Geometallurgical Model Research Articles

Geometallurgical modeling and reconciliation with production data in a fhosphate mine

The characterization of the ore and its performance prediction in the processing unit are essential for the success of mining projects and should be part of the resource model. The greatest difficulties encountered in the elaboration of these models are the characteristics of the metallurgical variables, such as non-additivity and non-linearity, lack of uniformity of the database and the few samples of process tests. The generalization of geometallurgical models from primary attributes is a valid option when the relationships between geological characteristics and metallurgical responses are well established and will be tested in this work. Bench tests that simulate the mineral concentration process were performed and compared with geological and chemical information. Multivariate statistical analyses and computational algorithms allowed the prediction of metallurgical attributes from the properties of the rocks simulated by Gaussian sequential simulation. The results were reconciled with the plant data over a period of one year and proved to be consistent.

A Quantitative Particle-Based Approach for the Geometallurgical Assessment of Tailings Deposits

Tailings generated during ore processing may host significant residual contents of valuable commodities, including critical metals. The particle properties of the tailings, such as mineralogy, particle size, and the surface liberation of ore minerals, strongly control processing behaviour. This study explores a novel combination of methods for incorporating particle data, derived from automated mineralogy, into geometallurgical models of tailings deposits to better understand their reprocessing potential and the economic feasibility of re-mining. This was achieved through binning of different particle types, geostatistical modelling of particle bin frequencies, and bootstrap resampling to reconstruct particle populations. The spatial distributions of processing-relevant particle properties throughout the tailings deposit were predicted with corresponding uncertainties. There are clear systematic trends in the spatial distributions of different particle types, resulting from the sedimentary-style deposition of the tailings. For instance, the tailings nearer the dam walls comprise coarser, silicate-rich particles, while fine-grained and well-liberated sulphide mineral particles are more abundant in the centre of the tailings deposit. As a result, robust models could be developed for the spatial distributions of particle size and mineralogy, which strongly control the sorting of particles during deposition, and other related properties, such as sulphide mineral grain sizes. Finally, a bulk sulphide flotation process was simulated and acid mine drainage potential estimated using the interpolated particle data. Around 58% of the sulphide minerals present could be recoverable by flotation, with the recoverable sulphide portion decreasing towards the centre of the TSF due to the fine-grained nature of the sulphide minerals. The acid mine drainage potential of the tailings is estimated to be moderate to high, indicating that the carbonate minerals present are not sufficient to neutralise the high acid-generating potential of the sulphide minerals. Overall, this study demonstrates how particle-based geometallurgical models can be developed and utilised for practical applications, with the aim of improving the accuracy of resource and reserve estimations of tailings deposits and the sustainable and responsible management of anthropogenic resources. The methodology proposed here can be easily transferred to other tailings deposits.

Contributions of computing in geosciences: a case study of geometallurgical forecasting

Geometallurgical modeling has been improved over the years with the incorporation of mineralogical, chemical, and metallurgical information. However, the diversity of the database brings challenges since the information is not unified in procedures, sample support, or quantities. Characteristics such as nonlinearity and nonadditivity of the geometallurgical database impose even more difficulties. The objective of this article is to demonstrate the feasibility of geometallurgical models from multivariate techniques combined with computational tools. The algorithms used in this study were developed in Python programming language. The multivariate analysis and machine learning techniques allowed the grouping of transitional typological domains and the quantification of uncertainty; the selection and reduction of the dimensionality of the heterotopic database; and the prediction of metallurgical behavior from exhaustive primary variables using a transfer function. The results were verified and proved to be satisfactory. Other benefits of the computational structure were: reduction of subjectivity; automation; and dynamic updates of models using new data.

