Search Documents

By K, J. MAUNDERS

This paper briefly describes a computerised model of the South African electricity system which has been developed jointly by the Atomic Energy Board and the Electricity Supply Commission. The results of an analysis of this system have been subjected to a fairly extensive sensitivity analysis which reveals that up to the year 1990 between 680 and 700 million tons of coal will be required for electric power generation, together with between 7 000 and 10 000 tons of uranium. This result is fairly insensitive to data variations, but the prediction of a cumulative requirement up to the year 2000, of between 1 300 and 1 800 million tons of coal and 25 000 to 70 000 tons of uranium, proves to be very data-sensitive.

Jan 1, 2014

By T. Bhambhani, J. Jankolovits

Sphalerite flotation using copper sulfate has been performed in the industry for decades and the mechanism of Cu activation is well established. Despite its ubiquity, many practical questions surrounding copper sulfate persist. It is not clear how iron selectivity is achieved in zinc circuits given the propensity of iron sulfides to be activated by copper. Also, the large variation in copper sulfate dosages across different plants and ore types cannot be explained. Lastly, Zn flotation with greatly reduced copper sulfate dosages attracts significant interest and has been demonstrated in certain laboratory settings, but no information is available on why those laboratory solutions have not been adopted in plant practice. Flotation studies are presented that demonstrate that Zn flotation with very low copper sulfate dosages is possible when using certain novel collectors that are not practised in the industry. However, these reagent schemes are impractical because of a heretofore unappreciated and critical role that copper sulfate plays in Fe rejection. With both very low and standard copper sulfate dosages, copper sulfate influences both Zn flotation and Fe rejection. The results highlight the multifaceted role of copper sulfate in zinc flotation and the need to holistically consider complex systems’ behaviour to understand flotation reagent mechanisms and optimise plant performance. Keywords: Sphalerite flotation, metal activation, copper sulfate, iron rejection

Jan 1, 2020

By A. H. Mokken, G. K. I. Blendulf, G. J. C. Young

A study of the arrangements for pulp discharge on pebble mills, and their influence on mill performance by A. H. MOKKEN, B.Sc. (Eng.) (Rand) (Fellow), G. K. I. BLENDULF, M.Sc. (Stockholm) (Member), and G. J. C. YOUNG, B.Sc. (London) (Member) Although used for many years in the gold-mining industry, the conventional pan lifter at the discharge end of a pebble mill has hindered attempts at optimization of speed of rotation and pulp flow. A theoretical approach to flow patterns in pan lifters has pointed to limitations in the capacity of these lifters, with capacity decreasing as speed of rotation is increased. Complete elimination of the pan lifter by the adoption of peripheral or end discharge has resulted in a marked increase in pulp flow and an improvement in power utilization. Assays of the gold content of pulp discharged after a mill stoppage from mills fitted with pan lifters show values well in excess of the current milled grade, which indicates an accumulation of gold particles in the mill and, therefore, a longer retention period. The same could apply to other heavy minerals and metals present in the ore feed or worn off mill linings and metallic grinding media. Examples are osmiridium, fine iron and iron alloys, and copper and sulphide minerals. This phenomenon is attributed to the higher specific gravities of these metals and minerals, causing them to migrate towards the mill shell, from which position they are not readily removed by the lifters at the discharge end of the mill. The close association of gold particles with iron and sulphide minerals in the crevices of mill liners might be conducive to the formation of coatings on the gold. Assays of samples of milled pulp taken from a mill designed for end discharge show a lower accumulation of gold. The accumulation and nature of these gold particles are being studied at present. With increasing diameters of pebble (or autogenous) mills, peripheral or end discharge appears to be the preferred method of pulp removal.

Jan 5, 1975

Palabora Mining Company is operating a copper block cave mine in the Limpopo Province of South Africa. The mine was developed in the late nineties with a target ore reserve of 220 million tonnes of carbonatite at 0.69% copper. At 30,000 tonnes per day the operation had an estimated life of mine of approximately 23 years. Cave induced open pit side wall failure occurred in late 2004 with 130 million tonnes of waste reporting to the bottom of the open pit. This failure resulted in potential sterilization and dilution of the ore reserve and subsequent reduction in life of mine. Physical and numerical modelling indicated potential losses of up to 30% of the original ore reserves. Cave Management tools based on throughput principles are utilized to manage the failed waste ingress and ensure an optimum life of mine. Extracting the maximum rate possible, as soon as draw points are commissioned, with the view of achieving the highest production in the short term, is any mining operation?s goal as more money is required sooner. This approach, of course, improves the cash flow but presents a risk in the cave propagation profile as overall draw control is compromised. At Palabora it is evident from a Mineral Resource Throughput Management (MRTM) point of view that maximum production in the short term is not the solution. If high production rates are achieved and the profile of the cave is not manage correctly, this results in failed waste reporting to draw points earlier with adverse effects on grade. This would effectively and significantly shorten Palabora?s Life of Mine. The applications of good MRTM principles are required to assist the value chain for sustained profitability. The process commenced with change management paving the way for a better understanding and a holistic view for all role players.

