Publications

This paper describes developments to the JKMRC AG/SAG mill model in which correlations, derived from a large database of pilot and full-scale milling tests, relate model grinding rates to mill operating conditions and feed size to ore breakage characteristics. These correlations allow mill performance to be predicted across a wide range of mill sizes and operating conditions, enabling circuit evaluation at the pre-feasibility stage and reducing reliance on costly pilot testing. Predictions are validated against measured performance, including power draw, for a range of AG/SAG mills and product sizes.

This paper reviews the state of HPGR equipment, circuit design and operation with reference to Australian practice, drawing on the four operations then using the technology: Argyle, Bendigo, OneSteel and Boddington. It examines twenty years of development and discusses practical issues including the effect of moisture, voidage and particle size on roll wear, the difficulties of wet and fine screening, multi-stage HPGR circuit alternatives and commissioning experience. Circuit design choices are assessed in terms of operability, maintainability and capital and operating cost, and likely future directions for HPGR flowsheets are outlined.

This paper reports on the commissioning and optimisation of the 12 MW, 10.97 m (36 ft) SAG mill installed during the 1995 expansion of KCGM's Fimiston plant in Kalgoorlie, then the largest SAG mill at an Australian mine. Extensive grinding circuit surveys were used to build JKSimMet models of the circuit, which was fed with secondary crushed ore. Simulations of options such as recycle crushing and feed size manipulation identified the operating strategy needed to reach the design throughput of 7.5 Mtpa on refractory sulphide ore.

This paper summarises the decision-making process that led to selecting high pressure grinding roll (HPGR) technology over a conventional SAG mill and pebble crusher (SABC) circuit for the 25 Mt/a Boddington Expansion gold-copper project in Western Australia. It reviews the ore mineralogy and competence, the metallurgical test work and feasibility studies, and the comparative evaluation of circuit options for the very hard diorite and andesite ores. The work documents one of the earliest major decisions to apply HPGRs to a large hard-rock gold-copper operation.

Addressing the long-debated but poorly documented question of whether AG/SAG mill aspect ratio (diameter/length) affects grinding performance, this paper examines pilot and full-scale mill data in both open and closed circuit configurations. A semi-empirical specific energy model based on 43 circuits and 70 ore types is used to strip out the effects of other operating variables so that aspect ratio can be studied in isolation. The analysis indicates that low aspect ratio mills trade throughput for a finer grind but that there is no significant difference in energy utilisation efficiency between high and low aspect ratio mills.

This paper applies the Central Composite Rotatable Design (CCRD) statistical technique to quantify how the length and diameter of the second (inner) vortex finder, cyclone feed percent solids and inlet pressure affect the performance of the JK three-product cyclone. The CCRD approach required far fewer experiments than a three-level full factorial design while yielding nearly as much information. The length and diameter of the inner vortex finder and the feed percent solids were found to be statistically significant in most cases, and the resulting model predicted the experimental data well within the tested range of conditions.

This paper describes JKMRC research supporting the Mine to Mill philosophy, in which blasting and comminution are optimised together rather than as separate steps to increase overall profitability. It outlines blast fragmentation models capable of predicting the finer end of the run-of-mine size distribution, models of crushing and AG/SAG milling, and on-line size measurement tools, and shows via simulation how changes to blast design can significantly increase mill throughput. Reported industry experience, including gains of tens of millions of dollars per annum, is used to illustrate the value of the integrated approach.

This paper develops a theoretical model of the power draw of industrial tumbling mills, known as the C-model, in which the grinding charge is treated as a continuum so that analytical solutions can be derived from its shape and motion. The model applies equally to ball, semi-autogenous and autogenous mills, accounts for the power drawn by both the cylindrical section and the cone ends, and explains the difference between grate- and overflow-discharge mills through the presence of a slurry pool. Its accuracy is demonstrated against a large database of industrial mill measurements.

This paper reports a JKMRC and Newcrest Mining study of the liberation performance of the New Celebration grinding circuit treating gold-bearing ore. Circuit surveys combined with QEM*SEM analyses of individual streams are used to build a detailed picture of the liberation and classification of gold-associated pyrite, and the QEM*SEM results are compared with laboratory leach tests. Mathematical models of the ball mills and cyclones were then constructed and simulations run to identify promising ways of improving circuit performance without costly overgrinding.

