An iron ore mine prepares pre-crusher stockpiles, feeding the processing plant with ore at target composition, in grade (iron and multiple contaminants) and physical and source characteristics. Weekly, ore blocks to be mined are allocated to the previous stockpile, or to a new stockpile. By week end, the previous stockpile is completed. The new stockpile becomes the starting stockpile for the following week. An Excel-based program using Visual Basic macros allocates ore to the stockpile builds, selecting mining blocks to build the stockpiles, so completed stockpiles are close to target grade and composition. It sequences the stockpile builds, so the stockpile has consistent grade and composition throughout its build, maintaining control under early termination of stockpile build or reclamation. Initial solution using the Excel add-on ‘Solver’ exceeded the integer variable limit, so was replaced by a greedy algorithm solving the problem much faster, without the limitation on the integer variables.

Maximising ore tonnage at a selected target grade requires that appropriate ore identification criteria be established. Iron ore mining block models may consist of many millions of estimated blocks, each block corresponding to the smallest mining unit, with grade interpolated from drillhole data. Each block can be considered as a potential ore candidate. Analysis of such a huge database can require large computational resources, especially for sensitivity analysis requiring multiple computations. However, an alternative streamlined approach is possible, where the data are compressed to an appropriately smaller and representative subset. The consequent data reduction not only makes it feasible to consider the simultaneous analysis of multiple pits but also considerably speeds up computation, enabling sensitivity analysis of the potential ore tonnage as a function of the required grade. Ore identification criteria can be successfully established using appropriately chosen reduced data comprising less than one per cent of the blocks. The ore identified is virtually indistinguishable in tonnage and grade from the ore that would be selected using the entire block model. Two methods of data reduction are tested: sampling (using a proportion of unaltered blocks), and binning (consolidating groups of blocks to a single average grade). The source blocks are first sorted (such as by location, analyte grade, Principal Component score, random order or value) before sampling or binning. In this study, the sorting options considered were by location and Fe grade. The effectiveness of each method was evaluated using the block model subsets to select the maximum tonnage of ore at a specified target grade. Block model data from an anonymous Pilbara iron ore project with about one million blocks averaging 1.132 kt were used. The blocks are from multiple resource models, to be blended to a single product. Sampling is shown to be superior to binning, because sampling preserves the full variability while binning, although matching the block mean grade, reduces the block variability. It was found that the appropriately sampled data sets (sorted by location) gave results virtually identical to the results for the full data set, even when the samples comprised less than one per cent of the original blocks. The reduction in computer time to identify maximum tonnage at a specified grade was approximately proportional to the sample percentage. The increase in computation speed greatly facilitated sensitivity analysis. A sampled data set of one-eighth the total block model was used to investigate the effects of changing each of the component grades around the original grade specification. The results of this sensitivity analysis are discussed, and illustrate the potential benefit of carrying out sensitivity analysis in determining feasible and marketable product grades.

