure of the platform at the third-order sequence boundaries. None of these boundaries are cored, so other data (e.g., FMI images and internal facies architecture) are used to define the third-order surfaces. Eliuk (1978) proposed periodic exposure of the platform, with the development of freshwater lenses, citing evidence of karst from the G-32 well (which he linked to dolomitization). Three-dimensional seismic slices of the carbonate platform in the deep Panuke area commonly show circular depressions, 100 – 700 m (330 –2300 ft) in diameter, which have been interpreted as karst

sinkholes (G. Syhlonyk, 2005, personal communication). We have examined the similar aged Candeiros formation carbonates in central Portugal. These have well-developed karst sinkholes, tens of meters deep and wide, as well as bedding-parallel dissolution (paleowater table?). These data support the model of periodic subaerial exposure of the Abenaki platform. The nature of the alternation between siliciclasticand carbonate-dominated deposition on the Scotian Shelf is seen in two cored wells to the northeast of deep Panuke: Peskowesk A-99 and Citadel H-52 (Figure 17). The former displays a change from dominantly carbonate to siliciclastic deposition. The base of the core consists of a dark-gray oolitic packstone with crinoids and shell hash, representing a middle foreslope setting. Above are fissile calcareous shales becoming nodular and fossiliferous toward the top (crinoids and corals), representing a transition from basin to lower foreslope environments. The top of the latter unit is extensively burrowed and mineralized, as is the top of the overlying thin crinoid-bivalve

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 415

FIGURE 17. Mixed carbonate-siliciclastic core from two wells northeast of deep Panuke. Cores show switching from one dominant sediment type to another after major depositional hiati.

416 / Weissenberger et al.

wackestone. This succession is interpreted as transgressive, with the borrowed and bored contacts at the tops of the two shoaling-upward cycles representing significant depositional hiati at two marine flooding surfaces. The base of the overlying succession is a black, fissile, occasionally sideritic shale. This is overlain by a thin, fine-grained sandstone, followed by shales, silts, and finally, fine- to medium-grained sand. This is interpreted as a basinal to lower shoreface-offshore transition. This is sharply overlain by planar to tabular, parallel-laminated, medium- to coarse-grained sandstone with common Skolithos, interpreted as upper shoreface deposits above a sequence boundary. The succession in Citadel H-52 begins with siliciclastics, shales, and silts passing upward into fine- to medium-grained, parallel-laminated sandstone, with a burrowed upper surface. Above is a dark-gray to black, very fine silt to sandstone, with Planolites and Terebellina. The top of the unit is irregular and mineralized. This succession is interpreted as an offshore transition to middle (and finally, lower) shoreface setting with a major submarine hardground at the top. Above the hardground is a fossil lag with abraded coral fragments, and bivalve fragments, overlain by fissile to nodular calcareous shale. This is, in turn, overlain by a thick branching coral-abraded stromatoporoid wackestone, with crinoids, bivalves, and sponges, capped by a massive, medium gray-brown, peloidal wackestone to packstone, with crinoids and bivalve fragments. The youngest interval is interpreted to represent basinal to middle-foreslope environments, initiated by an initial marine flooding event. These two cores demonstrate that carbonate and siliciclastic environments coexisted on the Abenaki shelf. However, although there are certainly mixed lithologies (e.g., ooids with quartz nuclei, calcareous sandstones), there is commonly a switching of dominant sediment type at a major event, such as a depositional hiatus, significant transgression, or drop in sea level. Prather (1991) describes in detail similar mixed carbonate-siliciclastic environments from the Jurassic of the Baltimore Canyon. Such a surface is interpreted at a fourth-order cycle (likely parasequence set, sensu Van Wagoner et al., 1988) from the M-79 well, illustrated in Figure 18. An irregular surface, observed at the top of a coralstromatoporoid packstone, is interpreted to represent a platform margin environment. This is abruptly overlain by 25 cm (10 in.) of fine-grained quartz sandstone, interpreted to have been deposited during exposure of the platform at a sea level lowstand and

reworked during the subsequent transgression. This is overlain by 30 cm (12 in.) of oolitic grainstone coralstromatoporoid-sponge packstone, reflecting transgression above the boundary.

DIAGENESIS Understanding diagenetic processes is critical for the correct evaluation of the deep Panuke reservoir. Examination of drill cuttings from PP-3C, immediately above the cavernous porosity zone, revealed the first Abenaki dolomite in coral-stromatoporoid packstones, suggesting that dolomitization may have been involved in vuggy porosity creation. However, logs from PP-3C indicated that the cavernous porosity was mostly in limestone. Detailed petrography, isotope, and fluid-inclusion analyses were conducted on the subsequent wells and presented by R. A. Wierzbicki, J. J. Dravis, I. Al-Aasm, and N. J. Harland (2005, personal communication). This work and some additional data are discussed below. Early diagenesis is characterized by grain suturing in grainstones and grainy packstones, suggesting little cementation before compaction. Stylolites and wispy microstylolites are common. However, abundant secondary limestone porosity is preserved in the Abenaki at relatively deep burial depths. This porosity was observed to cut stylolites. The porosity is preserved along stylolites, and stable skeletal grains (e.g., echinoids, calcispheres, corals, stromatoporoids, etc.) are partially dissolved, and calcite cement in healed fractures does not extend into adjacent secondary microporosity. These fabrics suggest the secondary limestone porosity formed after pressure solution and at least some of the fracturing. The observed porosity was dominantly vuggy and moldic, the latter typically reflecting dissolution of micritic peloids or micritized skeletal grains. Dolomites are dominantly fine to medium crystalline, nonferroan, and subhedral to anhedral in shape. This dolomite appears to be mostly fabric selective, replacing grainy limestone textures (packstonesgrainstones) proximal to the platform margin. Saddle dolomite was observed as a replacive mineral and pore-filling cement. The matrix dolomites overlie pressure-solution seams and massively replace relict sutured grain fabrics. Porosity in the dolomite section is vuggy and intercrystalline. Vuggy porosity crosscuts pressure solution seams in dolomites, with many pores appearing to be molds or partial molds. Saddle dolomites and dolomitized echinoids were also dissolved. This suggests that the pores may have formed

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 417

FIGURE 18. Formation image log (FMI) of a fourth-order sequence boundary from the Abenaki V of the M-79 well. Irregular surface capping grainy coral-stromatoporoid packstone (3884.9 m; 12,745.7 ft) is overlain by sandstone lag, oolitic grainstone, and finally, argillaceous coral-stromatoporoid-sponge packstone.

as a result of the dissolution of dolomitized relict grains or cement, instead of relict limestone. Fracturing is common in the Abenaki, normal vertical burial and tectonic fractures (including micro-

fractures and swarms of short macrofractures), whereas horizontal natural and crosscutting fractures suggest active wrench faulting. Vertical fractures commonly cut stylolites, suggesting their burial origin.

418 / Weissenberger et al.

FIGURE 19. Thin-section photomicrographs of Abenaki Formation reservoir, porosity, and diagenetic fabrics. (A) Secondary vuggy porosity associated with emplacement of replacement dolomite, as commonly observed in the Abenaki dolomitic limestones (H-08; 3448.6 m [11,314.3 ft]; plane polarized light, 5.5 mm [0.2 in.] field of view); (B) Well-developed micromoldic porosity in ooids and peloids, interpreted to result from deep burial conditions (F-09, 3520 m [11,548 ft]; diffused plane polarized light, 3.0 mm [0.12 in.] field of view); (C) Partial dissolution of saddle dolomite (replacing recrystallized limestone, with partial occlusion of porosity in saddles by nonferroan calcite (arrows). Such partial dissolution of dolomites is common in Abenaki dolomitic limestones (H-08, 3448.6 m [11,314.3 ft]; plane polarized light, 3.0 mm [0.12 in.] field of view); (D) A curved (saddle) dolomite rhomb adjacent to a pinpoint vug (PI-1A, 4029.5 m [13,220.1 ft], 3.5-mm [0.13-in.]-wide field of view). Original descriptions by J. Dravis.

Dissolution of dolomite crystals along fracture planes suggests that the fractures delivered diagenetic fluids. Well tests from deep Panuke suggest an effective fracture network, with fractured dolomites in the AB V as much as 1000 m (3300 ft) wide. Figure 19 illustrates some of the key diagenetic textures of the Abenaki in a series of photomicrographs. The samples show evidence for secondary porosity development related to partial dolomitization, small moldic porosity in limestones, and saddle dolomites (19A, B, and C/D, respectively). Geochemical analyses (R. A. Wierzbicki, J. J. Dravis, I. Al-Aasm, and N. J. Harland, 2005, personal communication) of the dolomites revealed carbon isotopes, which were not depleted, and oxygen isotopes that were relatively highly depleted. This suggests a lack of organic influence on dolomitization and recrystalli-

zation and cementation at depth caused by higher burial temperatures. Matrix dolomites were enriched in Sr87, whereas saddle dolomites were nonradiogenic, implying mixing of a radiogenic (juvenile) and nonradiogenic (basinal) source. Fluid inclusions in calcite cements showed homogenization temperatures ranging from 60 to 1818C, with salinities of less than 0 to 13.9 wt.% NaCl, i.e., saline to nonsaline fluids acting in a warm to hot regime. Dolomite cement (saddles and others) inclusions reveal temperatures from 85 to 1478C, with salinities from 7.6 to 12.0 wt.% NaCl. This suggests that warm to hot, moderately saline waters were involved in dolomitization. Evidence for faulting during diagenesis includes vertical stylolites, natural horizontal fractures, and twinned calcite. Helium is also present (0.02–0.03%) in the deep Panuke gas, and minor sphalerite is observed

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 419

FIGURE 20. Map of limestone and dolomite porosity trends in the central part of the deep Panuke field. Interpretation made from amplitudes extracted from the 3-D seismic volume.

in the reservoir. As mentioned above, the presence of juvenile fluids is suggested by the strontium isotope data. Vitrinite reflectance data from the M-79 well also showed a marked increase from 0.8 to 1.10 at about 4260-m (13,976-ft) drill depth, suggesting a thermal anomaly. Regionally, evidence for the presence of hot, fault-borne fluids includes zebra dolomite textures in the core of the Iroquois Formation from the P-23 well. These data suggest that deep burial and hydrothermal fluids were responsible for limestone and dolomite porosity at deep Panuke. High-porosity vuggy limestones in the PP-3C and H-08 wells occur as linear trends behind the platform margin, apparently influenced by wrench faulting. Vuggy dolomites occur as curvilinear trends subparallel to the platform margin (Figure 20). Figure 21 shows the stratigraphic distribution of these same porous limestone and dolomite intervals.

PETROLEUM GEOLOGY The PP-3C discovery well was drilled on the basis of the aforementioned predrill stratigraphic model; the location picked from a 1990 3-D seismic survey. An arbitrary dip line from that survey, near the discovery location, is shown in Figure 22A, with a schematic interpretation in Figure 22B. The apparent topography in the basement is interpreted as a basement horst and associated half graben to the west, the latter filled by synrift deposits. These are overlain by Mohican Formation siliciclastics. Three basic seismic facies are discernable from the data: stratified platforminterior (lagoon), chaotic reflectors at the platform margin, and basinward-dipping reflectors in the foreslope. The chaotic margin facies is between 500 and 1200 m (1640 and 3900 ft) wide in a dip sense. Seismic modeling indicated that some of the reflectors in the

FIGURE 21. Cross section of the deep Panuke wells, showing main occurrences of vuggy limestone and dolomite porosity. Color legend: light blue is porous limestone; purple is porous dolomite; dark blue is tight limestone; orange stippling reflects no significant vuggy porosity. Diagram is modified from Wierzbicki and Harland (2004).

420 / Weissenberger et al.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 421

FIGURE 22. Arbitrary dip line from the Panuke 3-D seismic survey (A) and schematic interpretation of the same, illustrating the end of the Abenaki sequence V (B). Interpretation after G. Syhlonyk.

422 / Weissenberger et al.

margin facies represent a porosity anomaly. This is corroborated by a velocity sag in reflectors below the margin, representing the effect of slower velocity through the overlying porosity. Bright reflectors in the adjacent foreslope were (later) interpreted as sands, which had bypassed the platform during sea level lowstands.

Exploration Drilling Phase The PP-3C well was intended to test such a velocity anomaly. Beneath the Cretaceous (Mic Mac–Missisauga Formation) siliciclastics, the wellencountered gas shows in calcareous shales and mudstones of the upper Abenaki. Upon penetrating less than 20 m (66 ft) of clean limestone, a lost circulation zone occurred at 3934 m (12,906 ft) drill depth. Although only 0.1 m (0.3 ft) of porosity had been drilled, the formation continued to take drilling fluid, including two plugs of lost circulation material (LCM). Gas and condensate were flared. Introduction of a third LCM plug sealed off the formation, and the well was suspended. Given the likelihood of penetrating further cavernous porosity, it was decided to drill out using the annular velocity control (AVC) drilling method, which had not previously been attempted on the Scotian Shelf. This entailed pumping sufficient sea water into the wellbore while drilling to control the formation. No drill cuttings can be recovered using this method. Once drilling recommenced, several bit drops of decimeters to more than 1 m (3.3 ft) were encountered, suggesting further significant porosity. More than 200 m (660 ft) of Abenaki carbonate was drilled in this manner, to a final total depth of 4163 m (13,658 ft). Logging revealed a porosity zone 44 m (144 ft) thick in the AB V, which was interpreted from logs to represent predominantly vuggy dolomitic limestone porosity. As mentioned above, dolomite had been observed in the drill cuttings immediately above the lost circulation zone (a coral-stromatoporoid packstone; Weissenberger et al., 2000). The variability of the porosity seen on logs (Figure 23) was interpreted to reflect a complex pore system, likely including vugs, caverns, and matrix. Testing yielded a maximum flow rate of 69 mmcf/ day, limited by the equipment employed. Wet gas was produced, with 600 ppm of hydrogen sulfide (H2S). Also observed was a relatively slow buildup that may reflect formation damage and possible reservoir phase changes caused by the introduction of

FIGURE 23. Well log of the Panuke PP-3C discovery well showing cavernous porosity in the Abenaki V.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 423

Table 1. Abenaki Formation penetrations, deep Panuke.* Well Name and Number

Spud

Panuke F-09 Panuke H-08 Panuke M-79 Panuke M-79A Panuke PI1A-J99 Panuke PI1B-J99 Panuke PP3C J-99 Margaree F-70 Marcoh D-41 Queensland M-88

August 23, 2000 May 24, 2000 July 10, 2000 October 11, 2000 August 27, 1999 November 11, 1999 July 17, 1998 May 21, 2003 August 28, 2003 December 14, 2001

Top AB V Total Vertical Depth, Subsea

Total Depth (Depth and Formation) Measured Depth

3255.1 m 3815 m Abenaki 4 3348.5 m 3682 m Abenaki 4 3351.5 m 4597 m Abenaki 1 3331.2 m 3934.2 m Abenaki 5 3323.5 m 4033 m Abenaki 4 3322.2 m 4046.4 m Abenaki 4 3315 m 4163 m Abenaki 4 3330 m 3677 m Abenaki 4 3306 m 3625 m Abenaki 4 3737.4 m (eq) 4443 m Abenaki 2

Net Pay

Maximum mmcf/ Test Rate day (106 m3/day)

0m Flowed gas, TSTM 104.7 m 1.55 0m No test 11.4 m TVD 1.75 1.5 m No test 13.9 m 1.49 44.6 m 1.96 70 m 1.5 74.2 m No test 0m No test

0 55 NT 62 NT 53 69 53 NT NT

*Well information summary table. Porosity cutoff for net pay calculation is 3%.

abundant LCM into the reservoir, as well as more than 300,000 bbl of seawater pumped into the formation during AVC drilling. Reservoir temperature dropped simultaneously, from 130 to 808C in the wellbore.