Jan 1, 2007

By D. Limpitlaw, A. Alsum, D. Neale

"Ecological footprint analysis (EFA) was first described in 1996 as a measure of carrying capacity appropriated by human activities. EFA is a resource and emissions accounting tool designed to track the demand on the biosphere’s regenerative capacity. Ecological footprints are increasingly used as indicators of organizational and corporate environmental performance and product sustainability. There is a compelling argument to develop an EFA tool for mining companies.To determine the size of a mine’s ecological footprint, land requirements for all categories of consumption and waste discharge must be summed. This land is only made up of the ecologically productive land and water in various classes (cropland, pasture, forests etc.) required on an ongoing basis to provide all energy and material resources consumed and absorb wastes.A challenge in conducting an EFA for a mine is the shortage of accessible data. Undertaking an EFA for a corporation or individual site entails compiling consumption and emissions data (which can be used for other reporting applications). The footprint results themselves highlight the most critical aspects of an organization’s impact on the environment and provide a platform for focusing actions and for educating the workforce to improve their contribution to best-practice operations. This paper discusses the benefits and challenges of undertaking an EFA for a mining company, and provides examples of the various components of ecological footprints associated with mines, as well as showing how an EFA can be used to understand and communicate some of the impacts associated with mining activities.IntroductionFundamental biophysical and economic limits are now being experienced in many countries around the world. Herman Daly, a former World Bank economist, described this scenario as ‘uneconomic growth’, where the costs of growth exceed the benefits (Daly, 2005). This occurs when the economy’s expansion encroaches excessively on the surrounding ecosystem, sacrificing natural capital. Under such conditions the sacrificed natural capital is more valuable that the resulting economic growth. Consequently, the world faces largescale threats to sustainability and especially to the viability and continued existence of the ecosystems that support human settlements (El Zein et al., 2014). In recognition of this sustainability imperative, mining companies are increasingly concerned with their ecological footprint (EF)."

Jan 1, 2017

By Z. Tshentu, M. Ngayek, P. Moleko-Boyce

This paper is focused on the development of iridium(IV)-specific quaternary diammonium cations containing electron-donating substituents in the benzyl group of methyl and electron-withdrawing substituents such as triflouromethyl and nitro groups. Quaternary diammonium cations used were tetra- 4-methylbenzyl-1,10-diammonium decane chloride (QuatDMDAMeBnz), tetrabenzyl-1,10- diammoniumdecane chloride (QuatDMDABnz), tetra-3-trifluoromethylbenzyl-1,10- diammoniumdecane chloride (QuatDMDACF3Bnz) and tetra-4-nitrobenzyl-1,10-diammoniumdecane chloride (QuatDMDANO2Bnz). Polyvinylbenzylchloride (PVBC) nanofibres were used as support material, and the functionalised nanofibres were used for separation of Ir(IV) from Rh(III). The sorbent materials were characterised by FTIR and elemental analysis. The efficiency of the quaternary diammonium cations was investigated in a column study under dynamic flow adsorption conditions. Column sorption of [IrCl6]2- and [RhCl5(H2O)]2- on nanofibres functionalised with diammonium cations showed [RhCl5(H2O)]2- was not adsorbed by the sorbent materials while [IrCl6]2- was loaded onto the column. The loading capacities of [IrCl6]2- on the quaternary diammonium sorbent materials increased in the order of F-QuatDMDAMeBnz (3.08 mg/g) < F-QuatDMDABnz (3.39 mg/g) < FQuatDMDACF3Bnz (5.46 mg/g) < F-QuatDMDANO2Bnz (7.10 mg/g). The loading capacity for Ir(IV) with quaternary diammonium cationic nanofibres increased with an increase in the electron-withdrawing nature of the substituent on the quaternising group. The charge delocalising ability of the 4-nitrobenzyl group resulted in the best interaction of the decamethylene diammonium cation with [IrCl6]2-. Keywords: Iridium(IV), rhodium(III), quaternary diammonium, polymer nanofibres, separation

