Characterising The Blasting

Characterising the Blasting Properties of Iron Ore A Scott1 and I Onederra2 ABSTRACT Most iron ores occur in mine benches as complex mixtures of lithologies and ore types. This significantly complicates their characterisation for both blast design and fragmentation modelling. A great deal of technical and research effort has been applied to the development of models to predict fragmentation from rock blasting. A number of quite useful engineering models are currently available to blasting engineers and consultants, and more sophisticated approaches using mechanistic descriptions of rock breakage continue to be developed and applied by researchers and explosives companies. The engineering or empirical models that are in more general use within the industry, such as the Kuz-Ram model, require appropriate data if they are to generate useful fragmentation predictions. Some of these models have reached significant sophistication and have made very useful contributions to both routine blasting operations and the exploration of new blasting paradigms. However, these models struggle to adequately deal with rock that is as complex as iron ore. This paper reviews the characteristics of the most successful blast fragmentation models available to the industry and discusses approaches to the derivation of suitable rock mass properties to adequately describe blast fragmentation in a range of iron ores.

INTRODUCTION It is obvious to argue that in order to design an effective blast, it is helpful to be able to predict how it will perform in the field. However, such predictions have always been, and to some extent still remain, fairly imprecise. The aspects of blast performance that are of most interest to an open pit mine include fragmentation, swell, muck pile movement and damage to adjacent benches. Rules of thumb, nomograms, empirical design rules and numerical models have been developed and used over the years to address each of these blasting outcomes. However, the design of most mine production blasts generally relies on some relatively simple geometric relationships guided by observations from the field – essentially design rules that have been honed by ‘trial and error’. This paper addresses one of these blast outcomes in a particular type of mine – the prediction of blast fragmentation in open pit iron ore mines. Fragmentation in these operations is of interest in order to: •• avoid boulders or oversized fragments that reduce the efficiency of excavation and require secondary breakage •• influence the proportion of fines in the run-of-mine (ROM) ore, which may affect the market into which the ore is sold •• affect the distribution of ROM particle sizes, which may influence the design and efficiency of ore processing operations. An understanding of the range of expected fragmentation outcomes is important to the design of ore handling and process plants prior to the commencement of operations as is the management of fragmentation performance throughout

the mine’s operating life. The prediction of blast fragmentation depends on the use of a model that adequately responds to the breakage processes and outcomes that are of importance to the mining operation and an adequate description of the relevant rock mass properties.

BLAST DESIGN Local experience A mining operation quickly develops a history of performance that can be used to guide future designs. This can be particularly effective in consistent geological environments where surprises are minimal. However, this process is always ‘chasing its tail’ when conditions change or the required blasting outcomes need to be varied. Industry experience can be of some assistance when designing for a greenfield development, but this experience is unlikely to apply to the detailed characteristics of the ores found in a new mining area.

Traditional design rules The literature abounds with texts providing blast design guidelines and rules. Although sometimes contradictory because of the diverse background of the authors, general trends are evident in the design rules available. These guidelines can be used to relate the burden to the blasthole diameter, the burden to blasthole spacing via a spacing to burden ratio, stemming lengths and subdrill lengths in terms of multiples of the hole diameter and even the size of the

1. FAusIMM(CP), Consulting Mining Engineer, Scott Mine Consulting Services Pty Ltd, PO Box 5126, Kenmore Qld 4069. Email: [email protected] 2. Senior Lecturer, School of Mechanical and Mining Engineering, University of Queensland, St Lucia Qld 4072. Email: [email protected] IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

481

A SCOTT AND I ONEDERRA

granular material used to stem the holes. Figure 1 shows the style of nomograph published by Langefors and Khilstrom (1978), which was a convenient way to capture these sorts of relationships. Many of these design rules have been refined and focused over the years and have been made available in industry handbooks and courses, conference papers, textbooks and advice from consultants. A problem with these design rules is that they offer advice on how to develop blast designs that would generate ‘satisfactory’ or ‘good’ fragmentation when blasting the type of rock in which the author had experience, but not specific or tailored blasting outcomes. These design rules are able to provide guidance to overcome common blasting problems (eg avoiding excessive oversize by improving the energy distribution within the blast) or influence the fragmentation outcome (eg using higher velocity of detonation explosive to increase the proportion of fines in the muck). However, they are still not able to predict what the particle size distribution of the blasted muck is likely to be when a particular blast design is applied to a particular rock mass. In fact, the rock and its properties are provided minimal influence in most blast design guidelines.

FIG 1 – Example of a blast design nomograph.

FRAGMENTATION MODELS Development paths Scott, Chitombo and Kleine (1993) provide a useful background to the status of fragmentation modelling 20 years ago. This paper reviewed the fragmentation models available at the time and suggested a path for their future development. The development of models to predict fragmentation from blasting has since followed the two distinct paths they identified: 1. an empirical or engineering approach that captures relationships between known rock mass properties, blast designs and the subsequent blasting results 2. a mechanistic or fundamental approach that focuses on the underlying physics of the detonation behaviour of explosives and rock breakage.

