Ore Grade Estimation Research Articles

A Chart for Judging Optimal Sample Spacing for Ore Grade Estimation: Part II

Resources of mineral deposits are commonly estimated using data extracted from rock cores obtained by drilling. Information of good quality and quantity is essential to achieve good estimates (unbiased and precise). Frequently, the geomodeller is required to provide an estimate of the grade of each block to be mined. Some deposits have more erratic spatial grade distributions than others, and therefore a better understanding of closely spaced samples is required. Given the importance of samples, due to their influence on the precision and accuracy of the model built from these collected data, it is deemed important to optimize their spacing (or sampling grid). The idea is to obtain information which will lead to the desired model precision while minimizing the overall sampling cost. Commonly, mineral deposit regions with high grade variability require closely spaced samples. In this paper, the variability is measured by the grades’ coefficient of variation. A chart relating the sample spacing to the estimation error for various data coefficients of variation and block geometries of three different deposits is proposed. A case study illustrates the results obtained for a copper grade, an iron grade, and an aluminum oxide grade. The aim of this paper is to establish a fast and reliable way to obtain the maximum error for a given sample spacing that may be used for short-term geological modeling and mine planning at different ore deposits.

A multidisciplinary approach considering geochemical reorganization and internal structure of tailings impoundments for metal exploration

Tailings deposits may host a significant amount of valuable elements and are of great interest as exploration targets for secondary mining. However, elements are distributed heterogeneously within an impoundment due to physical and chemical processes during and after deposition. Therefore, the complex internal structure has to be considered for representative sampling, estimation of ore grade and to prevent dilution of metals by bulk mining. This study shows that a multidisciplinary approach and a combination of methods are necessary to take account of that. The selected methods comprise surface sampling with few drillings, geochemical and granulometric analysis and Electrical Resistivity Tomography (ERT). They provide important information for a quick decision about the economic viability of further detailed exploration or even secondary mining. In order to understand the internal structure of tailings impoundments collection of 1D and 2D information from sampling and drilling should be accompanied by 3D resistivity mapping. In this work, indications for particle size distribution, metal distribution and water movement were extracted from the different data sets and a model for the internal structure and re-organization is proposed. General zones of metal enrichment and depletion can be projected from the model. Furthermore, water saturation and movement as an important factor for acid rock drainage and tailings dam stability can be deduced for an estimation of the environmental risk of the tailings impoundment. Therefore, the proposed combination of methods covers both the estimation of economic potential as well as the analysis of environmental impacts of tailings deposits.

Comparison of machine learning methods for copper ore grade estimation

In this study, machine learning methods such as neural networks, random forests, and Gaussian processes are applied to the estimation of copper grade in a mineral deposit. The performance of these methods is compared to geostatistical techniques, such as ordinary kriging and indicator kriging. To ensure that these comparisons are realistic and relevant, the predictive accuracy is estimated on test instances located in drill holes that are different from the training data. The results of an extensive empirical study in the Sarcheshmeh porphyry copper deposit in Southeastern Iran illustrate that specially designed Gaussian processes with a symmetric standardization of the spatial location inputs and an anisotropic kernel yield the most accurate predictions. Furthermore, significant improvements are obtained when, besides location, information on the rock type is included in the set of predictor variables. This observation highlights the importance of carrying out detailed studies of the geological composition of the deposit to obtain more accurate ore grade predictions.

Near Real-Time Classification of Iron Ore Lithology by Applying Fuzzy Inference Systems to Petrophysical Downhole Data

Fluctuating commodity prices have repeatedly put the mining industry under pressure to increase productiveness and efficiency of their operations. Current procedures often rely heavily on manual analysis and interpretation although new technologies and analytical procedures are available to automate workflows. Grade control is one such issue where the laboratory assay turn-around times cannot beat the shovel. We propose that for iron ore deposits in the Pilbara geophysical downhole logging may provide the necessary and sufficient information about rock formation properties, circumventing any need for real-time elemental analysis entirely. This study provides an example where petrophysical downhole data is automatically classified using a neuro-adaptive learning algorithm to differentiate between different rock types of iron ore deposits and for grade estimation. We exploit a rarely used ability in a spectral gamma-gamma density tool to gather both density and iron content with a single geophysical measurement. This inaccurate data is then put into a neural fuzzy inference system to classify the rock into different grades and waste lithologies, with success rates nearly equal to those from laboratory geochemistry. The steps outlined in this study may be used to produce a workflow for current logging tools and future logging-while-drilling technologies for real-time iron ore grade estimation and lithological classification.