Jan 1, 2020

By F. Reyes, M. Hilden, G. Forbes, M. Yahyaei

Since AlphaGo defeated the best human player in the game of Go in 2016 - something thought to be an impossible achievement - machine learning, deep learning and reinforcement learning have caught the attention of scientists, the media and the general public alike. This recent development in machine learning (let us remember that neural networks have been available for decades now) opens new possibilities in the areas of modelling and process control. However, little has been reported in the minerals processing industry about these new possibilities and opportunities. This paper presents and introduces the topic of reinforcement learning and the different platforms that allow its implementation. The paper shows the consequences of using reinforcement learning for the specific area of process control and optimisation. As an example, the authors present the use of reinforcement learning for the control of a SAG mill closed with a pebble crusher using industrial data from an operation. A comparison with advanced model-based process control is provided, showing the benefits and shortcomings of reinforcement learning as a new control and optimisation technique. Keywords: Reinforcement learning, deep learning, process control, process optimisation

Jan 1, 2020

By S. K. Murphy

A partial shaft pillar extraction project was embarked on at a mine in the West Wits mining region in South Africa. Prior to the extraction, a rock engineering assessment was conducted with the main objective being to establish which areas of the shaft pillar could be extracted without unduly affecting the stability of major infrastructure within the shaft pillar. Standard numerical modelling methods were used to conduct this assessment and several areas were identified for extraction. However, during the extraction process, it was found that the infrastructure was affected more than was originally predicted by linear elastic modelling. In addition to this, several large events resulted in far more large aftershocks than would normally be predicted. In order to gain an understanding of this phenomenon, seismic data was integrated with numerical modelling as follows: > The seismic data in the form of displacements was integrated in 3D space within the model to assess the effect of static stress changes associated with seismic activity > Data from large events was integrated onto specific structures, where displacements associated with large events were integrated onto the plane of structure in question. This paper describes the methodology and results obtained from each of the above processes and include insights into the mining process that were previously unfathomable using standard linear elastic numerical modelling. The way forward for this rock engineering assessment of partial shaft pillar extraction is also discussed.

Jan 1, 2012

By J. W. van der Merwe, A. M. Maleka, V. J. Matjeke

The 10.5–12.5% chromium, titanium-stabilized ferritic stainless steel (FSS) plates used to fabricate coal line jumbo wagons is susceptible to undesirable chromium depletion during welding with 309L electrodes. In this work, non-stabilized FSS plates were welded using 308L welding electrodes. The objective was to weld non-stabilized FSSs using gas-shielded metal arc welding without causing sensitization. The welding process was optimized by lowering the heat input. Sensitization was assessed by rigorous etching techniques and anodic polarization scans. In addition, the welds were subjected to tensile testing, and their fracture surfaces were examined for intergranular cracking. Non-stabilized FSS plates were successfully welded using gas metal arc welding.

Apr 5, 2023

By N. C. Joughin

A simplified economic model is derived for the financial return from a new gold mining project. The model is used to analyze the effects of changes in the mining rate, capital costs, working costs, value of the reef and the stoping width. The most effective avenues for improving the return from a new project are indicated. It is shown that those technological innovations which would improve the return significantly, and therefore make it possible to open more mines, are the use of smaller mines, means for improving the mining rate to capital costs ratio for smaller mines, planning and control of mines by computer simulation, an improved mine valuation procedure and means for mining at a decreased stope tramming width.

Jan 1, 2014

By I. Fedotov

One of the most effective methods of metal surface heat treatment is through laser hardening. In most cases, laser parameters are determined experimentally, which is expensive and time-consuming. A mathematical model for the surface heat treatment process will allow a significant decrease in costs, and reduce the time required for optimization. In the present work, the authors present a mathematical model for laser heat treatment with forced cooling which can be applied to medium carbon steel. The model is based on a three-dimensional heat transfer equation using Green?s function. The model allows one to calculate the temperature of the treated steel at different depths below the surface with and without cooling, for set values of the laser power and travelling speed, when the heat transfer coefficient is a known constant. It also takes variations in the radiation adsorption coefficient on the metal surface and laser beam spot size into account. The model was illustrated by comparing calculated and experimental temperature profiles of mild carbon steel with 0.5% C with and without cooling. A reasonably good agreement was achieved between experimental and predicted values to ensure that a high enough austenizing temperature with short enough cooling time was reached for direct martensite formation. It also predicted the depth of treatment accurately. The incorporation of cooling in the model substantially improves previous models based on heat transfer only. It delivers a tool that can with a reasonable accuracy predict the depth of surface heat treatment and cooling time required from operational parameters, such as laser power, speed of motion, and distance between laser and cooling source, that can be controlled in the work set-up. This can eliminate trial and error efforts and result insignificant time and cost savings. Keywords Laser, surface treatment, martensite transformation, carbon steel, heat treatment.