Empirical or engineering approaches As summarised in Figure 2, steady progress is evident in the development of empirical fragmentation models. The starting point can be traced back to Rosin and Rammler (1933), who derived the original function adopted by both Kuznetsov (1973) and Cunningham (1983) to model blast fragmentation. Bond (1959) also contributed by proposing ways to predict the mean fragment size, while Johnson (1962) developed an empirical relationship based on simple crater tests. The introduction of parameters that linked blast geometry and charging configurations to fragmentation and resulted in the Kuz-Ram model (Cunningham, 1983, 1987, 2005) has been the mainstay of practical blast fragmentation modelling for the past 30 years. The success of the Kuz-Ram model when compared with most other approaches is due to its use of most of the basic blast design parameters in a form that can be reasonably quantified or estimated from field data, the fact that it can be executed on a calculator or Excel spreadsheet and that it can generate reasonable results for simple blasting situations.

FIG 2 – Chronology of developments in empirical fragmentation modelling. 482

IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

CHARACTERISING THE BLASTING PROPERTIES OF IRON ORE

Further modifications to the Kuz-Ram model were intro­ duced in the late 1990s, when developments in fragmentation modelling saw the introduction of two component modelling approaches. This allowed for improvements in the prediction of fine fragmentation (Scott et al, 1998; Kanchibotla, Valery and Morrell, 1999; Djordjevic, 1999; Thornton, Kanchibotla, and Brunton, 2001). Over the years, two component approaches have also been improved through the introduction of more accurate ways of predicting the potential volume of crushed material resulting from the crushing and shearing stages of blasting (Onederra, 2004). These have also incorporated the Swebrec function proposed by Ouchterlony et al (2006). Developments in the prediction of very fine size fractions (to one micron) have also been documented by Scott, Michaux and Onederra (2009).

determination of fragmentation modelling parameters such as mean fragment size, fines cut off points and fragmentation uniformity indices. These modelling parameters are determined for any given combination of pattern geometry, explosive charging arrangement and rock mass conditions. As shown in Figure 3, the extent of breakage (defined by both crushing and fracturing zones) is discretely estimated on predefined layers along the length of a blasthole. The number of layers per blasthole can be defined by the user and depends on the degree of variability associated with both the rock mass and explosive charging condition.

Incorporating rock mass variability

Several attempts have been made to simulate the dynamic fracturing process, fragment formation and displacement in blasting. The numerical methods currently used are based on finite element, finite difference and discrete element analysis techniques. They have been applied in isolation and in combined forms. In most cases, shock wave propagation and the fracturing process have been modelled with finite element and finite difference codes. Discrete element codes have mainly been applied to model particle motion or rock movement problems. In all three types of numerical techniques, the geometry of the rock mass is divided into a number of small elements or zones using a process referred to as ‘discretisation’. The behaviour of each element/zone in the model is governed by a ‘constitutive law’. If the constitutive law is appropriate and all of the relevant mechanisms are represented, the model is expected to behave in the same way as the material being represented. The difference between the three numerical approaches is in the treatment of the elements and how the behaviour of each is ‘summed’ to represent the problem as a whole.

Notwithstanding the successful implementation of empirical fragmentation models, their common constraint is that they are unable to explicitly consider changes in rock mass conditions and charging configurations. Current models assume homogenous and isotropic conditions, with only mean values used as input parameters. As an attempt to consider the inherent variability of rock mass input data, stochastic techniques such as those documented by Thornton, Kanchibotla and Esterle (2001) and Onederra, Mardones and Scherpenisse (2010) have been successfully applied. However, these techniques do not consider the influence of actual rock mass variability on final breakage and fragmentation outcomes. As they are built upon current empirical models, the stochastic approaches still do not take into account the threedimensional distribution of explosive charges typical of nonuniform patterns and charging geometries such as stemming length variations and decking. As a way of addressing these limitations, a layered fragmentation modelling approach has been developed and is currently undergoing evaluation and validation. This newly developed model is called iFrag and is an improved version of a methodology first proposed by Onederra (2004) that has also been successfully incorporated into a fragmentation model for underground ring blasting applications. The iFrag model is based on a combination of wellestablished and newly derived empirical rules that allow the

It is likely that this model, or models of its type, will find useful application within the industry in future years.

Mechanistic or fundamental approaches

Sophisticated finite element/discrete element hybrid codes such as ELFEN, commercially developed and marketed by Rockfield Software Ltd, can perform both two- and threedimensional analysis of stress, fracturing and in-flight block interaction during the blasting process. The code’s capabilities were improved with the integration of the MBM2D model developed by Munjiza and Owen (1992), Owen, Munjiza and

FIG 3 – Estimated breakage envelopes within the iFrag modelling framework. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

483

A SCOTT AND I ONEDERRA

Bicanic (1992) and Minchinton and Lynch (1996). In MBM2D, the detonation process is modelled using a non-ideal detonation code called CPeX, which provides the velocity of detonation and borehole pressure history as inputs to the MBM2D model.

and the parameters required by the algorithms involved are complex. The data driving the thermodynamic codes and dynamic response of the rock substance in both their sonic and supersonic states require sophisticated laboratory testing and analyses.