Ore selection is generally aimed at producing product with a consistent target grade matching a perceived market requirement. A composite selection criterion 'Comp' (a linear function of the grade vector) can be used to maximise ore tonnage at a specified target grade (in Fe and contaminants) extractable from a block model. The blocks correspond to the selective mining units, with grades interpolated from diamond drill-hole data. An iterative procedure has been developed to identify the coefficients and cut-off value for the composite function 'Comp', maximising potential ore tonnage at target grade. Commonly, target grade is pre-determined by marketing. However, there is no guarantee that maximising ore tonnage at a specified target grade maximises the potential value of the mine. If the value of ore could be expressed as a linear function of the grade components, then these coefficients would be appropriate for the selection criterion 'Comp'; the cut-off value for 'Comp' would correspond to the marginal cost of production; all blocks with a 'Comp' score above the cut-off would be classified as ore, and the cumulative grade of these ore blocks would be the average product grade over the life of the mine. Unfortunately, no such value function for iron ore is generally available. Methods for estimating the value function are discussed, using the spot price and sensitivities based either upon operational tolerances for Fe and the contaminants or upon published price differentials. This approach has been applied to an anonymous Pilbara deposit with about one million blocks averaging 1.1 kt. It is shown how the optimal target grade, cut-off grade and maximum tonnage can all be explored as a function of marginal cost and iron ore price. Alternatively, given marginal cost and an iron ore price, the effects of changing the tolerances (and therefore the relative costs of the contaminants) can also be explored. The procedure can enable informed decision-making for an iron ore mine. Rather than setting final policy for the entire life-of-mine, it enables target grades and tonnage estimates to be revised by re-computation as the mine life proceeds, to recognise changes in market conditions and knowledge about the resource. It should be emphasised that this paper is concerned with ore selection rather than ore sequencing. Appropriate ore selection is a necessary but not sufficient condition for maximising value. Having identified the set of blocks that potentially maximise value, the sequence in which they are mined will present a trade-off between producing at constant grade and extracting high-value ore early. The sequencing issue is not considered here. It is shown that optimum ore selection for an open-pit mine is independent of any consideration of discount rates, although ore sequencing will depend upon the time value of money if the product grade is to vary across time. Although the study considers an iron ore deposit, the method can be extended to any open-pit mining where ore value can be expressed as a linear composite of multiple grade components.

Author Summary: The use of pre-crusher stockpiles to store ore and buffer short-term fluctuations in production processes is generally well recognised and accepted. However, the potential to reduce short-term grade variation of ore entering the crusher is rarely recognised and generally poorly understood. Pre-crusher stockpiles are commonly built and reclaimed in an ad-hoc manner whereas well-designed and disciplined build and reclaim procedures can reduce variability into the crusher at low cost. Design options for pre-crusher stockpiling should consider the four competing roles of storage, buffering, blending and grade control, to produce predictable and uniform crusher feed grades. The selection of alternative grade allocation methods requires careful consideration, as decisions at this early stage of the production process have been shown to flow on to shipping and to the customer. This paper reports conclusions from studies simulating the reduction of grade variability for a range of alternative pre-crusher stockpiling configurations and grade allocation methods. The benefits achievable in reducing grade variance by systematically building stockpiles of appropriate dimension are quantified.

Simulation modelling is a practice commonly used in the mining industry to evaluate alternative process designs. Such modelling is typically undertaken as an optimisation study to increase the efficiencies, productivity and product quality of operating mines. In this situation real short-term grade variability data of extracted ore are available from production records as input data into the simulation. There is also a need for simulation modelling to be performed before mines are approved for construction to clarify the grade variability characteristics that can be expected from the operating mine and assist in the optimisation of the process design. In this situation no historical short-term grade variability data are available. To achieve meaningful and reliable results from a simulation it is necessary to have input data for the ore to be extracted from the pits, representative of the short-term grade variability that would be expected for the operating mine. This paper describes an example of a path taken to generate realistic input data. A method described as composite cut-off criterion was used to distinguish ore from waste which gave substantially greater recovered tonnage at target grade compared to the conventional quadrant cut-off grade criteria. The only data available for the project was resource model data in the form of kriged block models, which are known to underestimate true ore grade variability. To achieve realistic results from the simulation it was essential to increase this variability while maintaining the accurate average grades. Areas of certain deposits that were modelled using both kriging and conditional simulation estimation techniques were quantitatively compared to establish the comparative variance and the kriged data was modified to match the conditionally simulated variance. A realistic mining model was generated via a process of discretisation and regularisation of the resource blocks. Quantitative assessment demonstrated that this method adequately compensated for ore dilution and that adjustment for ore loss was not required due to an ore skin surrounding the edge blocks. The conditioned data were then used to generate a schedule of daily mine extraction, considering grade variability, tonnage, equipment constraints and extensive blended-in-blended-out pre-crusher stockpiles to feed into the process design simulations. For the pre-crusher stockpiles a number of alternative allocation criteria were examined for ore being extracted from the pits to identify the best method in achieving reduced variability through the daily scheduling system. The study concluded that a single analyte separation criterion produced acceptable variability with minimal complexity. The blending efficiency of manually stacked and reclaimed pre-crusher stockpiles was studied to determine a realistic blending efficiency within the pile. A recommended method of building and reclaiming was determined to give maximum blending efficiency. Finally, the data were used as input into process design simulation models; simulation from crusher feed to ship loading demonstrated that control of shipment grade variability was achievable and that the conditioning of the data delivered realistic results.