Delineation and Predevelopment Drilling Phase The PI-1 well tested the extent of the deep Panuke discovery to the northeast (Figure 5). It was drilled on a seismic porosity anomaly analogous to PP3-C interpreted in the AB V. The well (PI-1A) encountered significant gas shows, some dolomite, but no cavernous porosity. Instead of testing the AB V, the seismic data was reinterpreted, and a sidetrack wellbore (PI-IB) drilled 70 m (230 ft) to the northeast. PI-1B penetrated moderate, dolomitic limestone porosity in the AB V. The well was completed and tested gas at significant rates (Table 1). The dimensions of the pool to the southwest were then tested by the H-08 well. The well encountered a thick section of cavernous porosity in the AB V. Annular velocity control drilling was again necessary to control the formation. Ultimately, logs showed 104 m (341 ft) of pay at a slightly lower structural elevation than PP-3C and PI-1A (Figure 24) and tested gas (Table 1). A further delineation well at M-79 had a similar result to PI-1A (no significant AB V porosity in the initial wellbore). However, M-79A, drilled 600 m (1968 ft) to the southwest, encountered extremely productive gas pay (Figure 24). The seismic section (Figure 25) shows M-79 drilled behind the interpreted platform margin, M-79A penetrating a significant interval of porous dolomite closer to the margin. The

M-79 wellbore did encounter significant, wet porosity lower in the Abenaki. Two further wells investigated the northeast extent of the pool: F-70 is another gas well, analogous to M-79A; and D-41, which was drilled in 2004, was still confidential at the time of writing. Two other wells, F-09 and M-88, were drilled off the main AB V bank margin during the delineation phase of the project. F-09 drilled 70 m (230 ft) of interbedded carbonate muds and siliciclastics in the AB VI (Figure 24). The underlying oolitic AB V interval had some gas shows, but was too tight to be economic. The M-88 well was drilled basinward of the Abenaki platform margin. It encountered bypassed siliciclastic sands in the interpreted lowstand of the AB VII (Figure 26). A new 3-D seismic survey, 450 km2 (173 mi2) in area, was shot across the deep Panuke area in 2002. These data were used to locate the subsequent wells and to model the porosity in the pool for reserves determination and depletion planning. Predevelopment geoscience and engineering work on deep Panuke was presented in Brown et al. (2004). To summarize, shear sonic logs from eight wells, vertical seismic profile data, high-quality seismic processing of the new 3-D data set, including Post-stack Time Migration, Post-stack Depth Migration (PSDM), and Compressional + Shear Wave impedances and Lambda-Mu-Rho volumes, were used to image porosity in the pool. Seismic reservoir characterization used a supervised neural network approach with seismic attributes from PSDM and amplitudeversus-offset cubes as input nodes to analyze basin, backreef, and reefal parts of the pool. Accuracy of porosity prediction was then tested by matching well

424 / Weissenberger et al.

FIGURE 24. Structural cross section, deep Panuke pool, indicating distribution of conventional and sidewall core data. Diagram modified from Wierzbicki and Harland (2004).

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 425

FIGURE 25. Seismic dip line through the M-79 and M-79a wells, deep Panuke pool. The platform margin is aggradational from the AB II upward.

porosity to seismic data and defining a set of significant seismic attributes. The attributes were first extracted from a 3  3 trace area around each wellbore and then from the entire volume and fed into the trained neural net. Numerous field-scale porosity-permeability models were developed by integrating geological and geophysical interpretations in a geostatistical framework. These data were integrated with engineering information into a ‘‘shared earth model, containing the full 3D representation of the reservoir’’ (Brown et al., 2004, p. 3). Numerous property models, generated using seismic attributes, well log, and core data, were simulated and compared to well test data. This was an iterative process, testing the validity of each reservoir model. The five wells in the pool with production tests provide large-scale permeability data, which was

used to history match the different reservoir models, thereby reducing uncertainty, particularly in permeability estimates for the pool. Figure 27 shows one such 3-D porosity model for the deep Panuke pool, with an extracted seismic profile. A geostatistical approach was also used to estimate the likely reserve size distribution of the pool (Deutsch et al., 2003). Geological and geophysical inputs (the latter from the 2002 3-D data set) were used to realistically represent the reservoir heterogeneity.

Critical Elements The deep Panuke pool is interpreted as a combined stratigraphic-structural trap. A common gaswater contact exists at 3504 m (11,496 ft) subsea (Figure 24). Top seal is provided by argillaceous carbonates and shales in the upper Abenaki (Abenaki

426 / Weissenberger et al.

FIGURE 26. Geological dip cross section interpreting platform to basin transition between the M-79 and M-88 wells. Note bypassed quartz sands in M-88, just below 3600 m (11,811 ft).

sequence VI and VII) and Mic Mac–lower Missisauga formations. Lateral seal is provided by these same formations and shale- and mud-filled reentrants in the platform at the northeast end of the pool, as well as tight, platform interior oolitic limestones. A regional structural doming also exists along the Abenaki carbonate margin, which supports the east–west closure of the pool. The source of the gas is believed to be organic shales in the Verrill Canyon Formation (Williamson and Des Roches, 1993). The principal reservoir zones are reefal boundstones and associated sediments, having both intercrystalline and vuggy porosity in dolomites and secondary and vuggy porosity in limestones. The diagenesis responsible for this porosity was discussed above.

Discussion When small amounts of H2S were encountered in tests of the PP-3C well, it raised the immediate question of provenance and possible extent of sour gas in any further wells drilled or discoveries made on

the Abenaki platform. Two hypotheses for the origin of the gas were contemplated: biogenic production caused by the prolonged interaction of sea water with the reservoir (while suspended and AVC drilling) and migration from anhydrite-rich strata where thermochemical sulfate reduction (TSR) had occurred. Sulfur isotope analyses were undertaken on a sample of the gas and on a sample of anhydrite from the Eurydice Formation of the Mohican I-100 well (Figure 4). The analyses (I. Hutcheon, 1999, personal communication) indicate that the sulfur in the deep Panuke gas matches the isotopic signature in the Eurydice Formation anhydrite. Subsequent gas analysis of wells where there was no significant fluid influx showed the same H2S concentrations. It is likely that the observed sulfur concentrations are related to TSR processes.

CONCLUSIONS Figure 28 illustrates the tectonostratigraphic evolution of the Jurassic carbonate hydrocarbon system on the Scotian Shelf. During the Early Jurassic, sea level rise allowed the deposition of the Iroquois Formation, a broad, shallow-water carbonate shelf over synrift deposits of the Argo and Eurydice formations (Figure 28A). The latter was largely restricted to topographic lows (half grabens) between tilted fault blocks. Overlying the Iroquois, a thick section of siliciclastics was deposited in the transition from the Lower to Middle Jurassic (Mohican Formation; Toarcian to Bajocian). These strata are provisionally defined as a second-order depositional sequence.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 427

FIGURE 27. Derived porosity distribution, Abenaki sequence V, deep Panuke pool. Seismic section shown is taken just north of the M-79 well. The red and black colors indicate amplitudes interpreted as the highest porosity zones in the porosity model. Interpretation by R. Tonn.

Later in the Middle Jurassic (Bajocian – Callovian; Figure 28B) the first Abenaki third-order depositional sequence (AB I) was deposited above the Mohican Formation with continued relative sea level rise. The second Abenaki sequence (AB II) comprises two parts: the basinal Misaine Formation (transgressive systems tract) and the overlying foreslope and shallow-water carbonates (highstand systems tract). The Misaine represents the MFS of both the third-order AB II and the second-order depositional sequence. Figure 28B illustrates some of the stratal geometries that are apparent on seismic data, and the possibility of reefs initiating on the drowned AB I. Beginning in the Upper Jurassic, the stratal architecture of the Abenaki platform was essentially aggradational. The coral-stromatoporoid reef margin became well established (Figure 28C). During relative lowstands in sea level, siliciclastics from the platform interior bypassed into the basin. Minor epikarst developed during these lowstands. Carbonate platform deposition continued through the lowermost Cretaceous (Berriasian; Figure 28D). The large-scale stacking pattern was aggradational to the top of the AB V in the deep Panuke area, then retrogradational

in the AB VI and VII. Northeast of deep Panuke, sedimentation was dominated by siliciclastics. From the Early to middle Cretaceous, sedimentation was dominated by siliciclastics (Figure 28E). Tectonic activity in the region began in the Aptian (Jansa and Pe-Piper, 1988). This caused the reactivation of basement faults in a strike-slip sense and the movement of hydrothermal fluids into the Abenaki carbonate platform. These fluids completed the creation of secondary limestone and dolomite porosity begun earlier in the burial history (R. A. Wierzbicki, J. J. Dravis, I. Al-Aasm, and N. J. Harland, 2005, personal communication). The Verrill Canyon source rock entered the gas-generation window in the Aptian (Williamson and Des Roches, 1993), likely filling the deep Panuke trap by the Albian. Drilling since 1998 has proven a large accumulation of natural gas at deep Panuke, extending 26 km (16 mi) along the Jurassic carbonate platform margin. EnCana has not published a revised reserves estimate since 2002, despite the drilling of several delineation wells. It remains the first and only significant hydrocarbon discovery in the Mesozoic carbonates on the entire continental shelf of eastern North America.

428 / Weissenberger et al.

FIGURE 28. Tectonostratigraphic evolution of the carbonate platform, deep Panuke area. (A) Early Jurassic; Iroquois Formation carbonate platform deposited during transgression over Triassic synrift deposits. Early dolomite formation in semiarid climate. (B) Bathonian– Callovian; maximum flooding surface of second-order depositional sequence (Misaine Formation) and progradation of Abenaki II highstand carbonate. (C) Oxfordian; exposure during a relative sea level lowstand at the base of an Abenaki third-order sequences (e.g., AB IV) yields meteoric karst and cross-bank transport (and foreslope deposition) of siliciclastics. (D) Kimmeridgian–Berriasian; relative transgression in Abenaki VI and VII sequences. Platform architecture changes from dominantly aggradational to retrogradational in the deep Panuke area. (E) Valanginian – Cenomanian; Missisauga to Logan Canyon siliciclastics deposited; subsidence of forereef; regional wrench tectonism moves hot diagenetic fluids up reactivated normal faults.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 429

APPENDIX 1: NEW CORE AND SIDEWALL CORE DATA, DEEP PANUKE

Location

Core Type

PP-3C

None

PI-1A

Whole core and sidewall

Core Interval

Number of Side-wall Core

Recovery (Whole Core)

Whole core 4029.28 – 4040 m (13,219.42 – 13,254 ft)

45

1.15 m (3.77 ft)

Analysis Done

Side-wall core 3895 – 4034 m (12,778 – 13,234 ft) PI-1B

None

M-79

Whole core and sidewall

Whole core 4532.7 – 4538.7 m (14,871.06 – 14,890.7 ft)

Routine Routine

99

5 m (16 ft)

None

Side-wall core 3512.5 – 4591 m (11,523.9 – 15,062 ft) M-79A

None

H-08

Whole core

Whole core 3446 – 3460 m (11,305 – 11,351 ft)

F-09

Sidewall

Side-wall core 3264 – 3798 m (10,708 – 12,460 ft)

31

F-70

Whole core and sidewall

Whole core 3434 – 3461 m (11,266 – 11,354 ft)

45

3.2 m (10.4 ft)

Routine and special core Routine

24.65 m (80.8 ft)

Side-wall core 3350 – 3636 m (10,990 – 11,929 ft)

Routine Special core analysis

APPENDIX 2: BIOSTRATIGRAPHIC SUMMARY, M-79 WELL, DEEP PANUKE

Depth (m)

Age Assignment

Cycle

3180 (top sample) to 3220

Early Cretaceous, Neocomian

AB 7

3240 – 3360

Portlandian, anguiformis or older

AB 6

3380 – 3440

Late Kimmeridgian, fittoni or older

AB 6

3460 – 3820

Late to early Kimmeridgian, elegans or mutabilis or older

AB 5/4/3

3840 – 4020

Late Oxfordian, rozenkrantzi or older

AB 3

4040 – 4240

Middle Oxfordian, glosense to Callovian

AB 2

4260 – 4460

Middle Callovian, coronatum and jason

AB 2/Misaine

4480 – 4600 (total depth)

Callovian

AB 1 Scatarie

B. G. T. van Helden

430 / Weissenberger et al.

ACKNOWLEDGMENTS We thank EnCana Ltd. for giving permission to publish our work on deep Panuke and Husky Energy for additional support in completion of this publication. This paper would not have been written without the contribution of our many colleagues and associates who worked on various aspects of the project, and we beg the pardon of those we do not mention. J. A. W. Weissenberger thanks John Hogg and Ian DeLong for the opportunity to work with them on the project from the beginning. We must single out the contributions of our geophysicists, Garth Syhlonyk, Robert Riddy, and Rainer Tonn, who worked through the delineation and development phases, as well as engineers Tom Craig and John Slade. We must also acknowledge several colleagues who supported this project through their specialized work: Ihsan Al-Aasm (isotopes and fluid inclusions), Jeff Dravis (petrography and diagenesis), Les Eliuk and Bev Harris (drillcuttings description), Bert van Helden (micropaleontology), and Ian Hutcheon (sulfur isotopes). Thanks again to John Hogg for critically reading the manuscript and to Phil Argatoff of Penta Graphix Ltd., who prepared many of the diagrams. The authors also appreciate the helpful comments of reviewers Sherry Becker, Niall Toomey, and Mitch Harris, which greatly improved the manuscript. Please note that the reference to proven reserves in the AAPG 2006 conference abstracts is withdrawn because it was included in error.