Jan 1, 2011

By M. S. Shishvan, J. Sattarvand, R. S. H. Meij

Navigating miners during an evacuation using a smart evacuation technology can significantly decrease the evacuation time of an underground mine in case of an emergency. This paper presents a mathematical programming model to calculate the most efficient escape path for each miner as a critical component of the smart evacuation technology. In this model, the total evacuation distance of the crew is minimised and scenarios with blocked pathways, and stamina categories for the miners are simulated. It was found that all the tested scenarios are technically feasible. Using the feature that filters out blocked pathways has no downsides, as safer routes are calculated, and there is no penalty in the computation time. The paper also discusses the social and technical issues that need to be overcome before the algorithm can be implemented as an actual escape solution.

Sep 1, 2021

By Brian Wyatt, Neil Webb, Marcel Roche, John Thirkettle

Recent developments in survey techniques and measuring equipment now allow for an accurate assessment of external coating condition and cathodic protection performance. These developments build on 20 years of experience of using Direct Current Voltage Gradient (DCVG) and Close Interval Potential Survey (CIPS) methods. The techniques use the latest advances in measurement systems coupled to a bespoke data processing system. The paper describes tests and trial surveys carried out on a hydrocarbon transmission pipeline in West Africa during 2005. The survey works described and conclusions drawn are from the data collected. Of particular interest are the apparent capacitance effects which are thought to occur under disbonded coating and the characteristics that these may show on the integrated CIPS and recorded DCVG profiles. This paper was originally presented at the Cefracor conference in Aix, France, in June 2006 The pipeline was originally installed on a phased basis between 1989 and 1994. The external coatings were three layer polyethylene factory coating and shrink sleeve field joint coating. An impressed current cathodic protection system was installed. The pipeline has subsequently been affected by disbondment of the coatings. It was a primary objective of the trials to attempt to identify such affected areas. In addition to a technically detailed account of the survey works undertaken and the results obtained the authors will consider the practical and commercial aspects of these surveys with input that provides a perspective from the specifier, pipeline surveyor, pipeline operations personnel and the pipeline owner. It is intended that the data presented will provide greater awareness of the benefits and limitations of the techniques used and their role for both stand-alone interpretation and the integration of the survey information with other inspection methods used for external corrosion assessment. The application of the described techniques is well developed and proven to locate and characterise through?coating defects and to provide a full audit of cathodic protection status. However, the possibility of the techniques being able to indicate disbonded coating is unproven and further tests and trials are planned which may allow a greater understanding of the electrical characteristics which are believed to be associated with disbonded coatings which may or may not be filled with an electrolyte. Further work is currently planned in West Africa, South Africa and Scotland in order to improve the understanding of this technique and to determine if the reported indications do actually represent a method of locating disbonded coatings. Keywords: External Coating, Cathodic Protection, DCVG, CIPS, corrosion assessment.

Jan 1, 2006

By A. O. Filmer

In the mining and metals industry, business strategy demands that leaders extract more from any set of capital assets. This improvement is achieved through a combination of continuous improvement and upgrading of the assets. Continuous Improvement involves a series of incremental changes in the workplace to stabilize the process, remove waste, and test the limits of process capability. Enhanced Technology involves the replacement of an existing as set with a different process or equipment, to achieve a similar function in a more efficient and effective manner. The paper compares and contrasts the two methodologies and outlines the key business considerations which dictate how and when to select one or other improvement method. The paper also addresses the implications for these two improvement methods for those undertaking R&D in support of the mining and metals business.