A discrete element code that has been used to model the combined mechanical effect of stress waves and highpressure gas flow is the particle flow code PFC3D developed by Itasca Inc (Potyondy, Cundall and Sarracino, 1996). In this technique, a model is composed of distinct particles that displace independently from one another and interact only at contacts. PFC3D allows finite displacements and rotations of discrete bodies (including complete detachments) and recognises new contacts automatically as the calculation progresses. Results from these attempts support the view that particle flow codes can be successfully implemented in an advanced blasting model.

It is simply not yet practical to obtain these data for the complex environment of an operating mine. Simpler approaches are required.

In general, the success of earlier models indicated that an improved code that could simulate detonation, rock fracture and finally displacement must incorporate fundamental principles of detonation theory (ideal and non-ideal), stress wave propagation and interaction in a jointed rock mass, gas flow, the ability to model continuum and discontinuum behavior, and the ability to explicitly consider the effect of different boundary conditions. This has been achieved within the Hybrid Stress Blasting Model (HSBM), for which final validation and commissioning is currently being conducted. The HSBM can be described as a sophisticated blast modelling research tool. The code has been under development for the last 11 years as part of an international collaborative research project funded by a consortium of companies comprising explosive and equipment suppliers and major mining houses. Over the course of its development, several improvements and modifications have been made to both the detonation and geomechanical modelling components in order to improve calculation speed and the size of the problem that can be modelled. A detailed description of the framework and validation work is given by Furtney, Cundall and Chitombo (2009) and Onederra et al (2013). Codes such as the HSBM are still limited to research applications because they are computationally demanding and require specialist input for calibration. Today’s practical problems have to be tackled using simpler approaches.

THE BLASTING PROPERTIES OF A ROCK MASS Concept It seems trite to suggest that the fragmentation expected from any blast will be heavily dependent on the properties of the rock mass to be blasted. However, most empirical models rely on a simple ‘rock factor’ to describe the influence of the rock mass properties on the blasting outcome. Previous, current and emerging mechanistic models need to use explicit rock mass properties that require sophisticated laboratory measurements and detailed rock mass structural data if realistic outcomes are to be generated. The industry has been caught between excessively simple rock mass descriptions for the empirical models and excessively complex data requirements for the mechanistic models.

Data required by mechanistic models The detailed mechanistic models attempt to honour the actual physics of the dynamic loading applied to a structured rock mass and the response of the rock mass in terms of its subsequent breakage and movement. The detailed behaviour of the explosive depends on the response of the rock mass 484

Empirical approaches A detailed discussion about historical approaches to quantify the blasting characteristics of a rock mass is provided by JKMRC (1996a) and is still relevant today. Most early approaches sought to identify a rock mass ‘factor’ or ‘coefficient’ that could be inserted into a general formula to generate the required blast design parameter or fragmentation outcome. An example of this is the rock factor in the KuzRam model, which was based on a general assessment of the competence of the rock mass to be blasted. In the Kuz-Ram model, a rock factor of seven represented medium rocks, ten represented hard, fissured rocks and 13 represented very hard, weakly fissured rocks. A more systematic approach was provided by Lilly (1986), who introduced a number of rock mass parameters to generate a rock factor for Cunningham’s Kuz-Ram model to more appropriately describe the blasting properties of iron ore. Lilly’s approach has proven to be very practical, relying on the field assessment of rock mass structure (massive, blocky or friable), joint plane spacing, joint plane orientation relative to the blast volume and the density and strength of the rock. Estimates of these parameters were combined by a formula to generate a single ‘rock factor’ for use in both blast design (related to powder factor) and the Kuz-Ram fragmentation model. A number of variants of Lilly’s index have been generated by various workers. Perhaps the most useful form of Lilly’s index uses the approach adopted by Bickers et al (2001) shown in Figure 4. The rock mass description follows the geological strength index approach developed by Hoek, Marinos and Benissi (1998). In this application, the blastability index (BI), or rock factor (RF), has been modified by combining the original rock mass description and joint plane spacing parameter into a single rating. The joint plane orientation, rock density influence and rock strength parameters remain as documented by Lilly. The value of Lilly’s approach is that it incorporates many of the properties that are important to blasting performance, including rock substance strength, rock mass structure and density. The form of the index also suits the assessment of these parameters in the field because each of them is the result of a field-based estimate rather than a refined measurement.