REFERENCES CITED Blomeier, D. P. G., and J. J. G. Reijmer, 2002, Facies architecture of an Early Jurassic carbonate platform slope (Jbel Bou Dahar, High Atlas, Morocco): Journal of Sedimentary Research, v. 72, no. 4, p. 462 – 475. Brown, S., R. Riddy, R. Tonn, and R. A. Wierzbicki, 2004, The integration of geology, geophysics and reservoir engineering for field appraisal (abs.): Canadian Society of Exploration Geophysics Annual Meeting Abstracts and Program, 4 p. Deutsch, C., R. A. Wierzbicki, R. Riddy, and J. Slade, 2003, Quantification of uncertainty in gas resources of deep Panuke (abs.): AAPG Annual Meeting Program, v. 12, 6 p., CD-ROM. Eliuk, L. S., 1978, The Abenaki Formation, Nova Scotia Shelf, Canada — Depositional and diagenetic model for a Mesozoic carbonate platform: Bulletin of Canadian Petroleum Geology, v. 26, p. 424 – 514. Eliuk, L. S., and R. Levesque, 1989, Earliest Cretaceous sponge reef mound, Nova Scotia shelf (Shell Demascota G-32), in H. H. J. Geldsetzer, N. P. James, and G. E. Tebbutt, eds., Reefs, Canada and adjacent areas: Cana-

dian Society of Petroleum Geologists Memoir 13, p. 713 – 720. Eliuk, L. S., S. C. Cearley, and R. Levesque, 1986, West Atlantic Mesozoic carbonates: Comparison of Baltimore Canyon and offshore Nova Scotian basins: AAPG Bulletin v. 70, p. 586 – 587. Ellis, P. M., 1984, Upper Jurassic carbonates from the Lusitanian basin, Portugal and their subsurface counterparts in the Nova Scotian Shelf: Ph.D. thesis, Open University, England, 283 p. Ellis, P. M., P. D. Crevello, and L. S. Eliuk, 1985, Upper Jurassic and Lower Cretaceous deep-water buildups, Abenaki Formation, Nova Scotian Shelf, in P. D. Crevello and P. M. Harris, eds., Deep-water carbonates: SEPM Core Workshop 6, p. 212 – 248. Ellis, P. M., R. C. L. Willson, and R. R. Leinfelder, 1990, Controls on Upper Jurassic carbonate buildup development in the Lusitanian basin, Portugal: International Association of Sedimentologists Special Publication 9, p. 169 – 202. Given, M. M., 1977, Mesozoic and early Cenozoic geology of offshore Nova Scotia: Bulletin of Canadian Petroleum Geology, v. 25, p. 63 – 91. Golonka, J., M. I. Ross, and C. R. Scotese, 1994, Phanerozoic paleogeographic and paleoclimatic modeling maps, in A. F. Embry, B. Beauchamp, and D. J. Glass, eds., Pangaea, global environments and resources: Canadian Society of Petroleum Geologists Memoir 17, p. 1 – 47. Gradstein, F. M., M. A. Kaminski, and F. P. Agterberg, 1999, Biostratigraphy and paleoceanography of the Cretaceous seaway between Norway and Greenland, in F. M. Gradstein and G. J. van der Zwaan, eds., Ordering the fossil record; challenges in stratigraphy and paleontology: Selected papers from a symposium held in honor of the 75th birthday of Cor Drooger: Earth Science Reviews, v. 46, no. 1 – 4, p. 27 – 98. Haq, B. U., J. Hardenbol, and P. R. Vail, 1987, Chronology of fluctuating sea levels since the Triassic: Science, v. 235, p. 1156 – 1167. Harland, N. J., J. R. Hogg, R. Riddy, G. Syhlonyk, G. Uswak, J. A. W. Weissenberger, and R. A. Wierzbicki, 2002, A major gas discovery at the Panuke field, Jurassic Abenaki Formation, offshore Nova Scotia (abs.): Canadian Society of Petroleum Geologists, Annual Meeting, Calgary, p. 154. Hogg, J. R., and M. E. Enachescu, 2003, An overview of the Grand Banks and Scotian Shelf basins development and exploration, offshore Canada (abs.): AAPG International Conference and Exhibition, Barcelona, Spain, CD-ROM. Jansa, L. F., 1993, Early Cretaceous carbonate platforms of the northeastern North American margin, in J. A. T. Simo, R. W. Scott, and J. P. Masse, eds., Cretaceous carbonate platforms: AAPG Memoir 56, p. 111– 126. Jansa, L. F., and G. Pe-Piper, 1988, Middle Jurassic to Early Cretaceous igneous rocks along eastern North American continental margin: AAPG Bulletin, v. 72, no. 3, p. 347 – 366.

Sequence Stratigraphy and Petroleum Geology of the Jurassic Deep Panuke Field / 431 Jansa, L. F., and J. A. Wade, 1975, Paleogeography and sedimentation in the Mesozoic and Cenozoic, southeastern Canada, in W. J. M. van der Linden and J. A. Wade, eds., Offshore geology of eastern Canada: Geological Survey of Canada Paper 74-30, p. 51– 105. Jansa, L. F., and J. Wiedmann, 1982, Mesozoic – Cenozoic development of the eastern North American and northwest African continental margins — A comparison, in U. von Rad, K. Hinz, M. Sarnthein, and E. Seibold, eds., Geology of the northwest African continental margin: New York, Springer-Verlag, p. 215 – 269. Leinfelder, R. R., 1994, Distribution of Jurassic reef types: A mirror of structural and environmental changes during the breakup of Pangaea, in A. F. Embry, B. Beauchamp, and D. F. Glass, eds., Pangaea: Global environments and resources: Canadian Society of Petroleum Geologists Memoir 17, p. 677 – 700. McIver, N. L., 1972, Cenozoic and Mesozoic stratigraphy of the Nova Scotia Shelf: Canadian Journal of Earth Sciences, v. 9, p. 54 – 70. Poag, C. W., 1991, Rise and demise of the Bahama – Grand Bank gigaplatform, northern margin of the Jurassic proto-Atlantic seaway: Marine Geology, v. 102, p. 63 – 103. Prather, B. E., 1991, Petroleum geology of the Upper Jurassic and Lower Cretaceous, Baltimore Canyon Trough, western North Atlantic Ocean: AAPG Bulletin, v. 75, p. 258 – 277. Pratt, B. R., and L. F. Jansa, 1989, Late Jurassic shallow water reefs of offshore Nova Scotia, in H. H. J. Geldsetzer, N. P. James, and G. E. Tebbutt, eds., Reefs, Canada and adjacent areas: Canadian Society of Petroleum Geologists Memoir 13, p. 741 – 747. Scotese, C. R., 1997, Paleogeographic atlas: PALEOMAP Progress Report 90-0497, University of Texas at Arlington, 20 p. Vail, P. R., R. M. Mitchum Jr., and S. Thompson, 1977, Seismic stratigraphy and global changes in sea level: Part 4. Global cycles of relative changes of sea level, in Seismic stratigraphy — Applications to hydrocarbon exploration: AAPG Memoir 26, p. 83 – 97. Van Wagoner, J. C., H. W. Posamentier, R. M. Mitchum, P. R. Vail, J. F. Sarg, T. S. Loutit, and J. Hardenbol, 1988,

An overview of the fundamentals of sequence stratigraphy and key definitions, in C. K. Wilgus, B. S. Hastings, C. G. St. C. Kendall, H. W. Posamentier, C. A. Ross, and J. C. Van Wagoner, eds., Sea level changes: An integrated approach: SEPM Special Publication 42, p. 125 – 154. Wade, J. A., 1977, Stratigraphy of the George’s Bank basin— Interpreted from seismic correlation to the western Scotian Shelf: Canadian Journal of Earth Sciences, v. 14, p. 2274 – 2283. Weissenberger, J. A. W., N. J. Harland, J. R. Hogg, and G. Syhlonyk, 2000, Sequence stratigraphy of Mesozoic carbonates, Scotian Shelf, Canada (abs.): GeoCanada 2000 Convention ( Joint Meeting of the Canadian Geophysical Union, Canadian Society of Exploration Geophysicists, Canadian Society of Petroleum Geologists, Canadian Well Logging Society, Geological Association of Canada, and the Minerological Association of Canada) CD disk (5 p., 4 figures). Welsink, H. J., J. D. Dwyer, and R. J. Knight, 1989, Tectono-stratigraphy of the passive margin off Nova Scotia, in A. J. Tankard and H. R. Balkwill, eds., Extensional tectonics and stratigraphy of the North Atlantic margin: AAPG Memoir 46, p. 215 – 231. Wenzel, A., and A. Strasser, 2001, Sedimentology, paleoecology and high resolution sequence stratigraphy of a carbonate-siliciclastic shelf (Oxfordian, Swiss Jura Mountains): Field trip guidebook A3, International Association of Sedimentologists, Davos, Switzerland, 19 p. Wierzbicki, R. A., and N. J. Harland, 2004, Diagenetic model: deep Panuke reservoir, offshore Nova Scotia, Canada (abs.): AAPG Annual Convention Abstracts and Program, CD-ROM. Wierzbicki, R. A., N. J. Harland, and L. S. Eliuk, 2002, Deep Panuke and Demascota core from the Jurassic Abenaki Formation, Nova Scotia: Facies models, deep Panuke, Abenaki Formation (abs.): Canadian Society of Petroleum Geologists Annual Meeting Core Convention Extended Abstracts, p. 71 – 101. Williamson, M. A., and K. Des Roches, 1993, A maturation framework for Jurassic sediments in the Sable subbasin, offshore Nova Scotia: Bulletin of Canadian Petroleum Geology, v. 41, p. 244 – 257.

12

Ramon, J. C., and A. Fajardo, 2006, Sedimentology, sequence stratigraphy, and reservoir architecture of the Eocene Mirador Formation, Cupiagua field, Llanos Foothills, Colombia, in P. M. Harris and L. J. Weber, eds., Giant hydrocarbon reservoirs of the world: From rocks to reservoir characterization and modeling: AAPG Memoir 88/SEPM Special Publication, p. 433 – 469.

Sedimentology, Sequence Stratigraphy, and Reservoir Architecture of the Eocene Mirador Formation, Cupiagua Field, Llanos Foothills, Colombia Juan Carlos Ramon BP Colombia, Bogota´, Colombia

Andres Fajardo Chevron-Texaco, Bogota´, Colombia

ABSTRACT

T

he stratigraphic architecture and facies distributions in a high-resolution time-space framework define the three-dimensional (3-D) reservoir zonation of the Mirador Formation in the Cupiagua field. A high-resolution genetic sequence stratigraphy study, using more than 731 m (2400 ft) of core and 40 well logs, is integrated with petrophysical information to populate a static structural model based on the interpretation of a 312-km2 (120-mi2) 3-D seismic volume. Dynamic data (pressure, gas tracers, gas-oil ratio behavior) are integrated with the geological model to better define sandstone bodies’ lateral continuity. Production logs (production logging tools) complement petrophysical data in the definition of fluid-flow units. Three scales of stratigraphic cycles are recognized in the Mirador Formation. Short-term (high-frequency) cycles correspond to progradational-aggradational units. Six intermediate-term cycles are identified by the stacking patterns of their component short-term cycles and by the general trend of facies successions, indicating increasing or decreasing accommodation-to-sediment supply (A/S) ratios. Two long-term cycles are defined from the stacking pattern of the intermediateterm cycles and by the general trend of facies successions. The lower half of the Mirador Formation consists of coastal-plain facies tracts and is composed of channel, crevasse splay, and swamp and flood-plain facies successions. A bay facies tract occurs in the upper half of the Mirador Formation and is composed of bay-fill, bay-head delta, and channel facies successions. Copyright n2006 by The American Association of Petroleum Geologists. DOI:10.1306/1215884M883276

433

434 / Ramon and Fajardo

The lower Mirador Unit was deposited over a wide flood-plain sequence. Each intermediate-term cycle is composed of aggradational channel deposits, progradational and aggradational crevasse splay bodies, and aggradational swamp and flood-plain facies successions. The first two intermediate-scale cycles (I, II) show a seaward-stepping pattern, and then the next cycle (III) shows a landwardstepping stacking pattern. The fall-to-rise turnaround is located at the base of cycle II. The upper Mirador shows the continuous landward-stepping pattern and places prograding bay-head delta and bay-fill facies successions over the alluvialplain setting of the Lower Mirador. This upper unit consists of three onlapping cycles composed of a succession of aggradational channel deposits, progradational bay-head delta, and bay-fill deposits with a landward-stepping stacking pattern. During this cycle, deepening-up bay-fill facies successions were deposited in the area as a consequence of the increasing accommodation conditions that prevailed in the area. Finally, the Mirador is capped by restricted-marine shales of the Carbonera Formation. The Cupiagua structure is a large, east-verging, asymmetric anticlinal fold that trends north-northeast in the hanging wall of the frontal fault. Average length and width of the Cupiagua structure are 25 and 3 km (15 and 1.8 mi), respectively. The original oil in place in the Cupiagua field is estimated between 1000 and 1100 MMSTB of oil and 3000 to 4500 mmcf of gas. The Mirador Formation accounts for approximately 51% of the recoverable oil in the Cupiagua field.

INTRODUCTION The Cusiana and Cupiagua fields, discovered in 1988 and 1992, respectively, are located about 150 km (93 mi) northeast of Bogota, Colombia, South America (Figure 1). The fields lie in the foothills trend on the edge of the eastern Cordillera. The first oil production from Cusiana began in September 1992 as part of an early production scheme via the existing pipeline systems. Commerciality of Cusiana and a small central area of Cupiagua were declared in June 1993. Phase I of the development of Cusiana allowed production to increase to 185 MBOPD during 1995. A second phase of the project was sanctioned during 1995. This consisted of an expansion of the Cusiana facilities, the construction of an 800-km (497-mi) pipeline between Cusiana and the oil terminal at the Atlantic Coast (Coven ˜as) and the construction of the Cupiagua central processing facilities. This second development phase allowed production to progressively ramp up from 185 MBOPD to reach a peak throughput in the first half of 1998 of 341 MBOPD, with gas handling rates in excess of 1.4 MMSCFD. During 1997, the Cupiagua South accumulation was discovered. The Cusiana and Cupiagua field developments are now close to complete, and the fields have entered their decline phase. At mid-2005, the field has 48 wells, 36 producers, and 12 gas-injector wells (Figure 2).