Predictive insights for copper recovery: A synergistic approach integrating variability data and machine learning in the geometallurgical study of the Tizert deposit, Morocco

This work involved the application of machine learning (ML) to predict copper recovery through variability data in the context of the geometallurgical study of the Tizert deposit in Morocco. For this purpose, geological and metallurgical variability investigations were conducted on drilling samples from Tizert ore, leading to the consolidation of a robust database for ML algorithms. The dataset comprised 345 observations across seven variables, namely GPS coordinates, geological units, lithological facies, the main copper deportment, and metallurgical responses from flotation tests, as well as the copper grade and the oxidation rate. The geological and mineralogical variability revealed two geological units: The Basal Series and Tamjout Dolomite unit, hosting copper mineralization dominated by chalcocite and malachite. Besides, the metallurgical evaluation demonstrated recovery variability, primarily influenced by oxidized copper. Copper oxides exhibit less efficient flotation than sulfides, with recovery decreasing as the oxidation rate increases. The prediction of copper recovery was performed using the variability data through the XGBoost algorithm. The resulting model showed good performance on the test dataset, with a coefficient of determination R2 of 0.83 and a RMSE of 2.47, improving the accuracy of copper recovery predictions. In addition, the model elucidated the most significant variables impacting Tizert ore behavior, notably the casing, copper-bearing mineral, and oxidation rate. These variables were considered key factors for advancing Tizert ore valuation and geometallurgical model development. Overall, metallurgical modeling based on the ML approach provided an innovative way to integrate variables into mine planning and processing. The use of such a predictive model gave valuable insights for improving efficiency, understanding the metallurgical behavior, optimizing production, and reducing costs along the value chain of the Tizert deposit. Further data processing efforts are underway, to improve the assessment of copper recovery quality for the Tizert ore and the development of a complete geometallurgical model on a plant scale.

Predictive Geometallurgical Modeling for Flotation Performance in Mixed Copper Ores Using Discriminatory Methods

Predictive Geometallurgical Modeling for Flotation Performance in Mixed Copper Ores Using Discriminatory Methods

Uncertainty and Value: Optimising Geometallurgical Performance Along the Mining Value Chain

To maximise the value of a mining operation and minimise its environmental and social impacts, all processes—from the ore deposit to the final product and waste streams—should be optimised together. However, mining and metallurgical processes are inherently variable and uncertain due to the natural heterogeneity of ore deposits and the limited information and incomplete models available on ore behaviour throughout the process chain. Propagating these effects to geometallurgical models is important because they are used to make decisions with potentially large environmental and economic impacts. In this paper, we describe the need for geometallurgical optimisation routines to account for the effects of uncertainties, and the tools needed to manage them, by summarising the routines that already exist and those that are still missing.

Data-Driven Synthesis of a Geometallurgical Model for a Copper Deposit

Geometallurgy integrates aspects of geology, metallurgy, and mine planning in order to improve decision making in mining schedules. A geometallurgical model is a 3D space that is typically synthesized from early-stage small-scale samples and is composed of several metallurgical units, or domains. This work explores the synthesis of a geometallurgical model for a copper deposit using a purely data-driven unsupervised approach. To this end, a dataset of 1112 drill samples is used, which are clustered using different methods, namely, k-means, hierarchical clustering (AGG), self-organizing maps (SOM), and DBSCAN. Two cluster validity indices (Silhouette and Calinski–Harabasz) are used to select the final model. To validate the potential of the proposed approach, a simulated economic evaluation is conducted. Results demonstrate that k-means exhibits a better performance in terms of modeling and that using the obtained geometallurgical model for mining scheduling increases the project’s Net Present Value (NPV) by as much as 4%. Based on these results, the proposed methodology is an appealing alternative for generating geometallurgical models within greenfield, brownfield and ongoing operations.

Water consumption assessment in mineral processing integrating weather information and geometallurgical modeling

Water is essential in mineral processing. Because water is a limited resource and many mining operations are located in water-scarce areas, proper water management is crucial. An alternative to facing water scarcity in mining is its reuse in the process. Water recirculation is limited because there are water losses associated with the plant operation and the tailings facility. Tailings composition and particle size affect the losses related to entrainment and rewetting. These parameters can be estimated by simulating the plant using geometallurgical models. Weather variables such as radiation, temperature, wind speed, and the fraction of possible sunshine, impact the evaporation expected in the tailings facility. The variability of weather conditions throughout the year implies that evaporation depends on the season. This information is used to assess the water balance in a tailings facility located in a specific region, estimating the water consumption for a given operation. Water consumption can be economically valued and included in the mine plan. This methodology is used to provide examples where a mine plan is optimized incorporating water consumption and using efficient water management criteria.