A successful empirical approach The following describes the application of an advanced empirical fragmentation model and the characterisation of the rock mass upon which it relies. The original Kuz-Ram model had the following shortcomings: •• it relied on a simplistic description of the rock mass (represented as a qualitative ‘rock factor’) •• it tended to overestimate the coarse end of the resulting size distribution because it did not adequately respond to the structure of the rock mass •• it tended to significantly underestimate the fine end of the resulting size distribution because it was forced to fit a Rosin-Rammler distribution curve. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

CHARACTERISING THE BLASTING PROPERTIES OF IRON ORE

FIG 4 – Modified blastability index based on the geological strength index approach (Bickers et al, 2001). The original concept behind the Kuz-Ram model has been significantly extended to develop a model that: •• uses a Kuz-Ram-like algorithm to estimate the mean particle size •• uses a modified RF (of the style proposed by Lilly (1986)) based on data typically available from drill core rather than face mapping •• uses an estimate of the in situ block size distribution for the rock mass to influence and limit the coarse end of the fragmentation curve •• uses a form of crushed zone model and the in situ block size distribution of the rock mass to influence the fine end of the fragmentation curve •• generates a final size distribution using a Swebrec function (Ouchterlony et al, 2006), which has been demonstrated to provide a much better fit to actual rock fragment size distributions from blasting than the popular alternatives •• responds to the basic explosive properties (density, effective energy and velocity of detonation), explosive distribution, subdrill and stemming lengths. The resulting model has been applied to a wide range of practical blasting situations and has been relied upon in a large number of feasibility and field studies. A rock factor (RF) is used to drive the mean fragment size calculation and guide the explosive energy requirements for blast design. RF is estimated based on a combination of three basic influences: rock strength, rock structure and rock density. The individual factors are estimated using the relationships shown in Figures 5, 6 and 7. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

FIG 5 – Relationship between unconfined compressive strength and ‘strength factor’.

FIG 6 – Relationship between density and ‘density factor’. 485

A SCOTT AND I ONEDERRA

Australia’s large-scale iron ore industry was initially focused on mining the Premium Brockman ores of Mt Tom Price and Mt Whaleback (O’Brien, 2009). Reserves of these premium ores are limited, and more recent developments have focused on other hematite-goethite, Marra Mamba and channel iron deposits. There are also very large magnetite resources currently under evaluation, but they are yet to reach large scales of production.

Blasting characteristics of iron ores FIG 7 – Relationship between fracture frequency and ‘structure factor’.

The blasting characteristics of Australian iron ores vary considerably. Table 1 provides a summary of some indicative blasting properties for a number of iron ore types.

The RF used by the model to estimate the mean fragment size is estimated from:

TABLE 1 Typical blasting properties for a range of iron ore types.

RF = 11.5 * (strengthfactor * structurefactor *densityfactor) - 1

(1)

The RF is also used to estimate a nominal energy factor that can be used to guide initial blast designs for the rock mass being considered. Energy factor is the powder factor resulting from a blast design modified by the effective energy allocated to the explosive being used relative to ammonium nitrate/ fuel oil (ANFO) (JKMRC, 1996a). An initial target energy factor can be calculated from:

Type

Strength (MPa)

Fracture frequency

Density (t/bcm)

Rock factor

Magnetite

180

2

3.5

10.1

Massive hematite

150

2

3.4

9.2

Blocky hematite

130

4

3.2

7.9

Banded iron

110

10

3

6.5

Hematite/goethite

70

5

3.1

5.4

Goethite/limonite

25

20

2.9

1.9

EF = 0.295 e(0.1222 * RF) (2) where: EF is the energy factor in kilograms per bank cubic metre RF is the rock factor defined previously Examples of the RFs for different iron ore types are provided in the next section. In addition to the basic RF parameter influencing the mean fragment size resulting from a blast, the model uses the fracture frequency data to influence the coarse end of the fragment size distribution curve. Higher fracture frequency indicates a more fractured rock mass, which will generate smaller fragments at the coarse end of the fragmentation curve. The rock strength influences the extent of fines expected from the blast using a near-field crushing model of the form used by Kanchibotla, Valery and Morrell (1999). Weaker rocks will suffer finer fragmentation closer to each charge than stronger rocks, and this crushing effect near each blasthole is added to the general distribution of fragment sizes generated by the model.

IRON ORE Why is it different? Most blasting research, and hence the aforementioned fragmentation models, have focused on hard rocks like those encountered in quarries and base and precious metal mines. Some approaches have been extended to adequately deal with weaker materials common in coal environments, but, other than Lilly’s original blasting index, little work has been directly focused on the fragmentation of iron ore. Australian iron ores can be very complex in a physical sense. They can vary significantly in terms of their strength, structure and density – the three principal characteristics that influence blast fragmentation. The impact of this variability is amplified by the fact that it can occur within a single bench or blast volume. 486

Modelling single species Armed with the parameters shown in Table 1, it is a simple matter to model the fragmentation expected from blasting each of these materials. Table 2 shows examples of each of these materials and the fragmentation predicted when they are blasted using the same blast design. The blast design was based on a 12 m high bench and utilised a pattern of 251 mm diameter blastholes on a 7.0 × 7.9 m pattern charged with a heavy ANFO explosive with a density of 1.15 g/cc. This design results in an average powder factor of 0.73 kg/bcm. The number of boulders (rocks greater than 1 m in dimension) expected per truckload is a convenient way to visualise the coarse end of the fragment size distribution. The 80 per cent and 50 per cent passing sizes are often used to guide excavation models, and the proportion of fines less than 15 mm is often considered to be a useful parameter for subsequent material handling and ore treatment.