Geostatistical and GIS analysis of the spatial variability of alluvial gold content in Ngoura-Colomines area, Eastern Cameroon: Implications for the exploration of primary gold deposit

Linear and nonlinear geostatistic is commonly used in ore grade estimation and seldom used in Geographical Information System (GIS) technology. In this study, we suggest an approach based on geostatistic linear ordinary kriging (OK) and Geographical Information System (GIS) techniques to investigate the spatial distribution of alluvial gold content, mineralized and gangue layers thicknesses from 73 pits at the Ngoura-Colomines area with the aim to determine controlling factors for the spatial distribution of mineralization and delineate the most prospective area for primary gold mineralization. Gold content varies between 0.1 and 4.6 g/m3 and has been broadly grouped into three statistical classes. These classes have been spatially subdivided into nine zones using ordinary kriging model based on physical and topographical characteristics. Both mineralized and barren layer thicknesses show randomly spatial distribution, and there is no correlation between these parameters and the gold content. This approach has shown that the Ngoura-Colomines area is located in a large shear zone compatible with the Riedel fault system composed of P and P′ fractures oriented NE-SW and NNE-SSW respectively; E-W trending R fractures and R′ fractures with NW-SE trends that could have contributed significantly to the establishment of this gold mineralization. The combined OK model and GIS analysis have led to the delineation of Colomines, Tissongo, Madubal and Boutou villages as the most prospective areas for the exploration of primary gold deposit in the study area.

Application of K-means and PCA approaches to estimation of gold grade in Khooni district (central Iran)

Grade estimation is an important phase of mining projects, and one that is considered a challenge due in part to the structural complexities in mineral ore deposits. To overcome this challenge, various techniques have been used in the past. This paper introduces an approach for estimating Au ore grades within a mining deposit using k-means and principal component analysis (PCA). The Khooni district was selected as the case study. This region is interesting geologically, in part because it is considered an important gold source. The study area is situated approximately 60 km northeast of the Anarak city and 270 km from Esfahan. Through PCA, we sought to understand the relationship between the elements of gold, arsenic, and antimony. Then, by clustering, the behavior of these elements was investigated. One of the most famous and efficient clustering methods is k-means, based on minimizing the total Euclidean distance from each class center. Using the combined results and characteristics of the cluster centers, the gold grade was determined with a correlation coefficient of 91%. An estimation equation for gold grade was derived based on four parameters: arsenic and antimony content, and length and width of the sampling points. The results demonstrate that this approach is faster and more accurate than existing methodologies for ore grade estimation.

A hybrid simultaneous perturbation artificial bee colony and back-propagation algorithm for training a local linear radial basis neural network on ore grade estimation

In this paper, a local linear radial basis function (LLRBF) neural network that uses a skewed Gaussian activation function, called LLRBF-SG, is applied to the problem of ore grade estimation of highly skewed data from the Esfordi phosphate deposit. The network is trained using SPABC-BP, a method that combines a novel simultaneous perturbation artificial bee colony algorithm (SPABC) and back-propagation (BP) method. The SPABC algorithm is an extension of the standard artificial bee colony (ABC) algorithm that includes a tournament selection strategy, simultaneous perturbation stochastic approximation method and new search equations. The predictive accuracy of the network trained with the SPABC-BP algorithm is compared with hybrid versions of evolutionary and swarm intelligence algorithms, such as standard artificial bee colony, covariance matrix adaptation evolution strategy (CMAES) and particle swarm optimization (PSO) with BP. From the experimental results one concludes that networks trained with the SPABC-BP algorithm outperform networks trained with the alternative algorithms in the process of ore grade estimation. The predictive accuracy of the LLRBF-SG-SPABC-BP is compared with LLRBF-SPABC-BP and standard radial basis function (RBF) trained with SPABC-BP algorithm. An analysis of the results shows that the proposed LLRBF-SG network has higher generalization ability and is better suited to the problem of predicting ore grade values for highly skewed data.

A Chart for Judging Optimal Sample Spacing for Ore Grade Estimation

Mineral deposit grades are usually estimated using data from samples of rock cores extracted from drill holes. Commonly, mineral deposit grade estimates are required for each block to be mined. Every estimated grade has always a corresponding error when compared against real grades of blocks. The error depends on various factors, among which the most important is the number of correlated samples used for estimation. Samples may be collected on a regular sampling grid and, as the spacing between samples decreases, the error of grade estimated from the data generally decreases. Sampling can be expensive. The maximum distance between samples that provides an acceptable error of grade estimate is useful for deciding how many samples are adequate. The error also depends on the geometry of a block, as lower errors would be expected when estimating the grade of large-volume blocks, and on the variability of the data within the region of the blocks. Local variability is measured in this study using the coefficient of variation (CV). We show charts analyzing error in block grade estimates as a function of sampling grid (obtained by geostatistical simulation), for various block dimensions (volumes) and for a given CV interval. These charts show results for two different attributes (Au and Ni) of two different deposits. The results show that similar errors were found for the two deposits, although they share similar features: sampling grid, block volume, CV, and continuity model. Consequently, the error for other attributes with similar features could be obtained from a single chart.