Recursive Feature Elimination and Neural Networks Applied to the Forecast of Mass and Metallurgical Recoveries in A Brazilian Phosphate Mine

Geometallurgical models are commonly built by combining explanatory variables to obtain the response that requires prediction. This study presented a phosphate plant with three concentration steps: magnetic separation, desliming and flotation, where the yields and recoveries corresponding to each process unit were predicted. These output variables depended on the ore composition and the collector concentration utilized. This paper proposed a solution based on feature engineering to select the best set of explanatory variables and a subset of them able to keep the model as simple as possible but with enough precision and accuracy. After choosing the input variables, two neural network models were developed to simultaneously forecast the seven geometallurgical variables under study: the first, using the best set of variables; and the second, using the reduced set of inputs. The forecasts obtained in both scenarios were compared, and the results showed that the mean squared error and the root mean squared error increase in all output variables evaluated in the test set was smaller than 2.6% when the reduced set was used. The trade-off between simplicity and the quality of the model needs to be addressed when choosing the final neural network to be used in a 3D-block model.

Geoenvironmental characterization of two REE deposits: the Montviel carbonatites and Kipawa silicates, Quebec Canada

Drainage water quality is the significant environmental concern for the rare earth elements (REE) mining industry. REE deposits are associated with other metals and radioactive bearing minerals. REE mining and refining activities can generate significant quantities of liquid and solid wastes. Therefore, a long-term integrated approach covering the full mine-life cycle is required to mitigate possible environmental concerns. In the present study, two REE concentrates were prepared and all deposit lithologies of carbonatites and silicates sampled and investigated for their mineralogy, geochemistry, and their environmental behavior using kinetic testing. For the Montviel carbonatite (enriched in light rare earth elements, or LREE), the majority of REE-bearing minerals are associated with carbonates (i.e., monazite, kukharenkoite, burbankite, etc.), whereas the REE-bearing minerals associated with the Kipawa silicates (enriched in heavy rare earth elements, or HREE) are fluorbritholite, eudyalite, mosandrite, etc. The kinetic tests showed a neutral to alkaline pH of leachates and a low leachability of REE (carbonatites <140 μg/L; silicates <15 μg/L) with a higher mobility of HREE than LREE. The reactivity of REE carbonates are one to two orders of magnitude higher than REE silicates. For sustainable mineral development, geological and environmental data was integrated into the geometallurgical model to identify and control the environmental risks associated with mining those two deposits.

Process simulation to determine blending and residence time distribution in mineral processing plants

Mineral processing plant performance depends on multiple factors, including the feed and the parameters to control the process. In this work, we show how to assess plant performance using geometallurgical modeling and dynamic simulation. Several models that describe comminution, classification, flotation, and residence time distribution (RTD) are implemented as modules and then connected to represent generic plant configurations. The estimation of the RTD is used to assess the ore blending generated within the plant through a methodology based on weighting the ore contribution at the plant discharge. Additionally, the RTD is used to display the ore permanence at different plant stages, which can be used as an operational input to anticipate the consequences of a perturbation in the feed. Different simulation scenarios are tested using synthetic data, including different plant configurations, time support for blending assessment, and ore feeding sequence. The results show that the simulation is sensitive to these attributes. Significant differences are detected in the generated product compositions when the plant configuration is changed. Also, distinct mine plans can be evaluated through simulation, predicting their processing performance. Therefore, the simulation tool developed can be used to evaluate real mineral processing operations and to test different operative strategies.