Modelling mixtures of ores The results from such modelling is often disappointing, usually involving an underestimate of the fine end of the blasted fragment size distribution. The reason for this is that it is quite uncommon to find a single species with consistent blasting properties in a mining face, let alone in an entire blasting block. The problem lies in attempting to provide the model with the properties of a mixture of rock mass components by treating them as a single species. Magnetite and some of the traditional massive hematites can present essentially as a single species within a blast volume. However, the majority of ores currently being mined are more like the blocky hematite, hematite/goethite and goethite/limonite examples in Table 2. These clearly present the blaster with a mixture of materials. In these cases, it is useful to consider the mixtures as consisting of three components: •• a host matrix •• a hard component •• a component of fines. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

CHARACTERISING THE BLASTING PROPERTIES OF IRON ORE

TABLE 2 Example of different types of iron ores and their relative blasting performance. Ore type

Appearance

Boulders/truck

80% passing (mm)

50% passing (mm)

Fines <15 mm (%)

Magnetite

6.8

761

275

9.5

Massive hematite

6.4

728

247

10.8

Blocky hematite

2.5

573

213

11.5

Banded iron

0

409

178

12.0

Hematite/goethite

1.3

460

174

17.4

Goethite/limonite

0

193

53

32.4

Most fragmentation models rely on the properties of the host matrix material and consequently ignore the contribution of the harder component or fine component. The resulting analyses are likely to describe the breakage behaviour of this matrix material reasonably, but ignore the contribution of the harder components, which will behave quite differently during blasting than the matrix, and also ignore the fine material that is unlikely to suffer significant further breakage during blasting. Table 3 summarises the results from modelling blast fragmentation for the ‘mixture ores’ in Table 1 using this composite approach. The blast design and fragmentation model are the same, but the rock mass has been characterised in greater detail. The predicted fragment size distribution in the resulting muck piles is generally finer for the multicomponent model because the materials that contribute to the generation of fines have been represented more realistically. A feature of these types of ores is that the components (the matrix, the harder components and the fines) tend to be consistent in character, but vary in proportion. Therefore, the next blast is likely to encounter a different combination of essentially the same materials. This means that once the fragmentation characteristics of the component materials have been modelled, it should be possible to estimate the fragmentation from the whole blast by calculating the weighted average or the per cent passing each size based on the proportion of each component. If additional components can be identified, then these can also be treated in the same way as the principal components shown in Table 1. The challenge is to then establish the blasting characteristics of each component of the rock mass and their relative abundance in any given blast volume. Dealing with the ore as a mixture of its significant physical components enables the range of physical properties relevant to blasting to be provided to the model rather than an approximation of these properties. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

TABLE 3 Fragmentation results for components and mixtures. Blocky hematite

Matrix

Hards

Fines

Proportion (%)

47.5

47.5

5

Density

3.0

3.2

Strength

40

120

Fracture frequency

5

3

Run-of-mine size

Boulders

X80 (mm) X50 (mm)

Fines (%)

Previous analysis

2.5

573

213

12

Component model

2.5

470

130

23

Matrix

Hards

Fines 10

Hematite/goethite Proportion (%)

30

60

Density

2.9

3.0

Strength

20

60

Fracture frequency

10

5

Run-of-mine size

Boulders

X80 (mm) X50 (mm)

Fines (%)

Previous analysis

1.3

460

174

17

Component model

0.6

360

74

33

Matrix

Hards

Fines

Proportion (%)

52.2

32.5

15

Density

3.0

3.2

Strength

20

30

Goethite/limonite

Fracture frequency

Run-of-mine size

20

Boulders

10

X80 (mm) X50 (mm)

Fines (%)

Previous analysis

0

193

53

32

Component model

0

190

30

44 487

A SCOTT AND I ONEDERRA

SOURCES OF DATA Drill core For a greenfield site, there is little choice other than to rely on drill core for the characterisation of the material to be blasted. The fragmentation model used here has been designed to utilise the data available from drill core. The basic data requirements can be met from the following sources.

Strength measurements •• Samples of core can be prepared in a laboratory for unconfined compressive strength (UCS) testing. •• Samples of core can be prepared in a core shed for point load strength (PLS) testing: •• As PLS tests are simple to undertake, many tests can be made. This allows an understanding of the variability of the rock strength, and any anisotropy can be identified by testing specimens along the axis and across the diameter of core samples. •• PLS is usually multiplied by a factor of 20–24 to estimate the equivalent UCS. •• Core samples can be used to measure sonic parameters (P and S wave velocities), which can be related to rock competence. •• Core samples may be used to measure energy-breakage relationships to guide the design of crushing facilities. The drop weight crushing indices (A*b) (JKMRC, 1996b) can also be related to static strength (Scott, Morrell and Clark, 2002). •• Most logging formats provide for the assessment of a ‘field strength index’, based on the response of the sample to being scratched or struck by a geologist’s pick. While indicative only, such indices can be used effectively to identify strong and weak zones within the core.