This study considers realistic data for a planned open-pit iron ore mine, but is applicable to any open pit situation. By interpolating drill hole data, grades are generated for a rectangular block model. Each block’s grade vector has components for each analyte (chemical element or compound) influencing ore value. Deriving the average (E-type) over many simulations leads to each block being assigned its expected value, and thus underestimates the overall grade variability. Alternatively, interpolation by means of conditional simulation is a method that implements random sampling from an infinite population of solutions. Each conditional simulation has appropriate overall grade variability, but estimating any block’s mean and variance requires sampling from multiple conditional simulations. For a block model, an ore/waste selection criterion maximises the expected tonnage at a target grade. This criterion is a linear composite of the grade components, with positive coefficients for the beneficial analyte (Fe) and negative coefficients for the deleterious analytes (such as SiO2, Al2O3 and P). Although conditional simulation often leads to a similar expected grade as kriging for each block, the expected maximum tonnage of ore selectable at a target grade may differ from that obtainable from the E-type solution. We apply the linear composite selection criterion to each of 25 conditional simulations, as well as to the E-type block model. Simulation confirms the distribution of product tonnage to have an expected tonnage that is over 20% greater than that of the E-type model. The method also enables a selection probability to be computed for each block, and thus a probabilistic pit boundary distribution to be identified and used in mine planning. Proposed extensions to this method will consider risk-based scheduling of the multiple selection solutions through minimisation of a derived stress factor and treating the mining process as an iterative system with actual or artificial depletions modelled in line with the mine plan, using the updated state (with new information) to re-evaluate the mine plan for subsequent periods.

Pre-crusher stockpiles are designed principally as buffers to decouple the mining and processing operations. They are usually paddock dumped or dumped over a face to form fingers by dumping haul truckloads and reclaimed by front-end loader in an ad hoc manner. In addition, they are often not built and reclaimed to completion but continually added to and extracted from. Whereas this design suits the mining process and is operationally simple, there can be little confidence in the grade of the ore that is fed to the crusher. This makes short-term grade control difficult and reconciliation back to the mine face imprecise. Well-designed and operated pre-crusher stockpiles can overcome these deficiencies. The different types of grade variability in mining are reviewed. The role of pre-crusher stockpiles in reducing short-term variability is discussed and their ineffectiveness in addressing long-term variability is highlighted. Their role in removing the serial correlation of the extracted ore is explained. Pre-crusher stockpiles carry out the four, at times competing, objectives of storing, buffering, blending and grade separation. Their actual design and operation result from a compromise between these four competing roles. These roles are explained in detail along with the advantages and disadvantages of the different types of pre-crusher stockpiles. The recommended blended-in blended-out stockpile (BIBO) design is discussed in detail. Matched pairs of BIBO stockpiles of limited tonnage built and reclaimed to completion retain the best knowledge of grade and facilitate the reconciliation process. Simulation studies have been carried out on BIBO stockpiles to understand the blending opportunities of the different options for building and reclaiming, and to identify the methods that achieve maximum blending. It was found that laying down rows in one direction and reclaiming across these rows gives the best blending. Matching the width of the reclaim face with the daily crushing requirements is also important. The length of the rows, and building in either one or two layers has little effect on the extent of blending. Appropriate design can provide a reasonable compromise between the four objectives.