This paper reviews operating experience at the Mount Isa Mines copper concentrator after a traditional crushing and primary grinding circuit was replaced by two 9.8m diameter autogenous grinding mills in June 1991. Closed-circuit operation with hydrocyclones caused severe throughput reduction, reduced power draw and over-grinding, revealing a fundamental slurry transport limitation within the mills, so the AG mills had to be run in open circuit to meet the 400 tph design throughput. The resulting flotation feed size coarsened from a p80 of 150 microns to 100 microns, with liberation and flotation studies showing no significant benefit from a finer feed.

This paper examines the issues involved in sizing the increasingly large ball mills that accompany today's big SAG mills, including power draw, residence time, feed size and the applicability of Bond's equations. Analysing operating data from 59 ball mills of up to 26 ft diameter, it shows that a power model with a diameter exponent of 2.5 predicts power draw without bias across all mill sizes. The author concludes there is no evidence that large mills behave anomalously and uses the data to define a safe operating envelope for large diameter ball mills.

This paper describes the SMC Test, a rock breakage characterisation test that needs only very small sample quantities such as quartered drill core, making it well suited to orebody profiling. The test produces a Drop-weight Index (DWi) that correlates with the JK drop-weight parameters A and b and can be used in power-based equations to predict the specific energy and transfer size of AG/SAG mill circuits, with accuracy demonstrated across a wide range of operating plants. The DWi is also correlated with the point load index and with the operating work index of HPGR circuits, allowing specific energy requirements for HPGRs to be estimated in conjunction with pilot or laboratory testwork.

This book chapter reviews the mathematical models developed at the JKMRC for predicting the power draw of comminution machines, covering ball mills, autogenous and semi-autogenous mills, and crushers. The crusher model links power to ore breakage characteristics and the size reduction achieved, while the mill models are based on the motion and magnitude of the grinding charge. Validation data from a wide range of industrial equipment are presented to demonstrate the accuracy of the models, which together allow both the size reduction performance and power utilisation efficiency of plant designs to be evaluated.

This paper reviews the design of the 40-foot diameter SAG mill installed at Newcrest's Cadia gold copper mine in New South Wales, which followed two years of extensive laboratory and pilot plant testwork showing the Cadia ores to be hard and competent, necessitating a variable speed mill drive and pebble crushing. Grinding power requirements were established using in-house grinding software, pilot plant data and JK SimMet simulation. Surveys of the SAG mill during its first months of operation are compared with design predictions, and changes made to the milling circuit are discussed.

This paper presents the mine-to-mill modelling and simulation techniques developed at the JKMRC for exploring how blast design changes affect downstream crushing and grinding performance. Because field experimentation is costly and hard to interpret, integrated blast fragmentation and comminution models are proposed as an economic alternative for evaluating run-of-mine size distribution changes. Case studies demonstrate how optimising blasting together with the comminution circuit can increase plant throughput and overall profitability.

This paper introduces a rock breakage test, the SMC Test, developed to predict autogenous and semi-autogenous mill performance when only limited rock sample quantities are available, such as quartered drill core in the 27-85 mm diameter range. The test generates a strength index, the DWi, which is directly related to the JK breakage parameters A and b and is correlated with the specific energies of a wide range of operating AG and SAG circuits, enabling both model-based and power-based throughput estimation. Because the DWi also correlates with the point load index and UCS, the test is well suited to mine-to-mill studies and orebody profiling.

This paper presents laboratory experiments carried out in a SAG mill fitted with grate-only and grate plus pulp lifter arrangements, using scale models of industrial pulp lifter designs, to study how pulp lifter design affects mill discharge. The results show the influence of pulp lifter chamber volume and demonstrate that over a wide range of conditions the discharge capacity of a pulp lifter assembly is always less than that of a grate-only configuration. The cause of this capacity shortfall is explained and different pulp lifter designs are compared.