The Cupiagua field is a rich, near-critical gas condensate reservoir over a very high (1800-m; 6000-ft) hydrocarbon column. The fluid composition changes slightly with depth. The condensate yield is about 250 bbl/MMSCF of gas, and the average API gravity is 408. The initial reservoir pressure was close to 6500 psia. The dew point pressure is 5350 psia. The main drive mechanism is gas reinjection and gas expansion. An efficient revaporization recovery mechanism exists. At mid-2005, the pore volume replacement is about 70%. A predicted decline of close to 30% yr1 is exacerbated by reservoir complexity, scale, and fluid issues like gas recycling. Recent optimization of the gas injection resulted in an average decline of about 23% yr1.

REGIONAL SETTING The Llanos Foothills are located between the undeformed Llanos foreland grasslands and the highelevation, highly deformed eastern Cordillera. The Llanos Foothills involves a zone of frontal deformation running northeast for hundreds of kilometers and about 20 km (12 mi) wide (Cooper et al., 1995). The foothills are limited by the Guaicaramo and the frontal-Yopal fault systems. The main outcropping structural feature is the Nunchia syncline.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 435

FIGURE 1. Location map of the Cupiagua and Cusiana fields in the Llanos Foothills. The map shows the different fields and their producing fluids.

The stratigraphic column (Figure 3) may be divided into three major depositional cycles, which, in turn, may be further divided. The first cycle covers the Paleozoic, the second begins in the Albian–Cenomanian and continues until the Paleocene (mainly Cretaceous succession), and the third extends from the Eocene to the present. The rock successions corresponding to these cycles are separated by the regionally significant sub-Cretaceous and sub-Paleocene unconformities. Early Cretaceous extension created a graben where the Cordillera Oriental mountain range is today, which was filled with as much as 3 km (1.8 mi) of Lower Cretaceous marine sediments (Dengo and Covey, 1993). The main Cretaceous transgression, which began in the Aptian, entered from the north and spread rapidly across the basin. These sediments were primarily derived from the Guyana shield to the east. Early Tertiary uplift and erosion have cut back the present edge of Cretaceous sediments well west of their original depositional limit. From west to east, progressively younger Cretaceous strata overlie progressively older Paleozoic rocks.

Cretaceous strata are dominated by shallow-marine, deltaic, and basal transgressive sandstones. Interbedded shales increase in thickness and frequency basinward. The Cretaceous cycle (Figure 3) represents a sedimentary wedge with a central shale unit (Gacheta Formation) enclosed by transgressive sands (the Une and Guadalupe formations, respectively). The Tertiary sequence began with deposition of the laterally amalgamated channel-belt systems of the Barco Formation. The overlying Los Cuervos Formation consists of aggradational flood-plain deposits and isolated singlestory channel belts. Paleocene deposits are preserved in a relatively narrow region along the foothills but have been mostly removed by subsequent erosion in most of the adjacent Llanos basin. The Eocene Mirador Formation transitionally overlies the Los Cuervos mudstones. The Carbonera Formation comprises four regressive sandy units (labeled C1, C3, C5, and C7) intercalated with four transgressive shale units (C2, C4, C6, and C8). The shale units can be correlated regionally, particularly the uppermost unit. The sandy units are interpreted to be nearshore, coastal plain, and predominantly deltaic. The Carbonera Formation, as with all previous formations, was deposited in a basin that extended to the west far beyond the present-day Llanos (Villamil, 1999). The Carbonera Formation thickens steadily westward toward the mountain front, reaching a thickness of more than 5000 ft (1500 m). The shaly Miocene Leon Formation, which overlies the Carbonera Formation, provides the first indication of the uplift of the Cordillera Oriental and the isolation of the Llanos basin in the east from the Magdalena basin in the west. The thickest Leon Formation sediments are located east of

436 / Ramon and Fajardo

FIGURE 2. Well base map of the Cupiagua field. Main structural features are shown. Producing wells are in green, and injector wells are in red.

more than 4000 m (13,000 ft). These sediments are mainly coarse-grained red beds of continental origin. The hydrocarbons were generated toward the west from the world-class source rock La Luna – Gacheta´ Formation, buried more than 6000 m (20,000 ft) below the main thick skin thrusting of the eastern Cordillera.

CUPIAGUA STRUCTURE

the present basin edge, where they reach as much as 1000 m (3300 ft) in thickness. Farther westward, a thinning trend (facies change into sandstones) is apparent before the formation reaches the mountain front. The Leon Formation consists of homogeneous shale across much of the central Llanos basin but is rich in sandstone layers along the foothills. Coarser clastics present in the west and northwest are interpreted as having been derived from the eastward reworking of Carbonera sands exposed in island ridges ancestral to the Cordillera Oriental (Cooper et al., 1995; Villamil, 1999). The Miocene –Holocene Guayabo and Necesidad formations represent very thick mollasse deposits associated to the uplifting of the eastern Cordillera. This last cycle is a thick, easterly thinning wedge of sands and shales, which reaches

The Cupiagua structure is an elongated asymmetrical anticline running northnortheast for approximately 30 km (18 mi) (Figure 4). This structure is an east-southeastpropagating hanging-block anticline associated with the east-southeast-verging frontal and core faults (Figure 5). The regional detachments for these faults are in the mudstone intervals of the Cretaceous Gacheta Formation. The Cupiagua structure lies below the Yopal fault, which separates the Nunchia syncline from the underlying structural deformation (Coral and Rathke, 1997). The west flank is 2.5 – 3 km (1.5– 1.8 mi) wide with average dip of 35–388. Locally, it can get as much as 4 km (2.5 mi) wide. The forelimb is quite variable from south to north along the structure (Figures 4, 5). In the south part of the Cupiagua structure, the forelimb is very high angle to inverted and is almost as high as the west flank (Figure 4B). To the north, the forelimb is structurally simpler, and finally, it disappears in the middle part of the field (Figure 4C). Farther north along the structure, the detachment position rises stratigraphically, and the whole structure bends strongly toward the east. Figure 6 is a time slice of the Cupiagua three-dimensional (3-D) volume showing

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 437

FIGURE 3. Generalized stratigraphic column of the Llanos Foothills. The main reservoir sandstones of the Mirador, Barco, and Guadalupe formations are shown.

438 / Ramon and Fajardo

FIGURE 4. Details of the Cupiagua structure in (A) 3-D view. Note the northeast bending in the Cupiagua structure. This change in strike is associated to a vertical feature that affects the whole sedimentary sequence and in geological maps is called the Golconda wrench. Structural cross sections across (B) the southern and (C) central-northern parts of the Cupiagua field show the change in structural style along strike (after Martinez, 2003). In the south, the structure has a high-angle to inverted imbricate. Toward the north, this imbricate disappears, and the structure becomes a simple hanging-block anticline.

the north–south trend of the syncline over the Cupiagua structure. Toward the northern portion, the syncline bends toward the northeast (Figure 6). Along the area of strike change, a lateral ramp or a later strikeslip fault is interpreted. This lateral ramp corresponds to a surface lineament called the Golconda wrench fault (Coral and Rathke, 1997). This feature affects the whole sequence from surface to the Cretaceous sequence, and the Cupiagua structure bends to the northeast and then continues to the north (Figure 4). The western limit of the structure is a west-verging series of faults. In the southern half (south of the lateral ramp) of the field, there is a major backfault that seems to cut completely the east-southeastverging system of thrusts (Figure 4B). As seen in Figure 5A, the seismic reflectors below the frontal fault are continuous, and the frontal fault looks broken in two by the west-verging backthrust. This fault is interpreted to be a backthrust of a deeper thrust (deeper than the frontal thrust) and to cut all overlying thrust faults. This backthrust seems to be genetically associated with the Golconda lateral ramp, as it dies and does not exist north from the ramp. On the contrary, in the northern part of the field (north of the lateral ramp), the western limit of the field is defined by a series of backthrusts of the frontal fault (Figure 5B). These west-verging faults are genetically related to the propagation of the Cupiagua structure toward

the east, and they formed where the frontal thrust bends (Figure 4C). Structural mapping, fault-seal analysis, well-test data, production data, and gas tracer results demonstrate that the faults in the anticline are not compartmentalizing the field. A few exceptions occur where these faults act as baffles. Associated with the Golconda lateral ramp or wrench in the central part of the structure are a series of eastto southeast-trending lineaments. These very highangle faults cause minor displacements of the Carbonera strata and seem to extend down and cut the reservoir section. Cross-line 1510 (Figure 7) running north-northeast shows several of these almost vertical faults. The Golconda lineament dipping to the north is the lateral ramp feature that causes the bend in the Cupiagua structure. South of this feature, an opposite thrust exists that breaks the Cupiagua structure. Several secondary faults occur both at the north and south but close to the lateral ramp. Seismic quality in the reservoir section is poor, and it is hard to determine if the vertical faults cut all the way down. Locally, dynamic data support the presence of some of these faults and evidence of their seal or baffle character.

RESERVOIR STRATIGRAPHY In the Cupiagua and Cusiana fields, the reservoir sandstones comprise Late Cretaceous to late Eocene shallow-marine to alluvial sandstones, sourced from

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 439

FIGURE 5. Dip seismic sections across the Cupiagua field. Cupiagua wells close to each line are shown. Inline 1200 (A) runs along the southern part of the Cupiagua structure. Note that the dominant structural feature is the Nunchia syncline. The Yopal fault separates two different structural styles of deformation. In this seismic line, it appears that the back-main west-verging thrust breaks the Cupiagua structure and the frontal thrust. This would suggest that the backfault is a younger event. This back-main thrust extends only south of the Golconda lineament. Inline 1900 (B) is located in the northern part of the Cupiagua structure. Note that the Nunchia syncline axis and the crest of the Cupiagua structure move eastward. The west-verging faults are, in this case, associated to the frontal fault.

This chapter concentrates on the sedimentology and the stratigraphy of the Mirador Formation.

FACIES, FACIES SUCCESSIONS, AND FACIES TRACTS IN THE MIRADOR FORMATION

the Guyana field to the east. The reservoirs display sheetlike packages of shoreface bodies; laterally and vertically amalgamated channel fills and overbank deposits and bay-head delta and bay-fill bodies. The Mirador Formation is the main reservoir zone and contains 51% of the hydrocarbons initially in place. Barco and Guadalupe formations have 28 and 21% hydrocarbons initially in place, respectively. Cupiagua field has moderate core coverage of the reservoirs, with six cores taken from Mirador, four from Barco, and four from Guadalupe Formation, for a total of about 1100 m (3800 ft).

Two facies tracts exist in the Mirador Formation in the Cupiagua field. The coastal-plain facies tract is located mostly in the lower half of the Mirador and is composed of channel, crevasse splay, and swamp and flood-plain facies successions. The bay facies tract occurs in the upper half of the Mirador Formation and is composed of bay-fill, bay-head delta, and channel facies successions (Figure 8). Each facies succession consists of a spectrum of facies that are arranged in regular vertical and lateral successions. These regular facies successions reflect lateral facies transitions along the depositional profile and their vertical superposition through progradation and aggradation.

440 / Ramon and Fajardo

FIGURE 6. Time slice at 2236 ms of the Cupiagua 3-D volume showing the attitude of the Carbonera strata. The axis of the Nunchia syncline runs along the left. Most strata over the time slice belong to the east flank of the syncline. Leon (yellow) and Carbonera (blue, red) markers are shown. Note the bending of the syncline and the displacement of strata along the Golconda lineament.

This section presents the results of the facies analysis and descriptions and interpretations of the facies, facies successions, and facies tracts recognized in this study. Interpretations are discussed in terms of their inferred hydrodynamic and environmental parameters in a stratigraphic context. Facies abbreviations are constructed from the dominant sedimentary structure and the texture of the rock. Initials of the dominant sedimentary struc-

ture are followed by initials of the texture of the rock. Trough cross-stratification is abbreviated as tx, ripples as rp, sandstone as Ss, mudstone as Md, and granule sandstone as gSs. For example, a facies of trough crossstratified granule sandstone is txgSs, sandstone with horizontal burrows is hbSs, and burrowed and irregular laminated mudstone is blMd. Figure 9 shows conventions of sedimentary structures, bed contacts, and other symbols that are used in figures and core description logs. Core logs for the Cupiagua A-1, C-3, and H-11 cored wells are presented in Figures 10 –12. These logs include a graphic representation of the core description, facies interpretation, inferred stratigraphic cycles, gamma-ray response, and important remarks. FIGURE 7. Strike line (crossline 1510) along the crest of the Cupiagua structure. North is to the right. The main feature in the center of the figure is the Golconda wrench (magenta fault). Several vertical faults occur associated to this wrench. North of this feature, the Cupiagua structure and the Nunchia syncline bend toward the northeast. Mirador and intraCarbonera horizons are shown. Note the change in thickness between the C5 and Mirador markers caused by the Yopal thrust fault. This fault separates the Cupiagua structure from the overriding Nunchia syncline.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 441

FIGURE 8. Gamma-ray (GR) log and interpreted stratigraphic cycles and facies tracts in the Mirador Formation of the Cupiagua A-1 well. The coastal-plain facies tract composes the lower half of the Mirador and the bay facies tract constitutes the upper half of the Mirador Formation. Intermediate-term and long-term stratigraphic cycles are shown.

Coastal-plain Facies Tract The coastal-plain facies tract occurs in the lower half of the Mirador Formation. Three facies successions were recognized: channel, crevasse splay, and swamp and flood-plain facies successions. The channel facies succession occurs at the base of the intermediate-term cycles and is composed of medium- to coarse-grained sandstones and sandstones with floating granule-size clasts. The crevasse splay facies succession overlies the interval of channel sandstones and is composed mainly of very fineto fine-grained sandstones. The swamp and flood-plain facies succession occurs associated with the crevasse splay complex and consists of massive, laminated, and burrowed mudstones intercalated with floodplain deposits and soils. The change from channel to crevasse splay to swamp facies succession defines an overall increase in accommodation space. Flood-plain deposits represent rock deposited during a period of decreased accommodation.

CHANNEL FACIES SUCCESSION The channel facies succession contains the main reservoir rock in the coastal-plain facies tract. Eight facies were identified in the channel facies succession; however, a single channel does not necessarily exhibit this complete spectrum of facies. The facies of this facies succession are pebbly conglomerate (pbCg), massive granule sandstone (mSs), granule

442 / Ramon and Fajardo

FIGURE 9. Conventions of sedimentary structures used in the core descriptions (used in the following figures).