Ore grade estimation of a limestone deposit in India using an Artificial Neural Network

This study describes a method used to improve ore grade estimation in a limestone deposit in India. Ore grade estimation for the limestone deposit was complicated by the complex lithological structure of the deposit. The erratic nature of the deposit and the unavailability of adequate samples for each of the lithogical units made standard geostatistical methods of capturing the spatial variation of the deposit inadequate. This paper describes an attempt to improve the ore grade estimation through the use of a feed forward neural network (NN) model. The NN model incorporated the spatial location as well as the lithological information for modeling of the ore body. The network was made up of three layers: an input, an output and a hidden layer. The input layer consisted of three spatial coordinates (x, y and z) and nine lithotypes. The output layer comprised all the grade attributes of limestone ore including silica (SiO2), alumina (Al2O3), calcium oxide (CaO) and ferrous oxide (Fe2O3). To justify the use of the NN in the deposit, a comparative evaluation between the NN method and the ordinary kriging was performed. This evaluation demonstrated that the NN model decisively outperformed the kriging model. After the superiority of the NN model had been established, it was used to predict the grades at an unknown grid location. Prior to constructing the grade maps, lithological maps of the deposit at the unknown grid were prepared. These lithological maps were generated using indicator kriging. The authors conclude by suggesting that the method described in this paper could be used for grade-control planning in ore deposits.

Relevance vector machines using weighted expected squared distance for ore grade estimation with incomplete data

Accurate ore grade estimation is crucial to mineral resources evaluation and exploration. In this paper, we consider the borehole data collected from the Solwara 1 deposit, where the hydrothermal sulfide ore body is quite complicated with incomplete ore grade values. To solve this estimation problem, the relevance vector machine (RVM) and the expected squared distance (ESD) algorithm are incorporated into one regression model. Moreover, we improve the ESD algorithm by weighting the attributes of the data set and propose the weighted expected squared distance (WESD). In this paper, we uncover the symbiosis characteristics among different elements of the deposits by statistical analysis, which leads to estimating certain metal based on the data of other elements instead of on geographical position. The proposed WESD-RVM features high sparsity and accuracy, as well as the capability of handling incomplete data. Effectiveness of the proposed model is demonstrated by comparing with other estimating algorithms, such as inverse distance weighted method and Kriging algorithm which utilize only geographical spatial coordinates for inputs; extreme learning machine, which is unable to deal with incomplete data; and ordinary ESD based RVM regression model without entropy weighted distance. The experimental results show that the proposed WESD-RVM outperforms other methods with considerable predictive and generalizing ability.

Ore Grade Estimation Based on Multi-gene Genetic Programming

Ore Grade Estimation Based on Multi-gene Genetic Programming

Improving the intelligent methods of management of processes of enrichment on the basis of visiometrics analysis of the grade of the ore

For the flotation process is effective using of open control loops, which are used as input parameter the ore quality. To determine the quality of ore is carried out on-line analysis of its elemental and mineral composition and structure. The basis of the operational analysis is the measurement of X-ray and visible light. To improve the accuracy of determining ore mineralogical composition, new techniques and facilities have been developed. At the conveyor ore analysis, special lighting systems were tested, using alternating focused and flat light fluxes. Such systems provide highly accurate in-stream measurements of ore properties. A special flatbed facility for optical spectrumbased estimation of fine-crushed ore has been developed, being incorporated into the existing system of ore sampling and analysis. Application of the flatbed facility renders possible a high accuracy of determining ore mineralogical composition. Modern systems for optical spectrum-based estimation of ore grade provide a basis for effective automated control of ore beneficiation processes, based on the principle of advanced ore grade control. The systems implemented at the Erdenet processing plant (Mongolia) contributed to increasing copper and molybdenum recovery.