When ore is being selected from a regularised block model, a common procedure is to accept as ore any block whose iron content exceeds a cut-off value, and whose content in each contaminant analyte is below an appropriate cut-off value. This method defines a quadrant within the multidimensional analyte space, such that any block lying within the quadrant is accepted as ore and any block lying outside the quadrant is rejected as waste. The cut-off levels can be adjusted to maximise ore tonnage at a target grade (or analyte vector). When the method is applied to a project comprising multiple pits with systematically differing grades, it is found that ore tonnage at target grade is maximised if different cut-off quadrants are used for the different pits. This is apparently illogical, because it implies that an ore block from one pit may be accepted as ore, but an identical block from another pit rejected as waste, even when ore from both pits is to be blended into the same product and there are no production cost differences between the pits. The paradox is an example of the Theory of the Second Best: if there are multiple conditions for optimality and one condition is broken, then obeying the other conditions may not be the best policy. The problem lies with the quadrant cut-off method, which is shown to be inconsistent even for selecting ore from a single pit. A criterion that is inconsistent cannot be optimal. It is shown that a linear composite cut-off function provides maximum ore tonnage, and that a single composite cut-off criterion is consistent and optimal over multiple pits. Further, it is shown that a composite cut-off function not only maximises ore tonnage at a specified composition, but also can be used to select ore so as to maximise project value, if marginal costs and values are available.

As an external consultant advising industry (mainly the mining industry), two contrasting experiences are frequent. One is that the industry practitioners have strong domain knowledge that the consultant lacks. The other is that the industry practitioners often suffer from paradigm paralysis, using models of thinking that are either fundamentally inappropriate, or, perhaps more commonly, have become inappropriate through the passage of time and changes in technology and computing power. Although, and perhaps because, the data available have often grown considerably, paradigm paralysis can cause decision makers to fail to make full use of the available information, or worse, can lead to outcomes contrary to those intended. By coming in as an outsider, external consultants can provide benefit by introducing new paradigms or updating previous paradigms, subject to the important proviso that they become familiar with the domain knowledge and work closely with the industry practitioners. In this paper, the author discusses a number of projects, in which he has been involved, which illustrate the occurrence of paradigm paralysis.

In response to the implementation of a new schedule for iron ore fines in the International Maritime Organization's International Maritime Solid Bulk Cargoes Code, improved measures for the management of moisture have been developed. In particular, prediction of cargo moisture allows management of iron ore fines within the supply chain to facilitate the safe shipping of iron ore fines. An Autoregressive Integrated Moving Average model, augmented to allow for shipment tonnage, has been developed to predict the moisture level from previous shipments of the same product. The model explains about 60% of the moisture variance. It was found that inclusion of other available information, such as ore composition, train/rake assays and recent rainfall at port and mine did not add to the predictive power of the model.

At Cliffs Natural Resources Ltd (CNR) crushing and shipping sample stations, precisions are estimated for the sampling, preparation and measurement stages of iron ore sampling in accordance with ISO 3085 (2002). In this standard, gross duplicate samples are prepared at each of the three stages, yielding eight measurement assays. Appropriate calculations of the differences between assay pairs are then made to estimate precision of each stage of sampling. The international standard for checking precision in iron ore sampling, ISO 3085 (2002), prescribes a method for identifying outliers that is ambiguous and possibly inappropriate. The procedure also makes an unnecessary implicit assumption, that the differences between assay pairs are normally distributed. The Anderson-Darling statistic is used to demonstrate that in some cases the assay pair differences have distributions significantly different from normal. When applied at CNR as specified, the standard leads to precision estimates well below the expected values, thus overestimating the sampling method's capability and limiting the opportunity for real process improvement. This paper suggests an improved method of estimating sampling, preparation and measurement precision. A bootstrap procedure is used to estimate confidence limits for the calculated precision estimates. The proposed method would be suitable for testing blast hole precision. The discussion is supported by extensive simulation modelling of realistic data.

The exponentially weighted moving average (EWMA) can be used to report the smoothed history of a production process, and has some considerable advantages over a simple moving average (MA).Discussion of these advantages includes comparison of the filter characteristics of the EWMA and MA in the frequency domain.It is shown that the EWMA provides a much smoother filter than does the MA, and the corresponding implications of this difference are examined in the time domain.In smoothing a production process, the successive entities being smoothed commonly have varying "weights", where the weights may be such quantities as tonnage, value or time interval.Standard textbook treatments of moving averages and exponential smoothing are generally confined to equal spaced data of equal weight.Adapting the average to cope with items of varying weight is shown to be trivial for the case of MA, but is not so obvious for the EWMA.This paper shows how the exponential smoothing constant has to be adapted to provide a consistent EWMA.Applications of the EWMA in process control are discussed, with particular reference to quality control in the mining industry.