This paper describes a new AG/SAG mill model developed to respond realistically to changes in most design and operational variables, addressing the shortage of published models validated against full-scale performance data. The model predicts throughput, power draw, product size distribution, load size distribution and slurry hold-up, and can be used as a stand-alone design tool, for scale-up from pilot data, or for optimising existing circuits. It is validated using data from more than 21 different full-scale grinding circuits.

To overcome the limitation that ball mill simulation models normally require calibration against an existing plant, this paper describes a JKMRC methodology in which laboratory ball mill test results are used to calibrate a perfect mixing ball mill model for greenfield design. Laboratory tests were conducted on ore samples from operating plants whose ball mill circuits were surveyed, and the laboratory-derived models were used to predict full-scale performance. Initial comparisons between predicted and measured plant results demonstrate the validity of the scale-up technique, offering a more versatile alternative to Bond's method for circuits fed by AG/SAG mills.

This paper analyses the global database of over 35,000 SMC Test results covering more than 1,300 orebodies in over 82 countries and 30 commodities. After first demonstrating that SMC Test parameters are accurate indicators of ore hardness through benchmarking against 186 industrial data sets, it examines trends in comminution circuit feed hardness associated with geographical location and mineral species. The result is a broad picture of how ore competence varies across deposits and regions worldwide.

This industry guideline presents a practical, condensed version of the Morrell method for determining comminution circuit specific energy and assessing the energy utilisation efficiency of industrial grinding circuits. It covers the required ore hardness characterisation data from the SMC Test and Bond Ball Mill Work Index test, the Morrell power draw and specific energy equations, and their application to circuits including crushers, high pressure grinding rolls, and AG/SAG and ball mills. Worked examples show how to benchmark the energy efficiency of existing circuits and support comminution circuit design and optimisation. First published in 2016, this is the 2021 revision.

First of a two-part series on slurry transport in grate discharge mills, this paper presents laboratory and pilot-scale experiments on the flow behaviour of discharge grates operating without pulp lifters. The results show that slurry hold-up builds from the shoulder of the charge, with the slurry toe moving towards the charge toe as flowrate increases. Grate open area and aperture position, charge volume and mill speed are all found to strongly influence mill hold-up, providing a basis for modelling the flow capacity of AG, SAG and grate discharge ball mills.

This paper describes the comminution component of a new geometallurgical model developed for the Rosario deposit at the Collahuasi copper operation in Chile. The model combines simulation and power-based approaches to relate ore hardness and flotation feed size to grinding circuit throughput, drawing hardness data from a block model populated with Bond ball work index, JK drop-weight and SMC Test results, and accounting for planned and unplanned maintenance downtime. Validation against weekly production data from January to October 2008 showed an overall accuracy of 4.9% with an R-squared of 0.95, and the model is in routine use for mine planning.

This paper traces the industrial and pilot-scale application of High Pressure Grinding Rolls in Australia, centred on the Argyle Diamond Mine, which installed smooth-roll HPGRs from 1990 and a studded-roll machine in 2002 to improve diamond liberation and throughput. It also covers the extensive HPGR pilot trials at Kalgoorlie Consolidated Gold Mines and Boddington during the 1990s that evaluated the technology as an alternative to SAG milling. Particular attention is given to roll wear experience and to Australian modelling and simulation work that allows throughput, specific energy and wear to be scaled from laboratory to full-size machines.

This chapter-style work addresses how to predict the overall specific energy of modern comminution circuits, arguing that Bond's equations are increasingly irrelevant for AG/SAG and completely inappropriate for HPGR circuits. It presents the Morrell energy-size relationship with a size-dependent exponent and work indices (Mia, Mib, Mic, Mih) derived from the SMC Test and Bond ball work index test. Worked examples compare SABC, HPGR/ball mill and crushing/ball mill circuits, showing the HPGR and conventional crushing routes require lower net comminution energy than the SABC alternative.

This paper reviews whether the transfer size (T80), the 80 percent passing size of AG/SAG circuit discharge, is an appropriate parameter for sizing AG/SAG and ball mills in power-based circuit design. Using measured discharge size distributions from a range of operating circuits, it argues that transfer distributions often violate the mathematical assumptions of power-based equations and that the T80 cannot be independently specified by a designer. It concludes that the transfer size is an outcome of factors such as mill geometry, feed size, ball load and pebble crushing, and should be treated accordingly in design.