Fracture frequency •• While some care is required to distinguish between natural fractures within the rock mass and breakage resulting from drilling and handling the core, the number of natural or in situ fractures observed over any length of core would be recorded by most logging schemes. It is then a simple matter to estimate the number of fractures per metre of core. •• All core should be photographed, and it is not difficult to get a reasonable estimate of fracture frequency from an inspection of core photographs. •• The relationship proposed by White (1977) allows the in situ block size to be estimated based on fracture frequency. •• Most logging procedures include the calculation of rock quality designation (RQD). This is simply the percentage of the core that is intact in lengths of 100 mm or longer. While higher values of RQD are likely to refer to better quality rock, the same RQD value can be correctly ascribed to rocks with a wide range of fracture characteristics. Nevertheless, relationships between RQD and fracture frequency can be useful (Palmstrom, 2005).

Density •• Samples of core can be subjected to laboratory density measurements. •• As some iron ores can be porous, it is useful to obtain both coated and uncoated density results so that the true substance density is known together with an estimate of the rock’s porosity. 488

Face samples and mapping In an operating mine, there should be frequent access available to bench faces. While safety requirements limit human access to this environment, there are a number of laser and photographic tools available that allow detailed mapping to be undertaken in an office environment. Coupled with geological interpretation and supplemented by strength and density data from hand samples collected when access is available, the basic characterisation requirements for fragmentation modelling can be met.

Logging drill cuttings Although drill cuttings are not available until the blasthole pattern has been decided and drilled, they do provide the opportunity to interpret the range of rock types present in the blast volume and their approximate location. These techniques have been routine in most operating mines to characterise the ore types, grade and contaminants present in the blast block. This work has been extended by some operations to capture data to influence the explosive charge design and estimate the resulting blast fragmentation.

Blasthole drill performance The concept of monitoring the performance of the blasthole drill and interpreting the relative strength of the rock to be blasted is not a new idea. The concept faced many technical challenges in the late 20th century, but with the development of accurate GPS systems, drill control and automation systems and communication and data management tools, there is no technical barrier to this concept being realised. Directly linking the resulting rock characterisation data to a ‘smart’ explosives truck is part of this emerging process. While drilling data can readily provide relative strength information, it struggles to reliably generate structural information. The real value of having access to drilling data is to track changes within the rock mass so that base characteristics determined from more fundamental measurements can be applied to the variable rock mass as a whole. In its simplest form, just tracking the location and thickness of hard or soft zones across a drill pattern is of value provided that the basic blasting properties of these zones and the parent rock mass have been determined from other sources.

Geophysics Downhole geophysical logs can provide a basic description of rock competence in an iron ore environment. Density, magnetic susceptibility and natural gamma have been used to distinguish a range of ore types. However, the situation is similar to the interpretation of drilling performance in that the basic geophysical and physical characteristics of the materials present need to be determined based on other measurements. The distribution of these materials within the blast volume can then be interpreted based on the downhole geophysical response.

Summary The characterisation of the fundamental blasting properties of a rock type is best achieved through physical sampling and testing, logging and mapping. This can be achieved from core or face mapping and sampling. Remote and potentially automated characterisation techniques, such as drill monitoring and downhole geophysics, can then provide valuable information about the distribution of the identified rock types within a blast volume. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

CHARACTERISING THE BLASTING PROPERTIES OF IRON ORE

AN EXAMPLE OF MODELLING THE FRAGMENTATION OF A MIXTURE OF ORE TYPES Core from a 12 m zone within a bedded iron deposit is shown in Figure 8. Core logging and testing of samples of these materials generated the blasting properties shown in Table 4. The core had been characterised as either being ‘massive’, ‘blocky’, ‘broken’ or ‘unconsolidated’. A visual assessment rated this 12 m interval to contain 14 per cent ‘massive’, 58 per cent ‘blocky’ and 28 per cent ‘broken’ ore. The blasting properties shown in Table 4 have been used to model the fragmentation expected when ore with the properties of each of these characteristics is blasted using a suitable blast design. In this instance, the design involved 251 mm diameter blastholes on a 7.7 × 8.9 m pattern charged with ANFO. This design results in an overall powder factor of 0.42 kg/bcm. When the ROM fragment size distributions for each of these components are weight averaged, the size distribution shown in Figure 9 results.

Having characterised the blasting properties of each of the component ore types and modelled the fragmentation expected for any given blast design, the size distribution expected from blasting any combination of these ore types can be easily calculated. Data from drill monitoring, analysis of drill cuttings, geophysical logging or face mapping can then be used to allocate the proportion of each of these types to any given blast bench to estimate the fragmentation expected.