Most internationally traded iron ore resources are derived from multiple mine sites, transported to a port facility where they are blended and possibly further processed before shipping. Provided the proportion of contributions varies over a number of shipments, an estimate can be made of the bias introduced by mine site sampling, analysis and materials handling systems, using zerointercept multiple linear regression. Further regression applied to the residuals allows the error variances for the source mines to be estimated. Various sampling strategies are used by the industry. All have shipment loading sampling on which payment or at least provisional payment is based. In most cases, sampling pre-stockpile is used at mine sites to verify estimated production grades. Train grades are estimated from mine stockpile grades and in some cases train grades are verified by port input sampling. The methodology used is presented in this paper. Simulated data have been used and do not reflect the actual performance of any operator.

A block model of the ore body is used to plan a mine. The first requirement is to identify ore to be extracted and discard waste, so as to meet target grade, generally in multiple analytes. Commonly, cut-off values are set for each analyte, so as to distinguish ore from waste. It will be shown that, if more than one analyte is important, this procedure is wasteful of ore and a composite cut-off function is preferable. A second requirement is to sequence the ore extraction so that the variability in ore grade is controlled: failure to do so will result either in low-quality ore being marketed, or excessive re-handling being required to blend the ore to reduce the grade variability. A third requirement is that the extraction sequence should be such as to limit the amount of movement of equipment to enhance the productivity of the equipment. The overall objective, to optimise the Net Present Value of the mined ore, is simply stated but complex in realisation, since so many factors, such as target grade, grade variability, equipment choice, equipment movement, and downstream blending all have alternatives which can be traded off against each other, and all of which contribute to the costs and benefits making up the total Net Present Value. There exist commercial packages for ore selection and mine planning. However, they are often of a black box nature (i.e. not transparent), and users are sometimes not fully aware of the criteria being applied. In particular, the treatment of quality as a multidimensional vector may be problematic. The purpose of this paper is to discuss some of the issues involved and to suggest an alternative set of approaches, which should complement existing commercial treatment. The discussion will be illustrated specifically by reference to the mining of iron ore, but the issues are relevant to a wide variety of mining situations.

An essential requirement of product quality control in the mining industry is to be able to reliably predict key quality properties of finished product from the data available before the extraction of the ore. From a production viewpoint, the unit of data collection is generally the input and output data set for each shift of crusher production but could be any period where mine pre-crusher data can be reliably matched with product data. Linear regression models can be used to predict crush grades from blast grades, even where the crush material is blended from multiple sources or pits for each of which differing regression models might apply. The best model for any application will be a balance between required predictability, available data and the tolerance of the business for complex models. The regression modelling approach has several advantages over the classic method of run of mine crusher trials. The models can use any predictor variable such as grade, geotype and in situ density provided the pre-extraction data can be reliably matched with post-crusher data and is significant as a predictor. The models have been used extensively in the generation of the daily crusher plan with the aim of maintaining finished product grade. This approach has also been used associated with exploration drilling and long term planning. It is acknowledged that there are inherent problems in fitting lump and fines grade to a linear model. However, these problems are minor when such information is used for interpolation within the window spanned by the shift blend records used to produce the model. This paper discusses some of the issues limiting linear regression models in this application, and suggests methods enabling consistent models to be formulated.