Robust LS-SVM regression for ore grade estimation in a seafloor hydrothermal sulphide deposit

Due to the geological complexities of ore body formation and limited borehole sampling, this paper proposes a robust weighted least square support vector machine (LS-SVM) regression model to solve the ore grade estimation for a seafloor hydrothermal sulphide deposit in Solwara 1, which consists of a large proportion of incomplete samples without ore types and grade values. The standard LS-SVM classification model is applied to identify the ore type for each incomplete sample. Then, a weighted K-nearest neighbor (WKNN) algorithm is proposed to interpolate the missing values. Prior to modeling, the particle swarm optimization (PSO) algorithm is used to obtain an appropriate splitting for the training and test data sets so as to eliminate the large discrepancies caused by random division. Coupled simulated annealing (CSA) and grid search using 10-fold cross validation techniques are adopted to determine the optimal tuning parameters in the LS-SVM models. The effectiveness of the proposed model by comparing with other well-known techniques such as inverse distance weight (IDW), ordinary kriging (OK), and back propagation (BP) neural network is demonstrated. The experimental results show that the robust weighted LS-SVM outperforms the other methods, and has strong predictive and generalization ability.

Hybrid self-adaptive learning based particle swarm optimization and support vector regression model for grade estimation

Ore grade estimation is one of the key stages and the most complicated aspects in mining. Its complexity originates from scientific uncertainty. In this paper, a novel hybrid SLPSO–SVR model that hybridized the self-adaptive learning based particle swarm optimization (SLPSO) and support vector regression (SVR) is proposed for ore grade estimation. This hybrid SLPSO–SVR model searches for SVR's optimal parameters using self-adaptive learning based particle swarm optimization algorithms, and then adopts the optimal parameters to construct the SVR models. The SVR uses the ‘Max-Margin’ idea to search for an optimum hyperplane, and adopts the ε-insensitive loss function for minimizing the training error between the training data and identified function. The hybrid SLPSO–SVR grade estimation method has been tested on a number of real ore deposits. The result shows that method has advantages of rapid training, generality and accuracy grade estimation approach. It can provide with a very fast and robust alternative to the existing time-consuming methodologies for ore grade estimation.

Advanced System for Control Milling and Flotation Processes Based on an Estimation of Ore Quality Grade

Ore processing plays a very important linking role between the mining and the metallurgical steps. The recent R&D achievements in this field resulted in the increase of the intellectual content of technological processes of mineral processing, as well as its automatic control. The main reason of mineral processing fluctuations is a mixing of ores from various sections of a deposit. Instability and non-optimized parameters of milling and flotation take 3% to 6% of losses of valuable component. Milling and flotation process are characterized by considerable fluctuations of all input, output and intermediate parameters. In these conditions, applying various mathematic models the process controlling by action on input parameters with further precise reaching the required values of output parameters of the process is not always possible. The application of the multilevel process models allows elaborating and using methods of the ore grade estimation and evaluation of the technological process situation by adaptive control.

Spectral pathways for effective delineation of high-grade bauxites: a case study from the Savitri River Basin, Maharashtra, India, using EO-1 Hyperion data

Bauxite, the only source of aluminium, is an aggregate of minerals, most of which are oxides and hydroxides of aluminium and iron such as gibbsite, bohemite, goethite and haematite. Bauxite is used in the chemical and refractory industries and its quality is controlled by the presence of impurities such as iron and silica. Bauxite commonly occurs together with iron-rich laterites as alteration products of parental igneous and metamorphic rocks. Aluminium-rich bauxites grade towards highly ferruginous laterites with a transitional Al-rich laterites or ferruginous bauxite, herein described as Al-laterites. In the Savitri River Basin, bauxite contains 58–75% gibbsite, 6–11% goethite and 19–26% haematite, whereas the mineralogy of Al-laterites and Fe-laterites are dominated by haematite (29–68%) and goethite (6–25%) with subordinate amounts of gibbsite. Conventional techniques to demarcate the high-grade pockets of bauxites rich in gibbsite are tedious, time consuming and involve detailed field sampling and geochemical analyses. Our work illustrates how spectral properties of these three litho-units can be effectively utilized in mapping of high-grade bauxites occurring over wide areas using hyperspectral remote sensing (HRS). The methodology adopted herein involves generation of noise-free field spectral database of target materials, linear unmixing of field spectra for constituent minerals, classification of preprocessed Hyperion images using field spectra and finally accuracy assessment for ore grade estimation. It is observed that bauxite mapping using Hyperion data and noise-free field spectra yielded results that correlate well with the chemistry and mineralogy of representative samples. By adopting the above procedure, we achieved classification accuracies of 100%, 71% and 89% for bauxite, Al-laterite and Fe-laterite classes, respectively.