CONCLUSION A great deal of technical and research effort has been applied to the development of models to predict fragmentation from rock blasting. A number of quite useful engineering models are currently available to blasting engineers and consultants, and more sophisticated approaches using mechanistic descriptions of rock breakage continue to be developed and applied by researchers and explosives companies. The mechanistic models struggle to adequately represent the rock mass properties that genuinely control rock breakage under the action of an explosive charge. The explosive–rock interaction that controls the velocity of detonation, dynamic breakage behaviour of the rock substance under super and subsonic loading conditions and the distribution and behaviour of the in situ rock mass defects make the provision of these explicit parameters for a detailed breakage model highly challenging in even simple blasting situations. Even the much simpler engineering models that are in more general use within the industry, such as the KuzRam model, require appropriate data if they are to generate useful fragmentation predictions. Some of these models have reached significant sophistication and have made very useful contributions to both routine blasting operations and the exploration of new blasting paradigms. However, these models struggle to adequately deal with rock as complex as iron ore.

FIG 8 – Core from a 12 m zone within a bedded iron deposit. TABLE 4 Blasting properties of the ore types in Figure 8. Form

Weathering

Strength (MPa)

Fracture frequency

Density (t/bcm)

Massive

Fresh

12

2

2.67

Weathered

8

2

2.67

Fresh

8

5

2.67

Weathered

4

5

2.67

Broken

4

15

2.67

Unconsolidated

4

150

2.67

Blocky

FIG 9 – Weighted run-of-mine size distribution for a 12 m bench in fresh bedded iron. IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015

Most iron ores in mine benches occur as complex mixtures of lithologies and ore types. This significantly complicates their characterisation for both blast design and fragmentation modelling. It is commonly observed that the properties of each of these components tend to be fairly consistent within a deposit, but that the proportion of each of them in a given blast bench can vary significantly. In order to generate a meaningful estimate of fragmentation from the blasting of a mine bench containing a number of component ore types, it is necessary to use a model that responds to the required breakage mechanisms and provide data that allows the model to respond appropriately. The complexity of most iron ores is such that this process needs to be undertaken for each ore component individually and the results combined to represent the outcome from the actual bench blast. Suitable information for this process can be obtained from drill core when working on greenfield projects. Field samples can be used to determine the blasting characteristics of the materials encountered in an operating mine. Routine data collection from the analysis of cuttings, drill performance, face mapping or geophysics can be used to map the distribution of the component ore types throughout the blast volume, enabling blast design and fragmentation modelling to be undertaken. It is not difficult to envisage automated systems being configured to achieve this and the results used to determine the detailed charge and initiation designs for blasts to meet targeted performance characteristics. Although they will continue to improve and develop with time, suitable empirical blasting models exist to meet these requirements. The development of detailed mechanistic models is well 489

A SCOTT AND I ONEDERRA

advanced, but the provision of suitable data to enable them to generate useful outcomes in practical blasting situations remains a challenge.

REFERENCES Bickers, C F, Dunbar, C T, LeJuje, G E and Walker, P A, 2001. Wall control blasting practices at BHP Billiton Iron Ore Mt Whaleback, in Proceedings EXPLO 2001, pp 93–102 (The Australasian Institute of Mining and Metallurgy: Melbourne). Bond, F, 1959. The work index in blasting, Quarterly of the Colorado School of Mines, 54(3):77–82. Cunningham, C V B, 1983. The Kuz-Ram model for prediction or fragmentation from blasting, in Proceedings First International Symposium on Rock Fragmentation by Blasting, pp 439–453 (Lulea University of Technology: Lulea). Cunningham, C V B, 1987. Fragmentation estimations and the Kuz-Ram model – four years on, in Proceedings 2nd International Symposium on Rock Fragmentation by Blasting, pp 475–487 (Society for Experimental Mechanics: Bethel). Cunningham, C V B, 2005. The Kuz-Ram fragmentation model – 20 years on, in Proceedings EFEE World Blasting Conference, pp 201– 210 (European Federation of Explosives Engineers: London). Djordjevic, N, 1999. Two-component of blast fragmentation, in Proceedings 6th International Symposium on Rock Fragmentation by Blasting, pp 213–219 (South African Institute of Mining and Metallurgy: Johannesburg). Furtney, J K, Cundall, P A and Chitombo, G P, 2009. Developments in numerical modeling of blast induced rock fragmentation: updates from the HSBM project, in Proceedings 9th International Symposium on Rock Fragmentation by Blasting, pp 335–342 (Taylor & Francis Group: Abingdon). Hoek, E, Marinos, P and Benissi, M, 1998. Applicability of the geological strength index (GSI) classification for very weak and sheared rock masses. The case of the Athens Schist Formation, Bulletin of Engineering Geology and the Environment, 57(2):151–160 Johnson, J B, 1962. Small-scale blasting in mortar, 22 p, volume 6012, US Department of the Interior, Bureau of Mines R1 6012. Julius Kruttschnitt Mineral Research Centre (JKMRC), 1996a. Open Pit Blast Design – Analysis and Optimisation (ed: A Scott), pp 112–145 (University of Queensland: Brisbane). Julius Kruttschnitt Mineral Research Centre (JKMRC), 1996b. Mineral Comminution Circuits – Their Operation and Optimisation (ed: T Napier-Munn), pp 49–94 (University of Queensland: Brisbane). Kanchibotla, S S, Valery, W and Morrell, S, 1999. Modelling fines in blast fragmentation and its impact on crushing and grinding, in Proceedings EXPLO 99, pp 137–144 (The Australasian Institute of Mining and Metallurgy: Melbourne). Kuznetsov, V, 1973. The mean diameter of fragments formed by blasting rock, Soviet Mining Science, 9(2):144–148. Langefors, U and Khistrom, B, 1978. The Modern Technique of Rock Blasting, third edition, 405 p (John Wiley and Sons: New York). Lilly, P, 1986. An empirical method of assessing rock mass blastability, in Proceedings Large Open Pit Mining Conference, pp 41–44, 89–92 (The Australasian Institute of Mining and Metallurgy: Melbourne). Minchinton, A and Lynch, P M, 1996. Fragmentation and heave modelling using a coupled discrete element gas flow code, in Proceedings 5th International Symposium on Rock Fragmentation by Blasting, pp 71–80 (Balkema: Rotterdam).