This paper describes the conceptual structure of a dynamic AG/SAG mill model being developed at the JKMRC to predict the response of power draw, grinding charge level, slurry level and product size distribution to changes in feedrate, feed size, feed hardness and water addition. Dynamic step-test data from two full-scale high aspect ratio autogenous mills, including trends in power draw, bearing pressure and flowrates plus a novel rock-counting device for coarse feed, are presented for model validation. The mills' measured responses are interpreted in terms of the physical changes occurring inside the mill.

Using data from a large number of pilot and full-scale autogenous, semi-autogenous and ball mill circuits, this paper demonstrates that Bond's classic energy-size equation does not hold over the particle size range of 100mm down to 0.1mm unless correction factors are applied. Morrell proposes an alternative relationship in which the particle size exponent is itself a function of size, removing the need for correction factors. The new equation is shown to predict the specific energy requirements of AG, SAG and ball mill circuits more precisely than Bond's approach.

This paper describes Mine to Mill optimisation programs instituted by MIM Holdings at its Alumbrera, Ernest Henry and Mount Isa base metal operations, aimed at manipulating run-of-mine fragmentation to maximise AG/SAG mill throughput. Because the interaction between blast fragmentation, primary crushing and mill performance is too complex for trial-and-error, JKMRC modelling and simulation techniques calibrated with site data were used to find optimum blast designs. The paper details the data acquisition and modelling phases, the predicted productivity gains and the extent to which those gains were realised.

This paper presents an overview of the JKMRC mathematical model for predicting the power draw of ball, semi-autogenous and autogenous mills, together with the database of over 70 Australian industrial mills assembled to validate it. It discusses the practical problems of obtaining accurate measurements of power draw, mill dimensions and charge levels, and describes procedures to keep measurement errors to a minimum. Using the model, the effects of design and operating conditions on power draw are illustrated, with particular attention to the influence of measurement errors.

Prompted by the Australian trend of closing AG/SAG circuits with hydrocyclones, this paper describes how inadequate grate and pulp lifter capacity leads to slurry pooling, depressed power draw and reduced mill performance. Mechanisms explaining these effects are presented and used to develop mathematical models linking grate flow capacity to slurry hold-up and power draw. The models' ability to predict flow capacity and its impact on power draw is demonstrated with industrial data from large diameter mills.

To understand how grinding media move inside tower mills, a fully scaled transparent glass model tower mill was built at the JKMRC and the motion of glass ball media was measured photographically using coloured tracers under a wide range of operating conditions. The study found that increasing stirrer speed increased the tangential velocities of the grinding media, while reducing ball size decreased tangential velocity due to greater slippage between adjacent balls. Slurry flowrates below the fluidization threshold had no effect on media motion.

This paper examines how feed size distribution, second only to ore competence in importance, affects autogenous and semi-autogenous mill performance, using plant data from a range of operations. It explains why AG and SAG mills respond so strongly to feed size, given their reliance on the feed ore for grinding media, and shows how operations have exploited this through blasting changes, stockpiling, secondary crushing and pre-screening. Recommendations are made for optimum feed size distributions for different mill and circuit configurations.

This paper presents the mathematical modelling of the power draw of vertical stirred mills such as tower, pin and cylinder mills, using data spanning laboratory, pilot and industrial scales. The development of the models is traced from simple empirical correlations through to detailed mechanistic models based on the motion of the grinding media inside the mill, and their predictive performance is compared. The authors also discuss how the mechanistic power models can be incorporated into size reduction models to give a complete description of the grinding process.

This paper presents JKMRC research on how grate and pulp lifter size and design govern the slurry discharge capacity of AG/SAG mills, an area with little prior literature despite its importance for closed-circuit operation. Models relating grate/pulp lifter design to flowrate capacity are described and validated with industrial-scale data from mills up to 12 m in diameter, and the flow-back and carry-over mechanisms responsible for pulp lifter inefficiency are explained. Results from industrial trials of the Twin Chamber Pulp Lifter at Alcoa's Wagerup operation are presented, showing that avoiding slurry pool formation increased mill throughput by more than 15 per cent.