Pebbly Conglomerate (pbCg)

sandstone with upper plane bed stratification (uppSs), trough cross-stratified granule sandstone (txgSs), trough cross-stratified sandstone (txSs), burrowed sandstone with relict trough cross-stratification (btxSs), and ripple-laminated sandstone (rpSs). The facies present in a single, aggradational channel succession are strongly dependent on temporal and spatial changes in accommodation space and base level. In general, under overall increasing accommodation conditions, there is a transition from facies pbCg, mSs, and/or uppSs into txgSs and/or txSs and then into btxSs, cvSs, and/or rpSs. A facies substitution diagram for the channel facies succession is presented in Figure 13. This diagram condenses the natural succession of facies during an overall increase in accommodation space for the channel facies succession cored in the coastal-plain facies tract of the Mirador Formation.

Facies pbCg appears very sparsely at the base of the aggradational channel succession in relatively thick beds (as much as 1.2 m [4 ft]) or in thin lag deposits (as much as 7.6 cm [3 in.]). It is overlain by facies mSs, uppSs, txgSs, or txSs. The lower contact of this facies is a scour surface, and the upper contact is transitional or sharp with the overlying facies. Bed thickness normally is between 5 cm (2 in.) and 0.6 m (2 ft); however, beds as much as 1.2 m (4 ft) occur. Coarse clasts are well rounded and generally pebble size, but some are granule size. Quartz clasts are the most common (more than 95%), and the remainder are lithic clasts. Sedimentary structures are uncommon. Figure 14A shows a photograph of this facies. In a stratigraphic context, this facies is common in relatively low accommodation conditions, where it constitutes a minor proportion of the aggradational channel facies and occurs in the thickest beds. In contrast, in high accommodation settings, beds of this facies are uncommon. Two hydrodynamic interpretations are possible. When this facies contains either normal or inverse grading, it is interpreted as deposited by a hyperconcentrated flow. When grading is absent, it is interpreted as rapidly deposited by a turbulent current in the upper flow regime.

Massive Granule Sandstone (mSs) This facies occupies the basal part of an aggradational channel succession. This facies is overlain by facies uppSs, txgSs, or txSs. The lower contact, in some cases, is a scour surface; in other cases, the contact is sharp or transitional with facies pbCg. The

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 443

FIGURE 10. Core description showing interpreted facies tracts and cycles of the Cupiagua A-1 well. The Mirador Formation is divided into six intermediate-term stratigraphic cycles and two long-term cycles. (See Figure 9 for legend.)

444 / Ramon and Fajardo

FIGURE 11. Core description showing interpreted facies tracts and cycles of the Cupiagua H-11 well. (See Figure 9 for legend.)

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 445

FIGURE 12. Core description showing interpreted facies tracts and cycles of the Cupiagua C-3 well. (See Figure 9 for legend.)

446 / Ramon and Fajardo

FIGURE 13. (A) Channel facies succession, example from Cupiagua H-11 well. Facies that compose these aggradational units define an overall increase in accommodation space. (B) Facies substitution diagram of channel facies succession. The horizontal axis represents the probability of the substitution of facies. The area occupied by each facies is directly proportional to the frequency of the facies in the channel facies succession. (See Figure 9 for legend.)

The quartz arenite is upper medium to coarse grained with a few scattered granules. Sedimentary structures are absent, but normal or inverse coarse-tail grading uncommonly occurs. In a stratigraphic context, this facies is present only in low accommodation conditions. Two hydrodynamic interpretations are considered. When normal or inverse coarsetail grading is present, this facies is interpreted as a hyperconcentrated flow deposit. When grading does not occur, it is interpreted as rapidly deposited by a turbulent current in the upper flow regime.

Granule Sandstone with Upper Plane Bed Stratification (uppSs)

upper contact normally is sharp. Bed thickness commonly ranges between 7.6 and 20 cm (3 and 8 in.), but, in exceptional cases, are as much as 0.6 m (2 ft).

This facies occurs in the lower part of the aggradational channel successions and is overlain by facies txgSs or txSs. The lower boundary is either sharp or transitional. The upper boundary is always sharp. Bed thickness commonly ranges between 0.15 and 0.6 m (0.5 and 2 ft). The quartz arenite is coarse to very coarse grained, with scattered quartz and uncommon lithic granules. The planar stratification is defined by alignment of pebble-size quartz clasts in horizontal or low-angle (less than 58) laminae.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 447

In a stratigraphic context, this facies occurs in relatively low accommodation conditions. Facies uppSs was deposited by a turbulent current in the upper flow regime.

Trough Cross-stratified Granule Sandstone (txgSs) This facies constitutes as much as 30% of the channel facies succession cored in the Mirador Formation. It is commonly overlain by facies txSs and, in very few cases, by facies btxSs. The quartz arenite is coarse to very coarse grained with scattered pebbles. Sometimes, the granule-size grains are arranged along the foresets. Thickness of trough cross-stratification sets is variable, commonly from 0.3 to 1.2 m (1 to 4 ft). Sets are commonly of constant grain size, but some sets exhibit a fining-upward trend. The angle of the foreset laminae range between 20 and 308, and the foreset laminae thickness ranges between 5 and 40 mm (0.2 and 1.5 in.). Pebbles are aligned parallel to the foreset laminae. This facies occurs in higher accommodation conditions than the previous facies. The bed set thickness changes with changes in accommodation space; as accommodation increases, bed set thickness increases. It was deposited by a turbulent current in the lower flow regime. Facies txgSs is the stratigraphic record of migrating 3-D dunes on the channel floor.

Trough Cross-stratified Sandstone (txSs) Facies txSs is the most common (as much as 60%) facies of the channel facies succession. It is overlain by facies btxSs, cvSs, and rpSs. The quartz arenite is medium to coarse grained. Fining-upward trends occur in some vertical successions of sets. Thickness of trough cross-stratification sets is variable, commonly between 0.15 and 0.6 m (0.5 and 2 ft); however, some sets are strongly amalgamated and range between 5 and 12.7 cm (2 and 5 in.) thick, whereas other sets are as much as 1.2 m (4 ft) thick. Foreset laminae are inclined between 15 and 258 and are between 3 and 30 mm (0.118 and 1.18 in.) thick. Foreset laminae commonly are locally broken by burrowing. Figure 14B is a photograph of this facies. In a stratigraphic context, this facies was deposited in higher or the same accommodation conditions as facies txgSs. The bed set thickness changes as a function of the accommodation space; as accommodation space increases, the bed set thickness increases. Facies txSs was deposited by turbulent currents in the lower flow regime. The trough cross-stratified sandstone is the stratigraphic record of migrating 3-D dunes on the channel floor.

Burrowed Sandstone with Relict Trough Cross-stratification (btxSs) This facies occupies the upper part of aggradational channel successions. This facies overlies facies txSs and substitutes with facies rpSs. Bed thickness is sometimes difficult to observe, but, in general, is less than 0.3 m (1 ft). The quartz arenite is medium to coarse grained. The most characteristic feature is the burrowing. More than 70% of the original trough crossstratification is destroyed, and relict foresets with variable angles of inclination are visible through the burrowing overprint. In some cases, the rock is completely burrowed, and some irregular laminations occur. In a stratigraphic context, this facies occurs in high accommodation conditions. When the burrow intensity is constant between two scour surfaces, this facies was deposited under low sedimentation rates. When burrow intensity decreases from top to bottom in sandstones between two scour surfaces, it is interpreted as fluctuations in rates of accumulation; periods of bed migration were interrupted by a time of nondeposition when organisms reworked the sediment.

Ripple-laminated Sandstone (rpSs) This facies is uncommon in the channel facies succession cored in the Mirador Formation. Facies rpSs, when present, appears on top of the aggradational channel successions. Bed thickness is between 5 cm (2 in.) and 0.3 m (1 ft). The quartz arenite is lower to upper medium grained. Occasionally, minor burrowing that partially destroyed the ripple laminae is typical. In a stratigraphic context, this facies occurs in high accommodation settings. It is interpreted as a waning flow cap of the hydrodynamic regime responsible for the deposition of the channel succession.

CREVASSE SPLAY FACIES SUCCESSION This facies succession was preserved during relatively higher accommodation conditions than the channel facies succession. Thin muddy beds of the swamp and flood-plain facies succession are commonly interlayered with this facies succession. Five facies are identified: sandstone with mudstone rip-up clasts (rupSs), small-scale trough cross-stratified sandstone (stxSs), ripple-laminated sandstone (rpSs), convolute sandstone (cvSs), and burrowed and irregular laminated sandstone (blSs). The crevasse splay facies succession generally does not contain good reservoir rock. For similar porosities, the crevasse splay sandstones have much lower permeability distributions as compared to channel sandstones.

448 / Ramon and Fajardo

FIGURE 14. Photos of the different facies present in the coastal-plain facies tract in the Lower Mirador in the Cupiagua field. (A) Pebbly conglomerate. Clast are well rounded and composed of white quartz and some cherts; (B) trough cross-stratified sandstone; (C) sandstone with rip-up clasts; (D) small-scale trough cross-stratified sandstone; (E) ripplelaminated sandstone; (F) convolute sandstone (3805 m [12,484 ft] measured depth); (G) burrowed, laminated sandstone; (H) laminated mudstone with starved ripples; (I) mottled mudstone; (J) rooted and irregular laminated mudstone; (K) brecciated mudstone. All cores are 10cm (4 in.) wide.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 449

FIGURE 14. (cont.).

450 / Ramon and Fajardo

Crevasse splay intervals normally occur between channel facies successions and swamp facies successions. This succession of facies tracts records increasing wetness and increasing preservation through time, beginning with a subaerially exposed surface of unconformity at the channel base. Facies successions in crevasse splay complexes constitute a series of commonly asymmetrical, shortterm, base-level fall hemicycles that contain progressively less crevasse-channel facies and progressively more distal crevasse splay facies (Figure 15). This means that initial crevasse splay progradations onto the flood-plain or ephemeral flood-plain swamps were replaced by crevasse splay progradations into wetter flood plains and permanent swamps. Thus, individual short-term stratigraphic cycles represent episodes of progradation, but each episode brought progressively more distal facies into the short-term facies succession. The short-term stratigraphic cycles are stacked to form an intermediate-term base-level rise cycle reflecting increasing accommodation through time.

Sandstone with Rip-up Clasts (rupSs) Facies rupSs occurs uncommonly and normally overlies scour surfaces; in some cases, it is interlayered with facies blSs. The upper contact is either transitional or sharp. Bed thickness commonly ranges between 12 and 20 cm (5 and 8 in.); however, beds as much as 0.6 m (2 ft) occur. This quartz arenite is fine to lower medium grained. Mudstone rip-up clasts are as much as 3 cm (1.2 in.) long in dimension and occasionally so abundant that the rock is almost a ripup clast conglomerate (Figure 14C). Variable degrees of burrowing and irregular lamination occur in facies rupSs. Where this facies overlies a scour surface, it is interpreted as a crevasse channel lag deposit. The mud clasts were either ripped up because of channel scouring and incorporated in the flow or were derived from slumps into the channel and incorporated in the flow. When facies rupSs does not overlie a scour surface, it is interpreted as related to an upstream scour event. Facies rupSs occurs in the proximal parts of a crevasse splay. In a stratigraphic context, it represents relatively lower accommodation conditions in a crevasse splay succession.

Small-scale Trough Cross-stratified Sandstone (stxSs)

FIGURE 14. (cont.).

This facies constitutes as much as 30% of the crevasse splay intervals cored in the Mirador Formation. It occurs in proximal portions in a crevasse splay

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 451

FIGURE 15. Crevasse-splay facies succession in cycle I of the Cupiagua A-1 well. Individual short-term stratigraphic cycles represent episodes of progradation, but each episode brings progressively more distal facies into the short-term facies succession. The short-term stratigraphic cycles are stacked to form an intermediate-term base-level-rise cycle reflecting increasing accommodation through time. (See Figure 9 for legend.)

complex. The lower contact is either a scour surface or a transitional surface from underlying ripplelaminated sandstones. The upper contact is sharp or transitional. Bed thickness commonly ranges between 5 and 15 cm (2 and 6 in.); uncommonly, beds

are thicker than 0.3 m (1 ft). The quartz arenite is lower to middle fine grained and well sorted. Smallscale trough cross-bedding is the most common sedimentary structure; however, different degrees of burrowing partially destroy the foreset laminae. Foreset laminae are inclined 10–208 and are 1–2 mm (0.04–0.08 in.) thick. The type of burrowing was not identified, but there is a low diversity of ichnofossils. Figure 14D shows a photograph of this facies. The small-scale trough crossstratification is the record of migrating 3-D dunes in both channelized and unconfined unidirectional flow in the lower flow regime. Burrowing is evidence of fluctuations in flow strength that allowed reworking of sediments by organisms. Facies stxSs is interpreted as part of a crevasse channel deposit where there is a succession from a scour surface into facies stxSs, then into rpSS, and finally, into facies blSs. When the succession is from facies blSs into rpSs and into stxSs, this facies is interpreted as a crevasse splay prograding into a swamp and flood plain. Facies stxSs occupies proximal portions in a crevasse splay complex. This facies commonly appears in the lower parts of the crevasse splay intervals cored in the Mirador Formation. In a stratigraphic context, facies stxSs represents relatively lower accommodation conditions in a crevasse splay deposit.

452 / Ramon and Fajardo

Ripple-laminated Sandstone (rpSs) Facies rpSs constitutes as much as 40% of the crevasse splay facies succession. Upper and lower contacts of the ripple-laminated sandstone are either sharp or transitional. Bed thickness generally ranges between 2.5 and 15 cm (1 and 6 in.); however, beds as much as 0.45 m (1.5 ft) thick are found. The quartz arenite is fine to very fine grained and well sorted (Figure 14E). Differential degrees of burrowing partially destroy the ripple laminae. This facies represents a migration of ripples in the lower flow regime in either channelized or unconfined flow. This facies appears in two different successions in a crevasse complex. It occurs in crevasse channel deposits, where it is interpreted as a waning flow cap of the crevasse channel succession. It also appears in the lower half of shallowing-upward successions of prograding crevasse splays and is interpreted as deposition in unconfined flow. Facies rpSs is better developed in more distal parts of a crevasse splay facies succession than facies stxSs and rupSs. In a stratigraphic context, those intervals with major proportion of facies rpSs are interpreted as deposited during higher accommodation conditions than intervals dominated by facies stxSs.