490

Munjiza, A and Owen, R J, 1992. Discrete element models of rock blasting, Engineering Modelling, 5:65–73. O’Brien, R, 2009. Australia’s iron ore product quality [online], Geoscience Australia. Available from: . Onederra, I, 2004. A fragmentation modelling framework for underground ring blasting applications, Fragblast: International Journal of Blasting and Fragmentation, 8:177–200. Onederra, I, Furtney, J, Sellers, E and Iverson, S, 2013. Modelling blast induced damage from a fully coupled explosive charge, International Journal of Rock Mechanics and Mining Sciences, 58:73–84. Onederra, I, Mardones, F and Scherpenisse, C, 2010. Application of stochastic approach to blast fragmentation modelling, Mining Technology, 119:221–232. Ouchterlony, F, Olsson, M, Nyberg, U, Andersson, P and Gustavsson, L, 2006. Constructing the fragment size distribution of a bench blasting round, using the new Swebrec function, in Proceedings 8th International Symposium on Rock Fragmentation by Blasting, pp 332–344 (Taylor & Francis Group: Abingdon). Owen, D R J, Munjiza, A and Bicanic, N, 1992. A finite element – dicrete element approach to the simulation of rock blasting problems, in Proceedings 11th Symposium on Finite Element Methods in South Africa, pp 39–58 (FRD/UCT Centre for Research in Computational and Applied Mathematics: Rondebosch). Palmstrom, A, 2005. Measurements of and Correlations between block size and rock quality designation (RQD). Tunnels and Underground Space Technology, 20:362–377. Potyondy, D O, Cundall, P A and Sarracino, R S, 1996. Modelling of shock and gas-driven fractures induced by a blast using bonded assemblies of spherical particles, in Proceedings 5th International Symposium on Rock Fragmentation by Blasting, pp 55–62 (Balkema: Rotterdam). Rosin, R and Rammler, E, 1933. Laws governing fineness of powdered coal, Journal of the Institute of Fuel, 7:29–36. Scott A, Chitombo, G and Kleine, T, 1993. The challenge of the prediction and control of fragmentation in mining, in Proceedings 4th International Symposium on Rock Fragmentation by Blasting, pp 507–517 (Balkema: Rotterdam). Scott, A, David, D, Alvarez, O and Veloso, L, 1998. Managing fines generation in the blasting and crushing operations at Cerro Colorado Mine, in Proceedings Mine to Mill Conference, pp 141– 148 (The Australasian Institute of Mining and Metallurgy: Melbourne). Scott, A, Michaux, S and Onederra, I, 2009. Characterising dust generation from blasting, in Proceedings 9th International Symposium on Rock Fragmentation by Blasting, pp 663–671 (Taylor & Francis Group: Rotterdam). Scott, A, Morrell, S and Clark, D, 2002. Tracking and quantifying value – from ‘mine to mill’ improvement, in Proceedings Value Tracking Symposium, pp 77–84 (The Australasian Institute of Mining and Metallurgy: Melbourne). Thornton, D, Kanchibotla, S and Brunton, I, 2001. Modelling the impact of rockmass and blast design variation on blast fragmentation, in Proceedings EXPLO 01, pp 331–345 (The Australasian Institute of Mining and Metallurgy: Melbourne). Thornton, D M, Kanchibotla, S S and Esterle, J S, 2001. A fragmentation model to estimate ROM size distribution of soft rock types, in Proceedings 27th Annual Conference on Explosives and Blasting Techniques, pp 41–53 (International Society of Explosives Engineers: Cleveland). White, D, 1977. Predicting fragmentation characteristics of a block caving orebody, MSc thesis (unpublished), University of Arizona, Tucson.

IRON ORE CONFERENCE / PERTH, WA, 13–15 JULY 2015