This paper describes a technique to predict the head grades and likely variances from target grade that can be achieved from mining an iron ore deposit. The method develops a production schedule from a block model. The example used was created from a wide-spaced drilling program and was classified as an Inferred Resource, but the method can be applied to any block model based on the available combination of Proved and Probable reserves. Marketing contracts require examination whether the target grade, i.e. Fe, P, SiO2, Al2O3, can be kept constant, or must be revised during the life of the mine. A model (written in Excel using Visual Basic macros, details available from the first author) develops a feasible, close to optimal, mine plan from block model data. Cut-off grades for minable ore are chosen. The model evaluates ore tonnage, grade and stripping ratio against depth. Outputs at this stage include plots of slices through the ore body, at nominated x, y and z coordinates, showing ore grade, waste, drill-hole locations and topography. The model then searches for a feasible satisfactory mine plan. The initial grade target is the average for the identified ore blocks. The available block list (ABL) at any time is the set of ore blocks, any one of which can be mined without removing any other ore block. The entire initial ABL has an unacceptably large total stress, defined as [(Grade–Target)/Tolerance]2, summed over the four relevant minerals. The model trims the ABL, removing the block most harming the total stress. Trimming is repeated until the Total Stress reduces to an acceptable level. The trimmed ABL defines the first mine campaign. Removing the trimmed block set, a second ABL is exposed, which is again trimmed and mined. The process is automatic: a few seconds computation provides a complete mine plan. Re-running the model with adjusted grade limits quickly shows whether a uniform target can be achieved or whether the target must be modified during the mine life.

Cliffs Natural Resources Pty Ltd (CNR) operates iron ore mines in the Koolyanobbing region of Western Australia, similar to 50 km north of the town of Southern Cross. Ore is trucked from three geographically isolated sources to the crusher at Koolyanobbing, where it is blended before and during crushing. Lump and fine products are produced and railed to Esperance for ship loading and export to Asian customers. The CNR is examining alternative processing paths, from mining to ship loading, with the aim of improving efficiency and reducing costs. Modifications to the system must be consistent with potential future expansions and maintain the low intershipment grade variability on which CNR prides itself and has built a strong relationship with its customers. In searching for the optimum process design, many options from mine face to ship loading must be evaluated and compared. Pilot plant studies are infeasible, while complex mineralogical interactions, competing goals and numerous possible system configurations limit the applicability of theoretical analysis. It was therefore concluded that simulation modelling would provide the confidence to take the next step into production trials. This paper describes techniques applied at CNR to simulate grade variability resulting from potential process design changes. The simulation models are easily run Excel based modules, with each module representing a different part of the process. The modules use extensive Visual Basic macros driven by Excel's user friendly interfaces. Presentation of the results is enhanced by Excel's excellent graphical capabilities. The simulation software stores and graphically presents time stamped data from a run, enabling detailed analysis of different process configurations. Final success of a simulation run is measured by intershipment variability (standard deviation and process capability) and in process ore tonnages. Meaningful results from the simulations require that the initial input data contain the same correlations present in the real production environment, between the mineral components, production linkages and across time. The data also have to allow simulation of potential changes to mining method and introduction of new pits into the blend. Mining data from the real operations under study are therefore used, with average grades and variability adjusted to match potential future development proposals. It is also necessary to filter out medium and long term variations from the production data, as this variability is best controlled through the conventional medium to long term mine planning process, not by the process design being studied. The filtering was carried out using a Fourier transform technique, which is described. For reasons of commercial confidentiality, detailed data, costs and quantitative conclusions are not reported in this paper.

Curcas Energy is developing a sustainable energy project in Thailand to produce Jatropha oil as a biological source of diesel fuel. The company is providing Jatropha seeds and advice to smallholder farmers. The crop is being grown as hedgerows on otherwise unused land. The nuts will be transported from widely distributed collection points to central plants where the oil will be extracted. The extraction plants can be any integer multiple of a basic unit, with larger plants having advantages of scale that have to be weighed against the lesser transport costs to smaller more widely distributed treatment plants. This paper describes a modelling tool, developed in Excel, which enables planners to compare the costs and benefits of alternative plant locations and sizes. The collection point locations and estimated production rates are entered into the model, together with potential treatment plant locations, costs and capacities, and transport costs per kilometre. The model computes the preferred plant destination for each collection point, after taking into account production rates and treatment capacities, and plots a map of all the collection points and plants, colour coded to indicate the destination plants for each collection point. The planner can iteratively adjust the plant locations, capacities and costs to explore the wide range of suggested alternatives to be considered.