Search Documents

By P C. J Blackney, I M. Glacken

Spotted Quoll is a massive sulfide nickel deposit located in the Forrestania greenstone belt in the Archaean Yilgarn block in Western Australia. Discovered in October 2007, mining has progressed from open pit to underground and currently uses longhole open stoping with paste fill as the primary stoping method. Immediately prior to 2019, the Mineral Resource estimate (MRE) was defined by a single grade domain of nickel greater than or equal to 0.8 per cent, irrespective of the host lithology, a legacy of the original ore wireframe being built using explicit methods. While a number of obvious structural offsets were honoured in this nickel grade domain, only one modelled fault was used to undertake domaining of arsenic. From 2019 onwards, implicit modelling in Leapfrog Geo™ was used to construct a mine-scale geological model, which triggered the definition of a high-grade core of massive/matrix nickel sulfides surrounded by a low-grade halo of stringer/disseminated nickel sulfides in the surrounding rock. As part of the evolution of the geological model and with the aid of Seequent Central™ to allow collaboration between site-based mine geologists, Perth-based resource geologists and Melbourne-based consultants, a detailed review of the structural mapping was carried out as additional information from underground development became available. This allowed numerous faults and shears to be identified – particularly in the lower half of the mine. These were used to define further structural domains and to refine the understanding of the nickel and arsenic grade distributions in the mineralised envelope. The increased definition to the domains provided by the geological model resulted in a significant improvement in the accuracy of the Mineral Resource estimate, as measured by reconciliation using voids of the mined stopes determined from the cavity monitoring survey (CMS).

Mar 22, 2022

By P Micale

Geological data provides the foundations for the mining value chain. It is used to underpin the modelling and planning and consequently optimise downstream business decisions. However, it is often forgotten or taken for granted. End users often assume the data to be sound without verification. Geological data can be considered a company’s number one asset. Subsequently, the storage, use and management of this data need careful curation and time invested in preserving and protecting it. An established polymetallic mine with 20 000 drill holes could have a data replacement value in excess of $500M at 2020 prices, excluding labour and salary costs. The Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves (the ‘JORC Code’) provides guidelines for Public Reporting. Within the code, Table 1 requires discussion around the different types of geological data, its collection and use. What is needed is additional details regarding the validation and integrity of the geological database. A comprehensive guideline specifying minimum geological standards for geological databases has been developed. This expands on the critical points outlined in Table 1 of the JORC Code. This guideline can be used by companies where geological databases are managed in-house or used to assess these services’ outsourcing. This guideline will ensure that geological data is reliable and unadulterated and can be used with confidence. Using this guideline will assist in preserving the investment of geological data in the mining value chain

Mar 22, 2022

By R Hargreaves, T Elkington, W J. Shaw, G W. Booth

Every day, around the world, thousands of tons of ore are loaded onto ships completing the miningproduction process and starting a multimillion-dollar economic transactions process. These transactions are based on the supply of tons of a commodity with certain specifications of assays of the main element (or physical properties) and concentration of contaminants, which are finally verified and evaluated by buyers, and also are considered ‘as true’ during Reconciliation process. On Sample Stations, this ‘true’ is based on a few kilograms of sample, often representing thousands of tons, that are sent to laboratories for further sample preparation and chemical analysis, where finally assay information is obtained from a few grams of pulverised sample, and are used for further production, planning, operational and financial-marketing decisions. Different International Standards (depending on the commodity) are used to establish the methodology to be followed, to set-up and obtain samples from the Sample Stations, and the specifications are visually inspected to give an estimation on the performance of the sampling point. This paper seeks to highlight with examples, the importance of the implementation of a continuously monitored Quality Program (QAQC and QM) on Sample Stations, focused on the parameters that directly affect the quality and representativeness of the samples. By this implementation, a proactive approach and better-informed decisions can be taken for mining companies to: a) quantify/improve production and reconciliation results, and b) to supply the right specifications of shipped ore, increasing the value of the Resources. Finally, laboratory results are normally used as an indication of the quality of the information applied for the economic transaction, but it should be considered, clarified and highlighted that the laboratory will only process and analyse a sample obtained from a Sample Station, and potential problems on bias or precision may occur in stages prior to the laboratory reception. Hence the importance of the implementation of a quality assurance program that protects multi-million dollar decisions.