Metal(loid) Deportment in Sulfides from the High-Grade Core of the Bingham Canyon Porphyry Cu-Mo-Au Deposit, Utah

Abstract Porphyry deposits supply the bulk of the world’s Cu and Mo and significant amounts of Au, as well as other minor and trace metal(loid)s, including Ag, Re, Te, Pd, Se, Bi, Zn, and Pb. Porphyry deposits are gaining in importance as a source of critical raw materials with the increasing global demand for these commodities. To date, minor and trace metal(loid)s are still commonly recovered as by-products from porphyry ores without prior characterization of their host mineralogy that could inform more efficient processing and improved recoveries. We report a comprehensive metal(loid) deportment study on a complete vein paragenetic series in samples from the northwestern high-grade zone of the Bingham Canyon Cu-Mo-Au porphyry deposit, Utah. The polyphase Bingham stock comprises an early premineralization equigranular monzonite phase that was intruded by a series of five successive, ore-related porphyry intrusions. Veins with hypogene Cu-(Fe) sulfide assemblages from all five porphyry intrusions were characterized for their trace metal(loid) contents by laser ablation-inductively coupled-mass spectrometry (LA-ICP-MS). It was found that bornite and digenite contain elevated Bi, Ag, Te, and Se relative to chalcopyrite, whereas the latter contains elevated concentrations of Co, Ga, and In. A stepwise decline in sulfide abundance occurs over the porphyry intrusion sequence and is more pronounced in digenite and bornite than in chalcopyrite. The related diminishing concentration per rock volume (inventory) of Bi, Ag, Te, and Se in the youngest porphyry dikes could have been caused in part by a geochemical change in the mineralizing fluid supply across successive intrusive-hydrothermal cycles. Element mapping of exsolved digenite within bornite revealed characterstic partitioning of metal(loid)s between bornite and digenite; most notably Ag, but also Te and Au are enriched in digenite relative to enclosing bornite. Bornite domains within these composite grains reveal complex zonation of Sn, In, and Bi, which are attributed to stress-induced diffusion within bornite, resulting from the digenite exsolution process. The selective partitioning of metal(loid)s between bornite and digenite is likely a common feature in many porphyry Cu deposits, given the fundamental mineralogical characteristics of these two sulfides. Our results contribute to an improved understanding of the distribution (from mineral to deposit scale) of critical trace metal(loid)s in porphyry deposits, particularly those containing exsolved digenite. This knowledge can be applied to determine more accurately the value of ore resources, to improve geometallurgical models and by-product recoveries, and to help limit the environmental effects of metal(loid) dispersion.

Validation of predictive flotation models in blended ores for concentrator process design

The use of geometallurgical modelling is becoming common in the design of a concentrator plant and during its operational phase. Predictive modelling aims to define optimized process parameters for the concentrator plant to accommodate the variability of the ore feed characteristics. The main objective of this study is to simulate the metallurgical response for distinct ore unit types and their blends using kinetic models derived from experimental flotation tests. Kinetic flotation tests, including both rougher and cleaner, were applied on four ore types containing different ratios of chalcopyrite, chalcocite and copper oxide minerals from Rich Metal Group’s Madneuli copper–gold mine, Georgia. The metallurgical performance of complex copper ores Madnueli XI and V differs from each other significatively due to the different mineralogical distribution of copper sulphides and oxides. The blend of ore units in various ratios affects the efficiency of the process by decrementing the recovery and mass pull due to copper oxide-sulphate in the ore Madnueli V. HSC Chemistry® software was used to simulate the metallurgical response of blended ores based on mineral compositions of the ores, Klimpel kinetic rectangular distribution model and flowsheet model. Simulated and experimental results of copper grade, copper recovery and concentrate mass pull correlated well with R2 values of over 0.95. These results demonstrate that the metallurgical response of ore blends can be simulated based on the flotation kinetics of distinct ore types. The use of predictive simulation can create savings during the test work phase and increase the flexibility and certainty in the process design stage.