A hybrid neural networks-fuzzy logic-genetic algorithm for grade estimation

The grade estimation is a quite important and money/time-consuming stage in a mine project, which is considered as a challenge for the geologists and mining engineers due to the structural complexities in mineral ore deposits. To overcome this problem, several artificial intelligence techniques such as Artificial Neural Networks (ANN) and Fuzzy Logic (FL) have recently been employed with various architectures and properties. However, due to the constraints of both methods, they yield the desired results only under the specific circumstances. As an example, one major problem in FL is the difficulty of constructing the membership functions (MFs).Other problems such as architecture and local minima could also be located in ANN designing. Therefore, a new methodology is presented in this paper for grade estimation. This method which is based on ANN and FL is called “Coactive Neuro-Fuzzy Inference System” (CANFIS) which combines two approaches, ANN and FL. The combination of these two artificial intelligence approaches is achieved via the verbal and numerical power of intelligent systems. To improve the performance of this system, a Genetic Algorithm (GA) – as a well-known technique to solve the complex optimization problems – is also employed to optimize the network parameters including learning rate, momentum of the network and the number of MFs for each input. A comparison of these techniques (ANN, Adaptive Neuro-Fuzzy Inference System or ANFIS) with this new method (CANFIS–GA) is also carried out through a case study in Sungun copper deposit, located in East-Azerbaijan, Iran. The results show that CANFIS–GA could be a faster and more accurate alternative to the existing time-consuming methodologies for ore grade estimation and that is, therefore, suggested to be applied for grade estimation in similar problems.

Ore grade estimation by feature selection and voting using boundary detection in digital image analysis

In mining, rock classification plays a crucial role at different stages of the extraction process ranging from the design of the mine to mineral grading and plant control. In this paper we present a new method to improve rock classification using digital image analysis, feature selection based on mutual information and a voting process to take into account boundary information. We extract rock color and texture features and using mutual information we selected 14 from 36 features to represent the data in a lower dimensional space. The original image was divided into sub-images that are assigned to one class based on the selected color and texture features using a set of classifiers in cascade. Additionally, using rock boundary information, a voting process for the sub-images within the same blob is performed. We compare our results based on sub-image classification to those obtained after the voting process and to those previously published on the same rock image database. We show that the RMSE on rock composition classification on a test database decreased 8.8% by using our proposed voting method with the automatic segmentation with respect to direct sub-image classification. The RMSE decreased 29.5% relative to previously published results with the same database using a mixture of dry and wet rock images. The RMSE decreased even more if we considered separately dry and wet rocks. Our proposed method could be implemented in real-time to estimate mineral composition and can be used for online ore sorting and/or classification.

Limestone chemical components estimation using image processing and pattern recognition techniques

In this study based on image analysis, an ore grade estimation model was developed. The study was performed at a limestone mine in central Iran. The samples were collected from different parts of the mine and crushed in size from 2.58 cm down to 15 cm. The images of the samples were taken in appropriate environment and processed. A total of 76 features were extracted from the identified rock samples in all images. Neural network used as an intelligent tool for ore grade estimation and the features of every image were combined with weighted average method. In order to feature dimensional decrease, principal component analysis method was used. Six principal components, which were extracted from the feature vectors, captured 90.661% of the total feature variance. Components were used as the input to neural network and four grade attributes of limestone (CaCO3, Al2O3, Fe2O3 and MgCO3) were used as the output. The root of mean squared error between the observed values and the model estimated values for the test data set are 6.378, 4.847, 0.1513 and 0.0284, the R2 values are 0.7852, 0.8663, 0.7591and 0.8094 for the mentioned chemical composition respectively. The magnitude of R2 indicates the correlation between actual and estimated data. Therefore, it can be inferred that the model can successfully estimate the limestone chemical compositions percentage.

Adaptive ore grade estimation method for the mineral deposit evaluation

Ore grade estimation is one of the most key and complicated aspects in the evaluation of a mineral deposit. Its complexity originates from scientific uncertainty. This paper introduces a new nonlinear and adaptive method to the problem of ore grade estimation, which is based on the Wavelet Neural Network (WNN) approach, and is designed to receive drill hole information from an orebody and perform ore grade estimation on a block model basis. The nonlinear ore grade estimation method combining the properties of the wavelet transform and the advantages of Artificial Neural Networks (ANNs) provides fast and reliable ore grade estimation, with minimum assumptions and minimum requirements for modeling skills. A number of case studies have been carried out using the new ore grade estimation method. The results obtained and the overall functionality of the method prove that Wavelet Neural Networks can offer a fast and robust grade estimation technique and a valid alternative to well established methodologies in this area.