Convolute Sandstone (cvSs) This facies is uncommon in the crevasse splay facies succession. Bed thickness ranges between 5 and 25 cm (2 and 10 in.) (Figure 14F). The quartz arenite is very fine grained and well sorted. Relicts of ripple laminae are common. This facies is interpreted as rapidly deposited sediments that collapsed because of water escape processes. This facies is associated with facies rpSs or blSs. It was deposited in similar accommodation conditions as facies rpSs.

Burrowed, Laminated Sandstone (blSs) This facies constitutes as much as 10% of the crevasse splay facies succession cored in the Mirador Formation. It occupies the distal part of a crevasse splay complex. Upper and lower contacts are either transitional or sharp. Bed thickness is most commonly between 5 cm (2 in.) and 0.6 m (2 ft). The quartz arenite is very fine to fine grained and well sorted. Clay matrix is present in some intervals. The most distinctive feature of this facies is the irregular arrangement of the lamination caused by a high degree of burrowing. Sometimes, the rock is completely or almost completely homogenized by organism reworking. Some intervals with lower degree of burrowing

exhibit carbonaceous or muddy irregular laminations and relict foreset laminae of short-term trough cross-stratification (Figure 14G). Facies blSs was deposited in an oxygenated environment where benthic organisms reworked the sediments that were deposited. It is found in distal parts of prograding crevasse splays and on top of crevasse channels. This facies was deposited in relatively higher accommodation conditions than facies stxSs or rpSs.

SWAMP AND FLOOD-PLAIN FACIES SUCCESSIONS The swamp and flood-plain facies successions are dominated by mudstones, and they generally occur adjacent to the crevasse splay facies succession. They are more common toward the top of the intermediate cycles of the Lower Mirador (Figure 10). Most mudstones correspond to wet flood plain and swamps. Locally, these are replaced transitionally by increasing thickness of soils. An unconformity is interpreted toward the top of the Lower Mirador in the Cusiana field. This unconformity is underlain by soils, caliche, and brecciated mudstones and is overlain by strata, which accumulated in wet flood-plain, marsh, and swamp to brackish bay environments (Fajardo, 1995). Three facies were differentiated in the swamp facies succession: massive mudstone (mMd), laminated mudstone with starved ripples (lsrMd), and burrowed and irregular laminated mudstone (blMd). Three facies where identified in the flood-plain facies succession: mottled mudstone (mtMd), rooted and irregular laminated mudstone (rtMd), and brecciated mudstone (brMd). A facies substitution diagram for swamp and floodplain facies succession is presented in Figure 16. This diagram represents the succession of facies under decreasing accommodation conditions from swamp to flood-plain facies succession. The left part of the diagram represents drier conditions. The right part of the diagram represents wetter conditions or even brackish conditions.

Massive Mudstone (mMd) This facies occurs between crevasse splay deposits or with other swamp facies. Lower and upper contacts are commonly sharp, where it is interbedded with crevasse splay facies successions, but where it is interbedded with other swamp facies associations, the contacts are generally transitional. Bed thickness is variable. Where this facies is interbedded with crevasse splay sandstones, the bed thickness ranges from

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 453

2.5 to 12 cm (1 to 5 in.). The color is commonly dark gray and medium gray. Some mudstones have high organic matter content. Facies mMd was deposited from suspension in open-swamp and openlake conditions. This facies is interpreted as the deepest swamp facies.

Laminated Mudstone with Starved Ripples (lsrMd) This facies consist of laminated gray mudstones intercalated with variable amounts of silt and few very fine sandstone ripples. Upper and lower contacts are either sharp or transitional. When it is interbedded with crevasse splay sandstones, contacts are sharp, and beds range between 2.5 and 15 cm (1 and 6 in.) thick. Beds of this facies in open-swamp intervals range between 0.3 and 1 m (1 and 3 ft) thick; in exceptional cases, beds as much as 1.8 m (6 ft) thick occur. Texturally, this facies varies from silty mud to muddy silt. It is commonly dark to middle gray and occasionally greenish gray or black. Plane parallel horizontal lamination, starved ripples, and lenticular and wavy laminations are the characteristic sedimentary structures (Figure 14H). This facies was deposited from suspension in open-swamp environments. It is interpreted as having been deposited at the same depth or even a shallower depth than facies mMd. The starved ripples and lenticular and wavy lamination are evidence of weak currents.

Burrowed and Irregular Laminated Mudstone (blMd)

FIGURE 16. Swamp and flood-plain facies successions in the Cupiagua A-1 well. The succession from swamp to flood plain defines an overall increase in accommodation space in intermediate-scale cycle III. Cycle IV is composed of wetter mudstones of marsh and brackish ponds. (See Figure 9 for legend.)

Lower and upper contacts are either transitional or sharp. Bed thickness ranges from 0.3 to 1 m (1 to 3 ft). It is generally composed of muddy siltstone but may be sandy siltstone. It exhibits different colors, including dark to light gray, greenish gray, and beige. Burrowing is the most distinctive feature of this facies. Irregular

454 / Ramon and Fajardo

laminations are frequently present. On top of the lower Mirador, facies blMd changes transitionally to mottled and rooted mudstones (mtMd or rtMd). This facies was deposited from suspension in a swamp or in ephemeral flood-plain ponds. In a stratigraphic context, where this facies is interbedded with facies mtMd or rtMd, it is interpreted as the result of an overall decrease in accommodation space. In contrast, where it is interbedded with facies mMd and lsrMd, it is interpreted as having been deposited during an overall increase in accommodation space.

Mottled Mudstone (mtMd) This facies’ most distinctive feature is its varicolored aspect. Upper and lower contacts are transitional. Bed thickness ranges between 0.6 and 2.7 m (2 and 9 ft). The mudstones of some intervals have a dark-gray background with irregular reddish-brown patches; other mudstones are light gray or beige with reddish and yellowish patches, and others are green or beige with blue and purple patches. A common feature is the vertical distribution and orientation of patches and irregular veins (Figure 14I). This facies is interpreted as deposited from suspension in a flood-plain environment where there are alternating dry and wet conditions. In a stratigraphic context, this facies represents a time of decreasing accommodation.

Rooted and Irregular Laminated Mudstone (rtMd) This facies is beige and light gray; it contains black to brown root traces composed of carbonaceous material and is interbedded with facies mtMd. The basal and upper contacts are commonly transitional. Bed thickness ranges between 0.3 and 2.1 m (1 and 7 ft). Burrows are another feature present in this facies (Figure 14J). This facies represents the soil alteration of a previous facies deposited from suspension. Two interpretations of this facies are possible. Where it is interbedded with facies mtMd, it is interpreted as deposited in a flood-plain setting where drier conditions predominated. Where it contains organic laminations and it is interbedded with facies mMd, it is interpreted as deposited in a wetter environment like an ephemeral pond.

Brecciated Mudstone (brMd) This facies constitutes a small percentage of the swamp and flood-plain facies successions. The lower contact is transitional, and the upper contact is sharp.

FIGURE 17. Facies succession and substitution diagram of the bay-fill and the channel and bay-head facies tract. Bay-fill and bay-head cycles tend to have a base-level-fall asymmetry with a thin base-level-rise cap. (See Figure 9 for legend and text for abbreviations.)

Bed thickness ranges between 1 and 1.5 m (3 and 5 ft). This facies appears brecciated, and it is composed of angular, irregular masses of brownish color with chaotic distribution (Figure 14K). In some intervals, ferruginous pseudonodules of less than 2 mm (0.08 in.) are present. This facies results from modification of previously deposited mudstones by soil processes. The facies has strong evidence of subaerial exposure.

Bay Facies Tract The bay facies tract occurs in the upper Mirador Formation. In this facies tract are channel, bay-head delta, and bay-fill facies successions (Figures 10 –12). A transition from channel to bay-head delta to bay-fill facies successions represents a continuous increase in accommodation space. A transition from bay-fill to bay-head delta to channel facies succession defines a continuous decrease in accommodation space. A facies substitution diagram of the bay facies tract is presented in Figure 17. This diagram was constructed from the bay facies tract intervals cored in three wells in the Mirador Formation in the Cupiagua field.

CHANNEL AND BAY-HEAD DELTA FACIES SUCCESSIONS Channel and bay-head delta facies successions contain the main reservoir rock in the bay facies tract. These successions are dominant in the lower half of the upper Mirador in the Cupiagua field area. Three

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 455

facies are recognized in these successions: trough cross-stratified sandstone (txSs); burrowed sandstone with relict trough cross-stratification (btxSs); and ripple-laminated sandstone (rpSs). Figures 10–12 show examples of these facies successions.

Trough Cross-stratified Sandstone (txSs) Facies txSs is the most common facies of the bay facies tract. It occurs both in channel and bay-head delta facies successions. Upper and lower contacts of trough cross-stratification sets are commonly sharp; however, when burrows are present, the contacts are indistinct. The quartz arenite normally is medium grained, but coarse-grained intervals occur. Finingupward or coarsening-upward trends are recognized in a vertical succession of sets. Thickness of trough cross-stratification sets is variable; it ranges from few inches to as much as 1.2 m (4 ft). Foreset laminae are inclined between 15 and 258 and are between 2 and 8 mm (0.08 and 0.3 in.) thick. Minor burrowing occurs in this facies. With the exception of uncommon burrows, this facies is virtually identical to facies txSs of the coastal-plain facies tract (see Figure 14B). Figure 18A is a photograph of this facies. Facies txSs was deposited by turbulent currents in the lower flow regime. The trough cross-stratified sandstone is the stratigraphic record of migrating 3-D dunes on the channel floor or in bay-head deltas.

Bioturbated Sandstone with Relict Trough Cross-stratification (btxSs) This facies is the second most common facies of the bay facies tract. It occurs in the upper part of aggradational channel successions and in the lower part of progradational bay-head delta successions. Trough cross-stratification set thickness is sometimes difficult to observe but, in general, is less than 0.45 m (1.5 ft). The quartz arenite is medium to coarse grained. The most characteristic feature is the burrowing. More than 60% of the original trough crossstratification is destroyed, and relicts of foresets with variable angle of inclination are visible throughout the burrowing overprint (Figure 18B). Macaronichnus, Ophiomorpha, Gyrolithes, Arenicolites, and crab burrows are recognized. Where the burrow intensity is constant between two scour surfaces, this facies was deposited under low sedimentation rates. Where burrow intensity decreases from top to bottom, it is interpreted as fluctuations in rates of accumulation; periods of bed migration are interrupted by times of nondeposition when organisms reworked the sediment.

Ripple-laminated Sandstone (rpSs) This facies is uncommon (2%) in the bay facies tract. It sometimes occurs on top of channel facies successions and uncommonly within bay-fill facies successions. In channel successions, the thickness is as much as 0.6 m (2 ft); in bay-fill successions, it is less than 20 cm (8 in.). The quartz arenite is fine grained and commonly contains clay matrix. It exhibits several degrees of burrowing, but relict ripples can always be easily identified. This facies was deposited in the lower flow regime, and it represents the migration of 3-D ripples. The intensive burrowing suggests low rates of deposition.

BAY-FILL FACIES SUCCESSION This facies succession is composed of burrowed sandstones and mudstones. It is more common in the upper half of the upper Mirador. Four facies are defined in this facies succession. From deeper to shallower, those facies are laminated mudstone (lMd), burrowed and irregular laminated mudstone (blMd), burrowed and irregular laminated sandstone (blSs), and vertical burrowed sandstone (vbSs) (see Figure 17). A transition from lMd to blMd to blSs and, later, into vbSs represents a continuous decrease in accommodation space (Figure 17). In some cases, facies vbSs is overlain by facies hbSs; this upward transition represents an increase in accommodation space.

Laminated Mudstone (lMd) This facies is uncommon in the bay facies tract. It occurs at the base of shallowing-upward bay-fill successions. Upper and lower contacts are normally sharp. Bed thickness is less than 1 m (3 ft). The mudstone is dark gray to black and exhibits discontinuous silt laminae or lenses (Figure 18C). This facies is interpreted as having been deposited from suspension. The lamination is evidence of weak currents. The absence of burrowing indicates that dysaerobic or anaerobic conditions occurred at the sediment-water interface.

Burrowed and Irregularly Laminated Mudstone (blMd) This facies constitutes less than 4% of the bay facies tract. It occupies the lower part of the shallowingupward bay-fill successions. Upper and lower contacts are either sharp or transitional. Bed thickness is variable, ranging from few inches to as much as 1.2 m (4 ft). This facies is composed of dark-gray or black

456 / Ramon and Fajardo

FIGURE 18. Photos of the different facies present in the bay facies tract in the upper Mirador in the Cupiagua field. (A) Trough cross-stratified sandstone; (B) bioturbated sandstone with relict trough cross-stratification; (C) laminated mudstone; (D) bioturbated and irregularly laminated mudstone (3797 m [12,457 ft] measured depth); (E) bioturbated and irregularly laminated sandstone; (F) vertically burrowed sandstone. All cores are 10 cm (4 in.) wide.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 457

silty mudstone. The more distinctive characteristic is the intensive burrowing. Burrows are mainly horizontal. In some intervals, Thalassinoides burrows are recognized (Figure 18D). Irregular and discontinuous laminations normally occur. This facies was deposited from suspension in openbay conditions. The intensive burrowing implies optimal substrate conditions for organisms to live.

Bioturbated and Irregularly Laminated Sandstone (blSs) This facies constitutes 30% of the bay-fill facies tract. Bed thickness normally is between 15 cm (6 in.) and 0.3 m (1 ft); beds as much as 1 m (3 ft) thick occur. It is composed of fine and very fine sandstone with variables amounts of clay matrix. It has carbonaceous or muddy irregular laminations and variable degrees of burrowing (Figure 18E). The major difference with facies vbSs is that single ichnofossils are not easily distinguished. Mudstone rip-up clasts occur in some intervals. This facies was deposited in open-bay environments at slow rates, which allowed intense reworking by organisms.

Vertically Burrowed Sandstone (vbSs) This facies is the most common of the bay facies tract. It occupies the upper part of shallowing-upward bay-fill successions. Upper and lower contacts are normally transitional with overlying and underlying facies. Bed thickness is variable, ranging from a few inches to 2 m (6 ft). The quartz arenite is fine and sparsely very fine or medium grained. The distinctive feature is the dominance of vertical burrows, such as Ophiomorpha, Skolithos sp., and Arenicolithes (Figure 18F). Horizontal burrows are uncommon and include Teichichnus, Planolites, and Paleophycus. This facies lacks ripple lamination or cross-stratification, but exhibits carbonaceous or muddy horizontal wispy laminations. This facies was deposited at higher rates than facies blMd and blss in open-bay settings. Facies vbSs was deposited in open-bay environments at slow rates, which allowed reworking by organisms.