Unsupervised textural classification of rocks in large imagery datasets

• Rock imaging characterisation techniques capture composition and texture. • A new feature extraction method for rock texture from imaging data is documented. • A large dataset of hyperspectral rock images was classified using this new method. • The results prove a robust and repeatable approach for textural analysis. • Textural outputs can be incorporated in geometallurgical modelling. The value of rock characterisation, whether it is for mineral exploration , extraction, or concentration, lies in the ability to describe its composition and texture. In the past century, extensive technological developments have provided new opportunities to assess compositional properties, both geochemical and mineralogical, and at different scales: X-ray diffraction, X-ray fluorescence, inductively coupled plasma mass spectrometry , reflectance spectroscopy , etc. More recently, the advent of imaging characterisation techniques and high-performance computing power has enabled the assessment of mineral texture in a robust and quantitative manner. This study proposes and validates an end-user focused workflow for the identification of textural families in a large drill-core hyperspectral imagery dataset, based on a novel textural feature extraction method named Mineral Co-Occurrence Probability Field (MCOPF). This workflow combines vintage image textural assessment methods with modern machine learning techniques for the automated unsupervised classification of textures within a drill core hyperspectral imagery dataset. The results demonstrate a meaningful and robust identification of rock textural families (clusters), enabling a wide range of applications in geology, mining, and metallurgy in the academic and industrial sectors.

MLA and optical microscopy as complementary techniques to the iron ore geometallurgical studies

Mineral liberation is an important variable to be considered in the iron ore geometallurgical studies, especially since it provides information leading to the understanding of the ore’s behaviour in the beneficiation process, mainly when harder ores are concerned. Nowadays, the professionals and researchers in the mineral industry have been using complementary techniques for mineralogical characterization. In this way, Mineral Liberation Analyser (MLA) and optical microscopy are amongst the most efficient and important tools in these studies. Together, these techniques can provide information leading to the development of geometallurgical models, including the degree of quartz liberation, mineral modal composition, mineralogical association and particle and mineral size distribution, as well as the spectrum of mineral liberation either by free surface or particle composition. The purpose of this study is to compare the degree of quartz liberation in four different iron ore types observed in optical microscopy and MLA. It also assesses the influence of mineral liberation in the quality of the resulting concentrate, according to the SiO2 content. All the results of the various quartz liberation degrees, determined by both MLA and optical microscopy were very close, as well as the correlation between the mineral liberation and the ore’s response in the flotation process. The investigation provided evidence that the mineral liberation study is an important tool in the prediction of the ore’s behaviour in the beneficiation process.

Determination of Abrasiveness in Copper-Gold Sulfide Ores: A Contribution to the Geometallurgical Model of the Sossego Deposit

The geological context of this study is established in the iron oxide-copper-gold (IOCG) deposit of Sossego (Canaã dos Carajás, Brazil), where hydrothermal alterations in shear zones concentrated the metals of interest and added new characteristics to the metavolcanic-sedimentary and granite rocks. The mineral transformation of rocks by hypersaline fluids enriched in metals and silica also modifies some metallurgical properties, such as abrasiveness. Special bench tests on rock drill cores are used in mapping the abrasiveness of rocks, with the Bond abrasion test being more commonly used in the mining industry, but it has a restrictive sampling protocol and mass requirement for geometallurgical studies. As a counterpoint, the test of the Laboratoire Central des Ponts et Chaussées/Central Laboratory of Bridges and Roads (LCPC) requires a smaller amount of fine material and a finer granulometric range. The study on the use of LCPC was implemented in 40 samples, using Bond Ai as a reference. The results showed a strong correlation between both methodologies (R2 = 95%), validating the use of LCPC to quantify abrasiveness in the Sossego mine. It was also possible to classify the most abrasive lithologies.

Understanding gangue acid consumption in copper sulfide heap leaching: Predicting the impact of carbonates, silicates and secondary precipitates

A geometallurgical model, linked to simple ore body characteristics, was established to understand short- and long-term gangue acid consumption (GAC) variability for a new copper sulfide chemical heap leaching process under standard metallurgical conditions, including an acid dose of 12 kg/t sulfuric acid during agglomeration at a particle size of 80% passing ¾”. Carbonate availability, or liberation, defined as a function of feed grade and alteration type, enabled the calculation of short-term GAC when combined with mineral abundance, assuming stoichiometric behavior. An adjustment to the acid balance needed to be applied for jarosite formation, which was made possible via a linear correlation, again related to carbonate feed content. The long-term GAC component of the ore could then be determined by difference, comparing data from the column program to the estimation of the two previous variables, resulting in the identification of biotite and chlorite as key drivers of long-term acid consumption during leaching. Final inputs for the three-stage approach to predict overall GAC were simplified to carbonate (CO3), magnesium (Mg), and iron (Fe), facilitating implementation in the 3D resource block model. Application of this methodology indicated that the magnitude of long-term GAC is reduced when higher levels of acid are consumed during the initial phases of processing. The study clearly shows how a holistic knowledge of the ore body can provide an in-depth understanding of future processing response, enabling key investment decisions to be made appropriately, prior to mine construction.