This paper describes the SMC Test and its sample requirements, showing how the test can use very small sample quantities, including quartered drill core, to characterise ore for AG/SAG circuit design. The test generates a Drop-weight Index (DWi) that correlates with the JK drop-weight parameters A and b, with the point load index and UCS, and can be used in equations to predict the specific energy and transfer size of AG/SAG circuits. Its accuracy is demonstrated with data from a wide range of plants, and applications to HPGR circuit design and Mine-to-Mill projects are also outlined.

This paper presents a performance model for high pressure grinding rolls (HPGR) comprising a fundamental throughput model, which accounts for slip at the rolls face, and phenomenological models for size reduction and power draw based on pre-crushing, compressed bed and edge effect breakage zones. The models are calibrated using laboratory-scale breakage and HPGR tests. After calibration they successfully scale up to predict the throughput, size reduction and power draw of pilot and industrial size machines, as demonstrated with diamondiferous ore data.

Building on the author's earlier method for estimating tumbling mill specific energy from laboratory ore characterisation data, this paper extends the approach to jaw, gyratory and cone crushers as well as high pressure grinding rolls. Comminution equipment is divided into tumbling mills, conventional crushers and HPGRs, each described by dedicated hardness indices (Mia, Mib, Mic and Mih) obtained from the SMC Test and Bond ball mill work index test. The technique is applied to three different comminution circuits whose overall specific energies are compared, with all calculations detailed in an appendix.

This paper reports a detailed investigation by Dominion Mining and the JKMRC of two gold treatment plants of very different design treating identical ore blends: Paddy's Flat, a single-stage SAG mill with recycle crusher followed by ball milling, and Haveluck, a traditional multi-stage crushing and ball milling circuit. Both plants were surveyed and modelled using JKSimMet together with recently developed power draw prediction techniques. The study evaluates the two designs in terms of energy efficiency and demonstrates the ability of simulation to accurately predict size reduction and power performance of complex circuit configurations.

This paper presents a new approach for determining the specific energy requirement of tumbling mill (grinding) circuits, applied to a database of 65 industrial circuits. Using this method it is demonstrated that there is no significant difference in energy utilisation efficiency between the various grinding circuit configurations studied, challenging long-held views that AG/SAG circuits are inherently less efficient than conventional ones. Laboratory results are also presented showing that classifier efficiency and recycle load in closed ball mill circuits can significantly affect apparent efficiency through changes in the gradient of the product size distribution.

This paper examines how the economic value created by 'Mine to Mill' improvement, in which more intense blasting in the pit generates downstream benefits in milling, can be tracked and quantified. It argues that traditional departmental cost-based management structures frustrate optimisation across the mining and processing boundary because the value created is much harder to measure than the costs incurred. The authors identify the measurements, reporting systems and organisational measures needed to justify increased blasting costs and to manage the value generated by intense blasting of ore.

This paper reports on JKMRC research programmes aimed at improving the widely used Variable Rates AG/SAG mill model in its three key sub-process descriptions: slurry transport, breakage rate relationships and rock breakage characterisation, including new experimental work on the influence of pulp lifters on slurry hold-up. In parallel, a phenomenological dynamic model of AG/SAG mills was developed and tested on-line at a number of operations to assess how well it mimics real plant behaviour. The latest results from these programmes are presented.

This paper describes a tumbling mill power draw model, developed by Citic SMCC Process Technology, that explicitly represents the slurry phase and its relationship to slurry level, slurry flowrate and grate/pulp lifter design. The media and slurry portions of the charge are treated separately and their power contributions summed, allowing the model to respond realistically to changes in slurry hold-up. Worked examples show how the model explains slurry pooling in AG/SAG mills and the characteristic drop in ball mill power draw after start-up when feed is first introduced.

This paper addresses the lack of a rock characterisation test suited to High Pressure Grinding Rolls (HPGR) circuit design, noting that Bond-style tumbling-mill indices do not correlate with HPGR performance. It shows that the drop-weight index (DWi) from the SMC Test, which requires only small quantities of drill core, is strongly related to the operating work index of HPGR machines. Using data from a wide range of ore types, it describes how the DWi can be applied to HPGR circuit design and performance forecasting.