Jan 1, 2009

By А. V. Aksenov, S. G. Rybkin, А. Y. Senchenko

Recovered activated carbon fines result from processing of ore and concentrates via cyanide leaching and sorption of valuable metals on carbon at gold-mining plants. This intermediate product is characterised by the particles sized smaller than 0.6 mm and gold grade up to 300-500 g/t. Processing of carbon fines with recovery of valuable metals is not available at many plants due to the lack of optimal technology. The paper describes the testwork and development of an efficient technology to process carbon fines with the recovery of valuable metals into the target gold alloy. This innovative technology involves pelletising of fine activated carbon particles using an organic binder, drying and further firing of pellets in a grate fired furnace at 800-900 °С. The produced ash is subject to sintering with sodium nitrate and hydroxide at 600-700°С. The sintered ash is then leached in water. The insoluble residue containing gold and silver is removed by filtration, dried and smelted to produce gold alloy. Water solution containing sulphur and arsenic is subject to detoxification. Processing of recovered carbon fines according to the developed technology implies the use of standard full-scale equipment. The offered method to treat gold-bearing carbon fines was patented by the Eurasian Patent Office. Key words: Fine activated carbon particles, valuable metals, pelletizing, firing, ash, sintering, leaching, smelting

Jan 1, 2020

By D. Marcotte, R. V. Dutaut

"In mineral resource evaluations, geostatistical methods known as global change of support allow prediction of the theoretical histogram and gradetonnage curves prior to interpolations or simulations of grades. Two methods commonly used by professionals to guide the choice of interpolation parameters and assess results are the discrete Gaussian model (DGM) and the indirect lognormal correction (IndLog). These models rely upon an estimate of the dispersion variance of the blocks, which is derived by numerical integration of the variogram model over a discretized block. Due to difficulties in obtaining well-formed traditional experimental variograms (especially in the presence of outliers and limited clustered data), many professionals prefer to use ‘normalized’ variograms such as correlogram (non-ergodic variogram), pairwise-relative, or variogram of the normal score transform. A series of simulations with different grade distributions and variogram models are used to assess the performances and robustness of the various variogram estimators with respect to the DGM and IndLog global change of support. Our results show that the traditional variogram, the correlogram, and normal score variogram have better performances, compared to pairwise, for both DGM and IndLog. Moreover, DGM provided better results than IndLog for the grade distributions that are not strictly lognormal. These findings provide valuable guides for geostatistics practitioners. IntroductionWhen reporting resource estimates,practitioners are required to follow generalstandards described in the NI43-101 (CIMguidelines), JORC, or SAMREC codes. One cangenerally recognize three main parts directlyrelated to the block model:•,Pre-processing: exploratory data analysis (EDA), domaining, capping, compositing, declustering, variography•,Processing: block size, neighbourhood, interpolation types (linear or nonlinear)•,Post-processing: classification, reporting.Pre-processing aims to simplify and strengthen the processing step. Ideally, domaining, capping. and compositing is aimed at defining a single homogenous population. During the processing phase, a common practice is to use a block size corresponding to the planned selective mining unit (SMU). In most precious metal deposits (gold, notably), the SMU size is frequently smaller than the recommended half data spacing (Journel and Huijbregts 1978), resulting in high estimation variance and either a high degree of smoothing if the regression slope is managed or a high degree of conditional bias if it is not. Most actual resource estimates are made using block sizes between one-quarter to one-sixth of the average data spacing, sometimes even smaller. In addition to the block size, the interpolation choices for the neighbourhood selection and interpolator (e.g. inverse distance (ID) or ordinary kriging) are often based on the Qualified Person’s (the QP) experience and some basic validation plots that often do not consider the conditional bias (e.g. trends or SWATHs plots, which are designed to compare two sets of population using a one-dimensional graph)."

Jan 1, 2018

By M. Frenzel, K. Bachmann, M. Rudolph, R. Tolosana-Delgado, E. Schach, M. Buchmann, L. Pereiraa, K. G. van den Boogaart, N. Kupka

Geometallurgy - the science of predicting the behaviour of ores along the value chain based on their geological and mineralogical characteristics - has barely moved beyond process minerology studies. These studies typically link process behaviour to the mineralogical and microstructural characteristics of pristine or processed ores in a qualitative or semi-quantitative manner. Eventually, a trial-and-error optimisation of the operation is achieved. However, to deliver on its promise, geometallurgy should progress beyond this current approach to a fully quantitative understanding of process behaviour, including the various uncertainties involved, namely predictive geometallurgy. One of the barriers currently hindering this development is the neglect of most ore characterisation data obtained in geometallurgical studies. This is mostly due to the large extent of this information exceeding the capacity of conventional software. To address this challenge, we developed a data mining platform consisting of an SQL database, an R front-end in-house package, and back-ends to interpret data from several analytical instruments. This platform enables the use of the statistical and data mining power of the R open-source environment for any geometallurgical study. Potential applications range from ore characterisation to process understanding and forecasting, and from geostatistical prediction to operational control and optimisation. The scripting abilities of R allow for the: 1) processing of many streams in loops, thus freeing the user from repeating tedious click-and-drop tasks; 2) computation of virtually any derivate or aggregate quantity from the data; 3) distribution of the work among several processor clusters; and 4) keeping of a log file for the calculations done. This contribution presents the building blocks of this platform, and illustrates with several examples its potential to enable predictive geometallurgy. Keywords: Data mining, entropy, kernel methods, LASSO regression, predictive geometallurgy