Mar 22, 2022

By P M. Hetherington, M P. Murphy, K E. Crook, F Pym

Resource characterisation is critical for informed decisions in both strategic and operational space. The sources of information for resource characterisation are limited by: direct measurement of resource attributes, such as grades, rock-types, mineralogy, and are available only from sparse locations, ie drill holes; or indirect geophysical measurements such as seismic, gravity or electro[1]magnetic survey. Data collection tools, methods, and processing technologies are constantly improving providing a broad spectrum of options for data collection strategies. The cost of each data survey varies from thousands to millions of dollars with the value of each additional data having diminishing return for a particular business decision. The value of information analysis provides the optimal decision on data collection strategy within a limited available budget. This work presents an example of value of information analysis for new Blast Hole Assay Tool (BHAT) and Reverse Circulation Grade Control (RCGC) data used in addition to the wider spaced exploration data. The value of information is quantified in terms of reduction of grade uncertainty, ore loss and dilution by comparing different combinations of data collection methods. Four different strategies were considered: exploration samples in addition to 50 per cent of blastholes surveyed with BHAT, exploration samples in addition to 33 per cent of blastholes surveyed with BHAT, exploration with wide-spaced RCGC data, and exploration without additional grade control. For each strategy the observed values have been simulated using advanced multivariate geostatistical simulations and compared with simulated ground truth models. The calculated uncertainty, ore loss and dilution informed the optimal data collection strategy. Opportunity to reduce sample density from the nominal 50 per cent of drill holes was evaluated.

Mar 22, 2022

By A Boucher, M Baudin, P Androszczuk, I Minniakhmetov, A Rigby, P Machado, M Wright, B Van Dijken

IGO Ltd’s (IGO’s) Nova Operation (Nova), which is 160 km east-north-east of the town of Norseman in Western Australia (WA), mines and processes nickel-copper-cobalt (Ni-Cu-Co) sulfide ore from the magmatic Nova-Bollinger Deposit (Nova-Bollinger), which is hosted by the Mesoproterozoic age high-grade metamorphic rocks of the Albany Fraser Orogen. Over its 56-month life-of-mine (LOM) operations to 31 December 2020, Nova has processed 5.63 Mt of ore grading 2.04 per cent Ni, 0.86 per cent Cu and 0.07 per cent Co from its underground mine. At the end of 2020, both the Nova-Bollinger JORC Code reportable Mineral Resource estimate (MRE) and Ore Reserve estimate (ORE) were >90 per cent classified in the highest JORC Code confidence categories of Measured Mineral Resource and Proved Ore Reserve respectively. The high JORC Code classification confidences assigned to the Nova-Bollinger MRE and ORE results from the 386 km of surface and underground diamond drill testing of the deposit to a nominal 12.5 by 12.5 m pierce point spacing in mineralisation. The close spaced high-quality drilling and bulk mining method employed at Nova has effectively removed the need for any grade control modelling outside of the MRE process. Importantly, having only the one fully drilled out deposit being mined and only the one process plant, Nova’s reconciliation process is in the fortunate situation of not being confounded by issues such as external ore sources and constant updates as new grade control information becomes available. IGO’s Nova reconciliation processes are focused on reconciliation ratios that quantify operational period differences between the MRE and ORE head grade and tonnage forecasts and the process plant head actuals. These metrics have been used to identify key reconciliation issues and have initiated several studies into the causes for specific reconciliation results, which are discussed in the paper. In late 2020, a period of sustained poor reconciliation results led to the recognition that increasing the understanding of risk between the MRE and ORE should be investigated. To assist with modelling stope level risk within the mine plan and subsequent reconciliation, an attempt was made to develop a formula to quantify ‘stope risk’ from MRE metrics such as sample spacing, kriging efficiency and so on

Mar 22, 2022

By S L. Martin, M Wimberger

The binary indicator coding of assay and lithological data according to a single criterion or multiple criteria is a powerful technique for representing mineralised shapes through the modelling of probabilities. This can be as simple as using a single indicator to model the probability of occurrence of a condition, usually a grade threshold or a lithology, but it can also be used with a number of categories. Case studies are presented to show the power of this approach, where the geological controls on mineralisation or on the distribution of a rocktype or alteration, often coupled with mineralisation, are unclear. The categorical approach is often used as a de facto grade filter in disseminated or stockwork mineralisation. The ultimate expression of the indicator formalism is in multiple indicator kriging (MIK), where a conditional distribution of probabilities is generated for each estimation panel. This has applications in recoverable resource estimation, as the input for continuous or categorical simulation, where the distribution of sample grades is multimodal, or when there are superimposed mineralisation events with different orientations. However, in some cases, MIK is being used as an excuse for ignoring a quality geological interpretation. Examples from the public domain show the abuse of the categorical and MIK approaches to modelling probabilities, together with a plea for greater consideration of the geology.