FOUR-DIMENSIONAL STRATIGRAPHIC ARCHITECTURE OF THE MIRADOR FORMATION

FIGURE 18. (cont.).

Three scales of stratigraphic cycles are recognized in the Mirador Formation. Short-term (high-frequency) cycles correspond to progradational and aggradational units. Six intermediate-term cycles were identified by the stacking patterns of their component short-term

458 / Ramon and Fajardo

FIGURE 19. Correlation of the cored wells in the Cupiagua field showing the long-term stratigraphic cycles defined in the Mirador Formation. Depth in feet. cycles and by the general trend of facies successions, indicating increasing or decreasing accommodationto-sediment supply (A/S) ratios. Two long-term cycles were defined from the stacking pattern of the intermediate-term cycles and by the general trend of facies successions (Figure 19). Stratigraphic cycles, regardless of scale, record a complete base-level cycle (Barrell, 1917; Wheeler, 1964). During a base-level cycle, accommodation-tosediment supply ratio increases (as base-level rises) to a maximum limit and then decreases (as base-level falls) to a minimum limit along the entire geomorphic profile connecting linked depositional environments.

Short-term Stratigraphic Cycles Short-term stratigraphic cycles are progradational and aggradational units that conform to Walther’s law. These short-term cycles are the building blocks of the stratigraphic framework. Aggradation occurs

by deposition on an almost horizontal surface; examples of aggradational units are channel, lake, and flood-plain deposits. Progradation occurs by deposition on an inclined surface; examples of progradational units are crevasse splay, bay-head delta, and bay-fill deposits. A single short-term cycle generally is composed of both aggradational and progradational components because of changes in inclination of depositional surfaces along the geomorphic profile. The definition of short-term cycles is based on facies successions, occurrence of surfaces of stratigraphic discontinuity, and measurements of sedimentologic attributes like bed set thickness or burrow orientation, diversity, and density. Short-term stratigraphic cycles in channel facies successions commonly exhibit base-level rise asymmetry, and less commonly with a thin base-level fall hemicycle. Asymmetry means that the time of increasing A/S is preserved as rock, whereas the time of

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 459

decreasing A/S is mainly represented by surfaces of stratigraphic discontinuity. In the Mirador Formation, typical thickness of short-term cycles composed of channel facies successions is 3– 6 m (10 –20 ft), and the range is 2.1 – 9 m (7 – 30 ft). When channel successions exhibit enough facies diversity, the recognition of base-level rise deposits is based on a vertical change of facies representing reduction in flow concentration and/or flow strength. A complete succession would include facies deposited as hyperconcentrated flows, followed by facies deposited in the upper flow regime, followed by facies deposited in the lower flow regime. This succession could be indicated by a transition from granule sandstone with upper plane bed stratification (uppSs) facies at the channel base to trough cross-stratified granule sandstone (txgSs) or trough cross-stratified sandstone (txSs) facies, followed by trough cross-stratification with some burrowing (btxSs), ripple-laminated sandstone (rpSs), or convolute-laminated sandstone (cvSs) facies (Figure 13). The base-level-fall hemicycle is commonly represented by a scour surface. However, if it is preserved as rock, it is recognized by facies successions that indicate an increase in sediment reworking and amalgamation. A complete succession would be a transition from facies deposited in the lower flow regime to facies deposited in the upper flow regime and even to facies deposited as hyperconcentrated flows. Short-term stratigraphic cycles in crevasse splay facies successions exhibit a base-level fall or baselevel rise asymmetry or are symmetric alternating rise and fall hemicycles (Figure 15). Thickness of these short-term cycles in the Mirador Formation ranges from 0.9 to 5.1 m (3 to 17 ft). Definition of cycles in crevasse splay facies successions is based on the recognition of shallowingupward intervals developed during the progradation of a crevasse splay into permanent or ephemeral flood-plain lakes. Sometimes, the crevasse facies become less frequent and thinner, and swamp and floodplain facies becomes thicker and more frequent. This transition is interpreted as a rise base-level. Short-term stratigraphic cycles in lake and floodplain facies successions are more symmetrical than in any other facies succession of the coastal-plain facies tract. Thickness of Mirador lake and floodplain short-term cycles ranges from 0.6 to 4.8 m (2 to 16 ft). The definition of short-term stratigraphic cycles in intervals composed only of swamp and floodplain facies successions is based on the interpreted wetness and dryness of the different facies that occur

in the interval. From deep to shallow, the swamp facies succession is from massive mudstone (mMd), to laminated mudstone with starved ripples (lsrMd), to burrowed and irregularly laminated mudstone (blMd), to gray mudstones with roots to mottled mudstones. This transition from swamp (wet) to dry flood plain is interpreted as a rise in base level. If flood-plain facies successions occur without intervening swamp and lake beds, short-term cycles are defined by the degree of soil development. A succession from mtMd to brecciated mudstone (brMd) facies indicates a base-level-rise hemicycle. In intervals composed only of mtMd facies, short-term cycles are defined by changes in the degree of mottling. An increase in mottling (increased duration of soil-forming processes) is interpreted as a base-level rise, and a decrease in mottling is interpreted as a base-level fall as more sediment is aggraded over the flood plain. Short-term stratigraphic cycles in the bay facies tract were identified in channel, bay-head delta, and bay-fill facies successions. A single short-term cycle can be composed of only one or a combination of these facies successions. Channel and bay-head delta facies successions commonly coexist at each geographic location in a single short-term cycle. Cycles are base-level rise or fall asymmetrical, or they are symmetrical. Channels are preserved during both short-term base-level rise and fall. In contrast, bay-head deltas only occur during shortterm base-level-fall time. Bay-head deltas are probably replaced by bay-fill facies successions during shortterm base-level-rise time. Cycle thickness is variable: cycles composed of channel and bay-head delta successions range from 3.3 to 9 m (11 to 30 ft), but cycles composed of either one or the other facies succession range from 1.2 to 3.6 m (4 to 12 ft). A short-term base-level-rise hemicycle in channel facies successions is indicated when a channel scour surface is overlain by trough cross-stratified sandstones (txSs), which constitutes the major part of the channel deposit and is followed by burrowed sandstone with relict trough cross-stratification (btxSs) and, in some cases, is capped by rippled sandstone (rpSs). The increase in burrowing toward the top and the decrease in flow velocity reflect increasing accommodation and gradual drowning of the channel. Short-term base-level-fall hemicycles in bay-head delta successions are recognized from facies successions and changes in cross-stratification set thickness. Bay-head delta successions exhibit both transitional and sharp lower contacts with underlying facies. A typical facies succession is a change from

460 / Ramon and Fajardo

FIGURE 20. (A) Channel and bay-head delta facies successions, Cupiagua A-1 well. Trough cross-stratified sandstone in the lower part of bay-head delta succession is intensively burrowed, and the burrowing decreases upward. The core photograph (B) shows the same interval. Note the better oil saturation toward the top of the core (right), indicating better reservoir quality. (See Figure 9 for legend.) Depth in feet. burrowed sandstone (bSs) at the base to burrowed sandstone with relict trough cross-stratification (btxSs) and then to trough cross-stratified sandstone (txSs). Burrowing intensity diminishes toward the top, and trough cross-stratification set thickness in facies txSs thins upward. These facies successions record an increase in the hydromechanical energy from base to top and increased cannibalization of migrating dunes during progradation, which are interpreted as an upward decrease in accommodation (Figure 20). Short-term stratigraphic cycles in bay-fill facies successions tend to have base-level-fall asymmetry with a thin base-level-rise cap. Typical cycle thickness ranges from 1.2 to 3.6 m (4 to 12 ft), but in exceptional cases, cycles are as much as 5.4 m (18 ft) thick. Short-term cycles commonly are recognized from facies successions of shallowing-upward bay-fill deposits. In a typical shallowing-upward bay-fill succession, there is a transition from burrowed and irregularly laminated mudstone (bMd) at the base to burrowed and irregularly laminated sandstone (blSs), followed by vertical burrowed sandstone (vbSs) toward the top. This succession represents a continuous decrease in accommodation-to-sediment supply ratio caused by progradation. The bayhead delta facies succession sometimes overlies bay-fill successions in a single short-term cycle (Figure 20).

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 461

Thin base-level rise caps are recognized in bay-fill deposits. They exhibit a succession from shallower to deeper facies, from vbSs to blSs facies, and eventually, to bMd facies. The short-term cycles have a thicker base-level-fall hemicycle with thin base-level-rise caps.

Intermediate-term Stratigraphic Cycles Six intermediate-term stratigraphic cycles are defined in the Mirador Formation in the Cupiagua field (Figures 10 – 12, 19). Intermediate-term stratigraphic cycles are defined from the stacking pattern of their component short-term cycles, from the general trend of facies successions, and from surfaces of stratigraphic discontinuity or dislocation. These cycles are designated with Roman numerals (I–VI). Facies assemblages of the three basal cycles (I, II, and III) belong to the coastal-plain facies tract. Cycles IV, V, and VI are composed of bay facies tracts.

Cycle I Cycle I is only cored in the Cupiagua A-1 well. This cycle exhibits mostly flood-plain and crevasse facies tracts. This core shows a transition from stacked flood-plain facies of the Los Cuervos Formation into crevasse channel and splay facies with some swamp and wet flood-plain facies successions. This trend of facies successions suggests a base-level fall to rise progression (Figure 10). Well-log correlation indicates that laterally, the crevasse sequences change into channel-belt sandstones (Figure 21). These channelbelt sandstones tend to be single story and laterally discontinuous. The transition from aggradational flood-plain deposits into crevasse sandstone bodies indicates the progradation of the crevasses over a distal part of the flood plain. The crevasse splay facies succession at the upper half of this cycle occurs in six short-term cycles (Figure 15). Short-term cycles cored in Cupiagua A-1 well consist of more distal crevasse splay facies, so each of these cycles steps landward (away from the channel belt) with respect to the underlying one defining an overall base-level rise in the upper part of this intermediate cycle. Fajardo (1995) showed that this lowermost intermediate cycle in the adjacent Cusiana field is composed of channel to crevasse to flood-plain facies tracts. His cycle starts with the erosion at the base of the channel belt and ends with the next channel belt base. This is interpreted as a base-level-rise cycle. Fajardo (1995) made an isopach map of his cycle I, which shows that the cycle occurs only in the southern part of the Cusiana field. Cross sections across the

Cusiana field show that the proportion of swamp and flood-plain facies successions increases from southeast to northwest, and the proportion of crevasse splay facies successions decreases in this direction. The lower proportion of the channel facies tracts with respect to the crevasse and flood-plain facies tracts in the Cupiagua field area suggests that depositional dip orientation was toward the northwest. This is consistent with the trend of paleovalleys mapped by Fajardo (1995).

Cycle II The lower boundary of intermediate-term cycle II is a base-level-fall surface. This cycle exhibits a baselevel-rise asymmetry. Cycle II consists of channel and crevasse splay facies successions. The flood-plain facies succession is only found in a couple of wells. Cycle II has a transition from channel to crevasse splay facies successions (Figure 10), which, in some cases, are overlain by the swamp and flood-plain facies succession. This trend of facies successions defines an intermediate-term base-level rise. Short-term cycles in channel and crevasse splay intervals have a general landward-stepping pattern in the base-levelrise hemicycle. Channel facies successions are dominant in the lower half of cycle II. Facies that compose these cycles have a general decrease in amalgamation and an increase in bedform preservation from base to top. Crevasse splay facies successions occur in the upper half of the intermediate-term base-level cycle II (Figure 10). These short-term cycles are successively composed of deeper or more distal crevasse splay facies; hence, these short-term cycles have a landward-stepping stacking pattern that defines an increasing accommodation regime. Intermediate-term cycle II has a higher proportion of channel facies tracts than cycle I. In addition, the channel facies tracts are thicker and show a much higher degree of amalgamation. Channel sandstone bodies are laterally more continuous (Figures 19, 21). These indicate that cycle II records a seaward stepping of the depositional systems of the alluvial plain. In Cupiagua, this cycle has the thickest proportion of granule facies. Also, channels at the base of this cycle are more amalgamated than cycle I (Fajardo, 1995). This evidences also a seaward-stepping stacking pattern.

Cycle III This intermediate-term cycle shows the highest degree of facies variability. In the three wells (A-1, C-3, and H-11; Figures 10–12) that cored this interval, the

deposits cored in A-1 change laterally into channel-belt sandstones. Depth in feet.

FIGURE 21. Well-log correlation of the A-1, Q-6, and U-23 wells of the Cupiagua field. Cycle I shows lateral facies changes. Crevasse and swamp and flood-plain

462 / Ramon and Fajardo

FIGURE 21. (cont.).

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 463

464 / Ramon and Fajardo

cycle has diverse proportions of channel, crevasse, and flood-plain facies tracts. Cycle III is more symmetrical than cycles I and II, but symmetry changes from well to well. In the A-1 well, the cycle is composed primarily of agradational flood-plain facies tracts. The base-level-rise hemicycle is composed of crevasse splay (20%) and swamp-flood-plain (80%) facies successions (Figure 10). The base-level-fall hemicycle is composed of swamp (65%), flood-plain (15%), and channel (20%) facies successions. In the C-3 well, the cycle consists mostly of channel facies successions (Figure 12). Minor intercalations of channel abandonment and crevasse facies successions occur. Well H-11 is composed of channel (60%), crevasse (20%), and swamp (20%) facies tracts (Figure 11). The baselevel-rise hemicycle is characterized by a transition from channel to crevasse splay to swamp and floodplain facies successions (Figure 11). This trend of facies successions defines an overall increase in accommodation space. The intermediate-term baselevel-fall hemicycle is characterized by a transition from flood-plain to crevasse to channel facies successions. This trend defines an overall decrease in accommodation/sediment supply ratio. The turnaround point from intermediate-term base-level rise to fall is located within the massive mudstone facies of the swamp and flood plain. Channel-belt sandstones of this intermediate-term cycle III are laterally discontinuous and less amalgamated as compared to the ones in cycle II. The highest proportion of flood-plain and crevasse facies tracts and the single-story character of the channelbelt bodies indicate that cycle III represents a landward stepping into more distal, lower energy facies tracts.