This paper presents a method devised by JKTech and SMC Testing for converting the A and b ore breakage parameters from JK Drop-weight and SMC tests into a more meaningful specific energy value, termed the SCSE, which represents the specific energy of a standard SAG mill circuit. The approach overcomes drawbacks of the widely used A*b product, which is qualitative, inversely related to impact resistance and non-linear. Results from an inter-laboratory comparison programme involving 27 laboratories worldwide were analysed using SCSE values and confirmed the high precision of both tests, with standard deviations of 3.82% for Drop-weight and 4.88% for SMC tests.

This paper gives practical guidance on planning ore comminution characterization programs for circuit design, brownfield expansions and geometallurgy, addressing the three key questions of which laboratory tests to use, how many samples to test and where to take them from. It reviews the most popular commercially available comminution tests and their accuracy, and uses statistics from a database of 50,000 test results covering more than 1,800 orebodies to show how much hardness varies within deposits. From this it derives guidelines for the number and distribution of samples needed to adequately characterize an orebody.

This paper reviews data from a range of comminution circuits and ore types to test whether certain circuit configurations are inherently more energy efficient for particular ores. It shows that the conventional approach of assessing power utilisation efficiency via the Bond operating work index is fundamentally flawed, since apparent efficiency differences largely disappear when a size-dependent energy-size relationship is used. An alternative basis for comparing circuit energy efficiency and describing ore competence, built around a drop-weight index (DWi), is proposed and supported with operating data.

Motivated by the growing need to grind fine-grained refractory ores to below 20 micrometres, this paper analyses survey data from an industrial tower mill circuit using the population balance models of the CSIRO and the JKMRC. The ability of both models to reproduce the effects of changes in feed size distribution, ball load and feedrate is assessed against the plant data. Comparisons between pilot and full-scale tower mill operation are also used to propose a prototype scale-up methodology for predicting the effect of changes in mill design.

This paper introduces the SMC Test, a rock breakage test designed to work with limited sample quantities such as small diameter drill core, and shows how its output index (the DWi) can be used to predict AG/SAG mill circuit specific energy via both power-based correlations and model-based approaches. The DWi is shown to be related to the JK drop-weight parameters A and b, to rock strength measures such as the point load index, and to HPGR performance, making it useful for orebody profiling and mine-to-mill studies. The predictive capability of the method is demonstrated against a wide range of industrial data and compared with traditional approaches such as Bond's.

This paper validates the JKMRC high pressure grinding rolls model and scale-up procedure developed by Morrell, Tondo and Shi, using four data sets from three industrial machines operating in the diamond industry (De Beers, BHP Billiton and Rio Tinto) fitted with both smooth and studded rolls. Laboratory-scale HPGR tests were used to calibrate the model, which was then scaled up and compared with full-scale performance. The authors conclude that product size distribution is predicted well, while throughput and power predictions are sensitive to inconsistencies in the gap-to-diameter ratio between laboratory and full-scale units, and that the model is mature enough for circuit evaluation and optimisation.

This presentation examines whether there is an association between the physical and morphological properties of iron ore deposits and the design of the comminution circuits used to treat them. It reviews the relevance of comminution to project economics, the main equipment types including crushers, high pressure grinding rolls and tumbling mills, and the hardness characteristics of iron ore bodies. Typical grinding circuit designs are then related to ore properties, showing how deposit characteristics drive the choice of comminution route.

This paper documents how Alcoa's single-stage SAG mills at its Western Australian alumina refineries were limited by insufficient pulp lifter capacity, which caused large slurry pools and depressed power draw at maximum throughput. Research at the JKMRC led to a new twin chamber pulp lifter concept that was engineered and installed at Alcoa. The installation delivered large gains in throughput and power utilisation efficiency, and the paper traces the project from problem recognition through research, design and full-scale results.