Rock-chemistry-to-mineral-properties conversion: Machine learning approach

A problem of prediction of ore minerals properties by whole-rock chemistry (47 elements) has been stated and solved for a case of the Kovdor baddeleyite-apatite-magnetite deposit (Russia, Murmansk region). Four regression methods have been used: linear multiple regression, artificial neural networks, random forests, and multivariate adaptive regression splines. The latter has turned out to be the best. Content of major and trace elements in the economic minerals, average grain size, and textural properties (intergrowths of the ore minerals with gangues) have been predicted with fairly high accuracy. Subsequently, a problem of finding an optimal number of predictors (i.e., a number of elements in whole-rock chemistry analyses) has been investigated to simplify regression models and, therefore, reduce time and cost of the prediction of ore mineral properties. We have found that 5–9 predictors are enough to predict one parameter of an ore mineral. It allows using the approach both for geometallurgical modelling of a deposit and for real-time control of ore quality and/or mineral processing.

The double wheel breakage test

The mining industry needs a low-cost and reliable breakage characterization test which can rapidly process a large number of samples for geometallurgical modelling. The more samples are tested, the better is the understanding of the ore hardness variability and lower the design or production risks. The proposed solution is a new testing device, herein named Geopyörä, which is a variation of a roll crusher with an adjustable gap and instrumentation to measure breakage forces and energy applied to rock particles during the breakage process. The principle utilises a controlled degree of crushing, with absorbed breakage energy being a response rather than an input. The new testing device is capable of rapidly testing rocks over a wide range of sizes and accurately measured energy levels. For a range of ores, the results were demonstrated to provide outputs that replicate the breakage modelling from full JK drop Weight tests. In addition to being suited to testing drill cores and small sample masses, the Geopyörä provides a distribution of particle strengths within every sample. This paper provides an introduction to the concept, development of the prototype device and breakage calibration results that indicate its potential to become a major player in geometallurgical ore testing.

Uncertainties in quantitative mineralogical studies using scanning electron microscope-based image analysis

Scanning electron microscope-based automated mineralogy studies are readily associated with quantitative results, providing one of the foundations of geometallurgical studies. Despite the importance of quantitative data for such studies, and efforts to reduce statistical errors, the reporting of uncertainties is rare. This contribution illustrates how bootstrap resampling can be used to provide robust estimates of statistical uncertainties for the modal mineralogy, metal deportment and all relevant textural attributes of a sample or a series of samples. Based on a case study of the Bolcana Au-Cu porphyry deposit in the South Apuseni Mountains, Romania, the impact of insufficient sampling statistics on quantitative mineralogical studies is illustrated. Quantitative analyses of the mineralogy and microfabric of milled ore samples from seven 40 m drill core intervals from the Bolcana Prospect were conducted using a Mineral Liberation Analyser (MLA), complemented by electron probe micro-analysis. Bootstrap resampling was then applied to assess how many grain mount surfaces should be analysed to achieve statistically robust results for both Cu and Au mineralogy, deportment and textural attributes. Despite variable mineralogy, grades and mineralisation styles, estimated statistical uncertainties on Cu deportment are consistently low. In contrast, uncertainties for Au deportment are so high that most reported values for important characteristics are statistically meaningless. This is mainly attributed to the pronounced nugget effect for Au mineralisation, exacerbated by the small sample size analysed by MLA. An unfeasible number of measurements would be necessary to provide robust figures for the deportment of minor/trace elements and minerals, along with other tangible mineralogical properties, such as mineral associations. The results of this case study demonstrate that statistical uncertainties need to be carefully incorporated when considering the results of automated mineralogical studies and their impact on geometallurgical models. This is particularly relevant for studies of precious metal ores.