Jan 1, 2020

By E. Moosavi, K. Tolouei

In mine production scheduling problems are well-known, where most are categorised into deterministic and non-deterministic NP-hard problems because of their complexity. Many researchers have attempted to solve the problem by applying various optimisation techniques. While using traditional methods they observed extreme difficulty in solving highly complex problems. In the 1990s, many researchers addressed long-term production scheduling (LTPS) by using simulated annealing, the genetic algorithm, the tabu search algorithm, and the ant colony algorithm. These algorithms are known as the meta-heuristic algorithms, and have proved to be the most efficient algorithms for solving LTPS thus far. The firefly algorithm is one of the best methods for visualising problems related to LTPS, and provides the best possible optimisation. It is one of the simplest methods, and easy to apply to any NP-hard problem. The experimental, and comparison with other recently published algorithms shows that the proposed algorithm is feasible, and is an effective approach for long-term production scheduling problems. The results indicated that a better solution is obtained by the proposed method, compared to other methods in terms of cumulative net present value, and average ore grade. Additionally, the CPU time via the proposed model is roughly 2.52% higher than the other method.

Sep 1, 2021

By A. R. Javadi

Pyrite (FeS2), chalcopyrite (CuFeS2), sphalerite (ZnS) and galena (PbS) generate hydrogen peroxide (H2O2) when placed in water, with pyrite generating the most amount of H2O2. The order of these minerals for generating H2O2 in water is: pyrite > chalcopyrite > sphalerite > galena. In previous research, the methods of generation of H2O2 in the grinding mill were studied. In this study, data from experiments were included by using MODDE 9 software for six parameters composed of water, grinding media, particle size; adding collector to the grinding, and adding depressant and pH. Finally the optimal condition was achieved. According to the results, most of the recovery is related to the particle 75 μ size and when using processed water, adding 100% collector and without depressant and pH 10. The most recovery of lead in the case of the use of fresh water is in the case of 100% collector, without depressant, pH 10 and 150 μ particle size. Maximum recovery in fraction size 45 μ in the case of water process is 100% collector, without depressant and pH 10. Maximum depressant of pyrite will be the fraction size 75 μ, distilled water, without collector, 100 depressant. Keywords: Sulfide mineral, hydrogen peroxide, flotation, oxidation

Jan 1, 2020

By S. Liaghat, J. Rosenkranz, P. Semsari Parapari, S. Chehreh Chelgani, A. Tohry

It is commonly accepted that the energy consumption for comminution within mineral processing is responsible for around 3% of the global electrical energy usage. However, it was clearly noted that inside mills, less than 1% of this energy efficiency is actually consumed for particle size reduction. Therefore, monitoring of grinding parameters and modelling of product particle size distribution (e.g. characterised by the 80% passing size P80) play an essential role in operating costs of a mineral processing plant. Several empirical models have been introduced for the prediction of the P80. Since in the industrial scale usually a few numbers of grinding variables are monitored, these models cannot identically be used in plants. In addition, these empirical models do not always indicate the magnitude of correlations between milling variables. This study proposes to introduce the Hilbert-Schmidt independence criterion (HSIC) as an intelligent model for exploring linear and nonlinear relationships between grinding variables and P80 for an industrial scale data assessment. HSIC can rank variables based on their importance and help to select the most effective one for the generation of predictive models. HSIC modelling through various monitored grinding variables (ball charge, feed input rate, F80, power draw and work index) of an industrial plant indicated that F80 as a feed property has the highest effect on P80 of the ball mill products. Modelling results highlighted that the P80 can be accurately predicted, based on effective variables by using a HSIC model. These outcomes showed that HSIC as a powerful modelling tool can assess the relationships between monitored grinding parameters, select the most effective grinding parameters and satisfactorily predict the P80. Keywords: Product particle size, variable importance measurement, Hilbert-Schmidt independence criterion, mill charge

Jan 1, 2020