This paper examines the sensitivity of AG and SAG mill performance to feed size distribution, using dynamic trends and plant data from a range of operations to show how changes in top size, intermediate size fractions and fines affect throughput and power draw. It discusses how feed size can be deliberately manipulated through primary crusher operation, additional crushing capacity, stockpile segregation control and, in particular, changes to blast design. The paper concentrates on demonstrating that modifying blasting practice to alter run-of-mine fragmentation can deliver considerable overall benefit to milling circuits.

Drawing on the Julius Kruttschnitt Mineral Research Centre's Mine to Mill research through a major AMIRA project and industry contracts, this paper describes how blast design can be tailored to optimise the whole rock breakage system spanning blasting, crushing and SAG milling of hard, brittle ores. It explains why such energetic blasts fall outside the conventional design envelope, the critical factors that must be considered to achieve their objectives, and the potential problems to avoid. Field experience shows that the quality of implementation is critical to the success of Mine to Mill blasting.

This paper reports pilot-scale experiments at the JKMRC on how the design and size of discharge pulp lifters affect slurry transport in AG/SAG mills. The results show that conventional pulp lifters with insufficient volume severely limit flow capacity, mainly because slurry flows back from the pulp lifter into the mill instead of discharging. A new Twin Chamber Pulp Lifter design that largely eliminates flow-back is presented, its benefit confirmed by an industrial trial on a 7.7 m SAG mill at Alcoa's Wagerup refinery where throughput increased by over 15 percent.

The second part of this two-part contribution describes a simpler empirical mill power model, the E-model, whose behaviour is based on the continuum C-model presented in Part 1. It documents in detail the extensive database of industrial ball, semi-autogenous and autogenous mills assembled for validation, including the difficulties of accurately measuring mill dimensions, fillings, speeds and power draws. The database is used to show that the E-model is only slightly less accurate than the C-model despite its much simpler structure.

This paper presents a new JKMRC model of AG and SAG mills based on charge dynamics, relating charge motion and composition to both power draw and size reduction. Breakage is treated as separate impact and abrasion/attrition processes linked to the energy available in the mill, the charge size distribution and the relative motion of the grinding media, with ore-specific breakage characterised by laboratory tests over a wide range of energies and sizes. Slurry transport is handled by the JKMRC's grate-design-sensitive discharge model, and the paper describes the overall model structure and its main sub-processes.

Discrete element method (DEM) simulations of a scale model SAG mill are quantitatively compared with earlier experimental photographs taken at the JKMRC, examining charge shoulder and toe positions for one fill level, three mill speeds and three liner profiles. Traditional two-dimensional DEM with circular particles is shown to under-predict shoulder and toe positions by around 10 degrees, and particle shape is found to make an important contribution to the charge profile. Full three-dimensional DEM modelling is demonstrated to quantitatively capture the real charge dynamics observed in the experimental mill.

Drawing on the SMC Test database of more than 17,000 tests covering over 650 ore bodies, this paper studies how ore hardness variability differs from deposit to deposit for AG/SAG, crushing and HPGR circuit design. The distribution of hardness variability is shown to be bi-modal, with coefficients of variation typically in the 15-20% or 35-40% ranges. Based on classical statistics applied to these trends, the paper recommends minimum sample numbers for orebody characterisation programmes, suggesting a staged approach as projects progress from pre-feasibility to bankable studies.

This paper addresses the historical lack of a formalised, validated method for estimating the specific energy requirements of AG/SAG-ball mill circuits and the newer HPGR-based circuits. It presents a simple but accurate procedure in which coarse and fine ore work indices (Mia, Mib and Mih) derived from the SMC Test and the Bond ball mill work index test are used in a general size reduction equation to determine overall circuit specific energy. The approach is then applied to compare the energy usage of three different comminution circuit configurations.

This PhD thesis develops validated models for predicting the power draw of wet industrial grinding mills, addressing the historical lack of models tested against comprehensive full-scale data. Charge motion was measured photographically in a glass-fronted laboratory mill and the results incorporated into a theoretical power draw model that explicitly accounts for grate or overflow discharge and planar or conical mill ends, together with a media-size-sensitive variant and a simple empirical version. Calibrated and verified against 76 data sets from ball, SAG and AG mills spanning 7 to 7900 kW, all three models predicted power draw with high accuracy, outperforming existing published models.