Cycle IV Intermediate-term cycle IV marks a major change in the depositional setting in the Cupiagua field area. Cycle IV is composed of channel (30 –60%), bay-fill (0 10%), and bay-head delta (30 –70%) facies successions. The lower boundary is generally a baselevel-fall surface, whereas the upper boundaries are base-level-rise flooding surfaces. This cycle commonly exhibits base-level-rise and base-level-fall hemicycles of similar thickness. Mudstones of cycle IV are commonly black or dark gray (Figure 18D) and are burrowed as compared to mudstones in cycle III, which exhibit light- to medium-gray, brown, and purple mottling (Figure 14J, K). The change from mottled mudstones to dark-gray mudstones is inferred to reflect a change from drier, soil-forming conditions to wet coastal-plain or bay conditions.

Although facies changes from well to well, the base-level-rise and base-level-fall hemicycles are present in all wells. Well A-1 records the base-level rise as the transition from channel to bay-fill and the baselevel fall with the change from bay-head delta back into estuarine-channel facies tracts (Figure 10). Well H-11 records the rise-to-fall cycle as the transition from estuarine-channel to bay-head and back into channel facies tracts (Figure 11). Finally, well C-3 shows a transition from channel to bay-head facies tracts (Figure 12). The presence of new more distal brackish to shallow-marine facies tracts indicates that cycle IV records a major landward stepping of all facies tracts. The alluvial facies tracts dominant in the Lower Mirador (cycles I, II, and III) shifted toward the east at the time of cycle IV.

Cycle V Intermediate-term cycle V records a continuous landward shift of facies tracts. This cycle is composed by bay-head delta (40 –70%) and bay-fill (30– 60%) facies successions. Channel facies tracts are only locally present toward the top of the cycle. The lower and upper boundaries are turnaround points (fall to rise) in conformable strata and are located at the position of maximum thickness of channel and bay-head delta facies successions. Cycle V is generally symmetrical. The base-level-rise hemicycle of cycle V is characterized by a transition from bay-head delta to bayfill facies successions. Short-term cycles show landwardstepping stacking patterns. Progressively, bay-head facies becomes thinner and more bioturbated, indicating preferential preservation of the distal portions. In the same direction, bay-fill facies tracts become thicker. Bay-fill mudstones are better developed to the middle part of the cycle. This cycle has the highest proportion of distal bay-fill facies tracts. This suggests a continuous landward-stepping pattern. The base-level-fall hemicycle of cycle V records the transition from bay-fill to bay-head delta facies successions, which is sometimes capped by channel facies successions. This trend of facies successions indicates decreasing A/S. Short-term cycles have a seawardstepping stacking pattern. This transition represents a shallowing-upward profile produced by progradation.

Cycle VI This intermediate-term cycle is asymmetrical where mostly the base-level rise is preserved. This cycle is only cored at well A-1. The lower boundary is located within conformable strata at the turnaround point

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 465

from decreasing to increasing accommodation-tosediment supply ratio. Cycle VI is composed of bayfill (40– 70%) and bay-head delta (30 –60%) facies successions. In the rise hemicycle, the proportion of deeper facies progressively increases, and shallower facies decreases within short-term cycles. This upward increase in the proportion of deeper facies means that short-term cycles have a landward-stepping stacking pattern. In few wells, the base-level-fall hemicycle is made up of short-term cycles with a general seaward-stepping stacking pattern. The upper half of cycle VI shows channel facies successions directly over the bay-fill facies tract. This would indicate an anomalous seaward shift in facies (downhill offset in facies). Above the channel facies tract, a rapid landward shift in facies without stratigraphic discontinuity occurs. This landward offset places deeper bay-fill facies of the C8 member of the Carbonera Formation on top of channel deposits. The higher proportion of channel facies tracts in cycle VI relative to cycle V suggests a seaward-stepping stacking pattern. Above the seaward facies offset, the landward-stepping stacking pattern is maintained up to the top of the Mirador Formation. Both the seaward and landward shift in facies are interpreted as anomalies in the stacking pattern and are not considered boundaries of intermediate-term stratigraphic cycles. These stratigraphic anomalies were identified in the cored well (A-1, Figure 10) and might be present in uncored wells.

RESERVOIR ZONATION OF THE MIRADOR FORMATION High-resolution stratigraphic correlation, defined as establishing temporal equivalency of rock units regardless of rock type, mineralogy, texture, or environment of deposition is the base for a more accurate and predictive reservoir characterization. Because the basic lithological, petrophysical, geometrical, and reservoir continuity attributes of rocks originate during sediment accumulation, accurate correlation permits accurate representation of these rock properties in a four-dimensional (time-space) context. High-resolution correlation is the best means of identifying stratigraphically controlled fluid-flow compartments, compartment boundaries, and fluid-flow pathways in reservoirs. This is because the natural bounding surfaces of episodically accumulated stratigraphic successions compartmentalize strata into time-bounded packages of varying spatial and temporal scales. These natural

stratigraphic boundaries commonly coincide with the most laterally continuous shifts of facies, and consequently, the most pronounced changes in lithological properties commonly occur across those bounding surfaces. Thus, petrophysically significant boundaries of fluid-flow compartments commonly are coincident with time-significant stratigraphic bounding surfaces. Numerous occurrences of natural stratigraphic boundaries are not associated with petrophysical boundaries; but occurrences of petrophysical boundaries that are not associated with stratigraphic boundaries are far less common (Cross et al., 1993). A second stratigraphic attribute that can contribute to reservoir analysis is the relation between dynamic changes in accommodation/sediment supply ratio and the resulting systematic changes in stratal architecture and diversity, successions, and continuity of facies preserved in the stratigraphic record. For example, Allen (1978, 1979) and Bridge and Leeder (1979) suggested that fluvial channel-belt sands deposited during conditions of slow subsidence (equivalent to low accommodation) form vertically and laterally interconnected, blanketlike sandstone bodies that could function as single compartment reservoirs in terms of fluid flow and pressure. By contrast, channel-belt sandstones of identical depositional systems may occur as isolated, stringerlike reservoir sandstones if deposited during periods of faster subsidence (equivalent to high accommodation). Each sandstone body would function as a separate fluidflow compartment. The Lower Mirador cycles clearly show these differences in channel-belt architecture. Intermediate cycles I and III deposited under relatively higher accommodation conditions have thinner channel-belt sandstone bodies associated with thick crevasse and flood-plain facies tracts. Channel-belt sandstone bodies are laterally discontinuous between wells and show strong thickness changes. Facies tract proportions change from well to well, but crevasse and floodplain facies tracts are generally thicker than in cycle II (Figures 10, 19). As observed in Figure 10, the Cupiagua A-1 well has only a thin channel sandstone in cycle I. Close wells show fining-upward, bell-shaped gamma ray response that is interpreted as channel successions (Figure 21). Stratigraphic position and thickness of the channel facies successions change between wells. In contrast, channel-belt sandstone bodies are thicker and laterally more continuous in cycle II (Figures 10, 11). The relative proportion of overbank facies tracts in cycle II is lower as compared to cycles I and III. Channel sandstone bodies are laterally more

466 / Ramon and Fajardo

continuous and are present in almost every single well in the Cupiagua field (Figures 19, 21, 22). Finally, facies tracts control the quality of the fluidflow properties of the sandstone bodies. Porosity versus log permeability as a function of facies successions were plotted for intermediate-term cycles I to III (Figure 23). This plot shows that channel facies successions have higher porosity and permeability values than do crevasse splay facies successions in intermediate-term cycles I, II, and III of the Lower Mirador. These figures show two distinctive populations: one with higher porosity and permeability, which corresponds to the channel facies succession, and the other with lower porosity and permeability, which corresponds to the crevasse splay facies succession. Because both channel and crevasse splay facies succession intervals are laterally continuous throughout the field, these differences in petrophysical properties become critical in the reservoir zonation of the lower Mirador long-term cycle. Similar differences are found in the bay facies tract of the upper Mirador (cycles IV– VI; Figure 24). Channel and bay-head facies successions have higher permeability values for the same range in porosity as compared to the bay-fill facies successions. The plot shows that for a given porosity, the channel and bayhead sandstones have as much as two orders of magnitude higher permeabilities than the equivalent

bay-fill facies successions. In single wells, the different facies tracts control the porosity and permeability vertical distribution. Well H-11 has massive sandstones with only minor variations in the gammaray response over most of the Mirador Formation (Figure 25). Interpreted facies tracts correspond to major changes in the porosity and permeability values. These different facies tracts then define different units with different fluid-flow capabilities. The crevasse-splay facies successions typically have porosities in the 3–4% range and permeabilities lower than 0.1 md. Bay-head facies successions typically exhibit porosities in the 5–7% range and permeabilities of tens of milidarcys. Water saturation data from porous plate and Dean Stark analysis performed in core plugs indicate the existence of different rock types. Channel of the Lower Mirador and channel and bay-head facies successions of the upper Mirador show water saturation values ranging from 5 to 7%. Crevasse and bay-fill sandstones show water saturation values higher than 17%.

FLUID-FLOW UNITS AND STATIC MODEL OF THE MIRADOR FORMATION Previous reservoir characterization studies identified four mega-rock-types for the three producing reservoir sandstones (Guadalupe, Barco, and Mirador

FIGURE 22. Well-log correlation of the wells in the southern part of the Cupiagua field. Depth in feet.

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 467

FIGURE 23. Plot of porosity versus log permeability as a function of facies successions for the coastal-plain facies tract (intermediate-term cycles I to III).

FIGURE 24. Plot of porosity versus log permeability as a function of facies successions for the bay facies tract (intermediate-term cycles IV to VI).

468 / Ramon and Fajardo

FIGURE 25. Vertical variation of porosity and permeability in the Cupiagua H-11 well.

formations): mudstone (nonreservoir), lithoarenites, quartzarenites, and phosphatic lithoarenites. The lithoarenites have poor reservoir quality; the quartzarenites and phosphatic lithoarenites have mod-

erate reservoir quality. For each of these reservoir rock types, a porosity-permeability transform was obtained by comparing core with wire-line data. For static modeling purposes, to capture the reservoir quality

Sedimentology, Stratigraphy, and Architecture of the Eocene Mirador Formation / 469

variability in the quartzarenites, a total of four facies associations and their four porosity-permeability trends were used to cover the entire rock quality range of this mega-rock-type (Figures 23, 24). The facies associations identified in cored wells were extrapolated to the rest of the wells using electrical log signatures (gamma ray, spectral gamma ray, neutron, density, and resistivity). The interwell static modeling was generated using geostatistics techniques to populate the facies, followed by the porosity and permeability based on the facies tracts. Net pay is defined at a permeability cutoff equal or higher than 0.1 md, which is a 3.5% porosity and about 388 API gamma-ray cutoff for the Mirador Formation quartzarenites. These cutoffs have been evaluated at different scales from pore scale (mercury injection data) to centimeter scale in Repeated Formation Test flow response, to meter scale using core fluorescence presence ending with unit flows, producing intervals confirmed by production logging tools (PLTs) response. As mentioned before, facies associations were populated within the gross rock volume using stochastic methods. The 3-D stochastic reservoir model was built using the statistics of rock type distribution gained from the Cupiagua core suite and semiregional depositional model. Porosity and water saturation distributions were assigned to each of the rock types in the gross rock volume. Average porosity and permeability values in the geocellular static model are consistent with the well average values in each formation and during the upscaling process to generate the reservoir-simulation model; those averages were maintained representative of the well average values. The stock tank barrels of original oil initially in place (STOOIP) volumes were obtained from the initialization of the dynamic reservoir model, and they are good representations of the static stochastic 3-D model.

ACKNOWLEDGMENTS We are grateful to BP for the support to this work and permission to publish. The authors thank Steven Bachtel, Paul (Mitch) Harris, and an anonymous reviewer for their helpful suggestions.

REFERENCES CITED Allen, J. R. L., 1978, Studies in fluviatile sedimentation: Bars, bar complexes and sandstone sheets (low sinuosity braided streams) in the Brownstones (L. Dev.), Welsh borders: Sedimentary Geology, v. 26, p. 281 –293. Allen, J. R. L., 1979, Studies in fluviatile sedimentation: An exploratory quantitative model for the architecture of avulsion-controlled alluvial suites: Sedimentary Geology, v. 21, p. 129–147. Barrell, J., 1917, Rhythms and the measurement of geologic time: Geological Society of America Bulletin, v. 28, p. 745 – 904. Bridge, J. S., and M. R. Leeder, 1979, A simulation model of alluvial stratigraphy: Sedimentology, v. 26, p. 617 – 644. Cooper, M. A., et al., 1995, Basin development and tectonic history of the Llanos basin, eastern Cordillera, and Middle Magdalena Valley, Colombia: AAPG Bulletin, v. 79, no. 10, p. 1421 – 1443. Coral, M., and W. Rathke, 1997, Cupiagua field, Colombia: Interpretation case history of a large, complex thrust belt gas condensate field: Cartagena, Memorias del VI Simposio Bolivariano ‘‘Exploracio´n Petrolera en las Cuencas SubAndinas,’’ Colombia, Tomo 1, p. 119 – 128. Cross, T. A., et al., 1993, Applications of high-resolution sequence stratigraphy to reservoir analysis, in R. Eschard and B. Doligez, eds., Subsurface reservoir characterization from outcrop observations: Proceedings of the 7th Exploration and Production Research Conference: Paris, Teichnip, p. 11 – 33. Dengo, C., and M. Covey, 1993, Structure of the eastern Cordillera of Colombia: Implications for trap styles and regional tectonics: AAPG Bulletin, v. 77, p. 1315 – 1337. Fajardo, A. A., 1995, 4-D stratigraphic architecture and 3-D reservoir fluid flow model of the Mirador Fm., Cusiana field, foothills area in the Cordillera Oriental, Colombia: M.Sc. thesis, Colorado School of Mines, Golden, Colorado, 171 p. Martinez, J., 2003, Modelamiento estructural 3D y aplicaciones en la exploracio´n y explotacio´n de hidrocarburos en el Cinturo´n de Cabalgamiento del Piedemonte Llanero, Cordillera Oriental, Colombia: VIII Simposio Bolivariano — Exploracio´n Petrolera en las Cuencas Subandinas. Villamil, T., 1999, Campanian – Miocene tectonostratigraphy, depocenter evolution and basin development of Colombia and western Venezuela: Paleogeography, Paleoclimatology, Paleoecology, v. 153, p. 239 – 275. Wheeler, H. O., 1964, Baselevel, lithosphere surface, and time-stratigraphy: Geological Society of America Bulletin, v. 75, p. 599 – 610.