This paper describes the design of the 12m (40-foot) diameter SAG mill installed at the Cadia Hill Mine, the world's largest at the time, whose selection followed two years of laboratory and pilot plant testwork that showed the Cadia ores to be hard and competent, requiring a variable speed drive and pebble crushing. Surveys during the first four weeks of operation revealed feed size distribution, transfer size and pebble rate lower than the design assumptions, and the SAG ball load was increased to compensate, resulting in total comminution circuit power about 0.5 kWh/t higher than predicted. The JK SimMet SAG model was modified to account for these variations.

This keynote paper traces the history of comminution circuit modelling and reviews the modelling approaches now available to circuit designers, notably power-based methods and simulation models, in an era when pilot testing has become rare. The pros and cons of each approach are discussed, with particular attention to how accurately the models size comminution equipment. Model predictions are benchmarked against measured data from full-scale operating plants to indicate the accuracy that designers can expect.

This technical note extends the author's earlier method for predicting the specific energy of crushing and high pressure grinding roll circuits to situations involving relatively coarse feeds. The extension is achieved through a size-dependent hardness parameter that modifies the Mic and Mih work indices, and the conditions under which its use improves prediction accuracy are described using data from pilot and full-scale machines. A worked example for a primary crushing circuit illustrates the calculation procedure.

This paper presents a new AG/SAG mill model developed at the JKMRC that is based on charge dynamics, relating charge motion and composition to both power draw and size reduction. Impact and attrition/abrasion breakage are treated as separate processes linked to the energy available in the mill, the charge size distribution and the relative motion of the grinding media, with ore-specific breakage described from laboratory tests over a wide range of energies and particle sizes. The overall model structure and its main sub-processes, including a slurry transport model incorporating grate design effects, are described.

This paper reports laboratory SAG mill experiments with scale models of conventional pulp lifters, studying the effects of pulp lifter size and design, grate open area and aperture position, mill speed and charge volume on discharge capacity. The results show that conventional pulp lifters with insufficient volume severely limit flow capacity, mainly because of flow-back of slurry from the pulp lifter into the mill. A new design developed at the JKMRC, the Twin Chamber Pulp Lifter (TCPL), is described which largely eliminates flow-back and brings mill operation close to ideal grate-only discharge conditions.

Drawing on experience from copper and gold projects treating highly competent ores such as Geita and Boddington, this paper discusses recurring problems in comminution circuit design test work and data interpretation. Issues examined include modified or non-standard test procedures, worn or damaged test equipment, conflicting results between different SAG mill specific energy tests, and modelling based on poor benchmarks or incorrect use of JKSimMet. Using case studies of Bond crushing and rod mill work indices and various SAG specific energy tests, the authors clarify these issues to reduce conflict in the interpretation and application of design test work data.

This paper reviews a century of published grinding mill power draw models and argues that most relied on unrealistic assumptions of a fixed charge position and shape, and were never validated against comprehensive industrial data. It then describes a model, developed from a detailed study of charge dynamics combined with a large database of industrial mill power draws, that treats AG, SAG and ball mills as a single class of machine. The model is shown to predict the power draw of all wet tumbling mills to within about 10% at the 95% confidence level.

This paper describes JKMRC developments to an energy-based AG/SAG mill model in which ore-specific breakage distributions, measured in laboratory high-energy impact and low-energy abrasion tests, are combined with in-mill energy levels estimated from the mill load and dimensions. Material transport is handled through a grate classification function and an empirical mass transfer relationship linking fines hold-up to feed rate, within a perfect mixing framework. Analysis of an expanding mill database reveals a systematic relationship between ball load and grinding rates, which is presented and discussed.

Second of a two-part series, this paper examines how conventional radial and spiral pulp lifters perform in combination with discharge grates, based on extensive test work in a 1 metre diameter pilot SAG mill. It shows that conventional pulp lifter designs restrict slurry flow considerably, so the discharge rate falls short of what the grate alone could pass at a given hold-up. The findings quantify how pulp lifter design, grate design and operating conditions such as mill speed and charge level determine the ultimate flow capacity of AG, SAG and grate discharge ball mills.