Mar 22, 2022

By O Dominguez

Mine value chain reconciliation (MVC) is a necessary procedure to evaluate a mine’s performance against the assumptions and models used to plan mining and processing operations. Using nomenclature from Parker (2012), F2 reconciliation is the ratio between material received at the mill, compared with material sent to the mill, commonly reported in terms of tonnes, grade and metal. Measurements of material sent to the mill can be estimated in many ways, including: fleet management system (FMS) records, truck counts, or volume reduction. Most time-based reconciliations (for example, month-end reconciliation) use short-term model depletions of in situ rock measured by periodic topographic scans or surveys.

Mar 22, 2022

By H Kurth

Oyu Tolgoi copper and gold mine in the southern Mongolia holds mineral resources that make it one of the largest known copper and gold deposits in the world. The project is operating an open pit mine on the near surface of Southern Oyu deposit and developing underground block-caving mine at the Hugo Dummet North where 80 per cent of total resources exists. Oyu Tolgoi (OT) commenced open pit mining in 2011 and implemented an industry standard blasthole sampling, and ordinary kriging grade control estimation with block-out process. Since 2017, grade control improvement projects have successfully completed that have led to the implementation of current, improved grade control system. Geological model is updated by blasthole logging on Leapfrog® and used for daily grade estimations. GC estimation is proceeded solely on Vulcan® by running Lava script and integrating with a relational database. Combination of block-out procedure produces GC model which is updated in Vulcan® and as well as in the database simultaneously. Switching to the integrated database eliminated many manual processes such as importing and exporting data and enabled access to the database anytime, anywhere and from any device. Thereby, these improvements allowed to increase an efficiency, create a comprehensive reconciliation system and reduce operational risks. We are looking forward to expanding applications of integrated grade control data stored in the database. This paper outlines how the grade control system was developed at OT’s open pit mine, and how it is integrated into the short range mine planning, mine reconciliation workflow and ultimately enabled development of various automated reporting systems.

Mar 22, 2022

By J R. Ruiseco, D La Rosa, L Julian, W Hunt

The DeGrussa copper-gold deposit is located 900 km North of Perth. Copper-gold bearing massive sulfide lenses occur in lower greenschist facies meta-sedimentary and mafic volcanics of the Bryah Basin. Regional folding, faulting, and tilting of the stratigraphy primarily occurred during the Capricorn Orogen (Johnson, 2013). The Shiraz Fault and Merlot Fault are regional oblique-sinistral faults that offset the DeGrussa massive sulfides lenses into three fault blocks by ~350 m and ~120 m, respectively. A network of deposit-scale brittle-ductile faults splay from, and in one example are cut by, the Shiraz and Merlot faults. This results in localised ore displacement, selvage alteration and degradation of the rock mass integrity. Historically, only the Shiraz and Merlot Faults were modelled and validated during mining and grade control. The requirement for a comprehensive structural model became more relevant as mining became advanced and confining stresses on local rock mass became more complicated. The DeGrussa Structural Model has been developed to delineate areas of structural complexity and is an important tool in mitigating rockfall risk through well informed technical design. During this process, every accessible excavation was walked and structurally mapped. New workflows in Leapfrog Geo and new technology, including an app that runs on mobile phones streamlined the validation process. Drill hole orientation was a limiting factor during validation. At the time of writing this submission, the DeGrussa Structural Model consists of 34 faults, and is now an important tool for the Technical Service team when developing mine plans and schedules. Applied structural geology is often seen as exclusive to those with a PhD in the field. Faults at DeGrussa have been modelled and reconciled by a Mine Geologist and Geotechnical Engineer, using first principles geological characterisation. This is a simple process that demonstrates any geologist can contribute to a structural model, provided they understand the fundamentals of unbiased structural measurements and consistent record of observation

Mar 22, 2022

By M Watson, A V. Engelbrecht, E Murray-Hayden

Whereas mine reconciliation is a well-established month-end process for most modern mining operations, the resource sector has yet to endorse uniform reporting factors against which model accuracy and metal recovery can be formally assessed. Logically, any comprehensive reconciliation framework should facilitate comparison between period reserve, mine and mill and production figures. Historically, Parker’s F Series ratios (Parker, 2012) are often used in some form by mine operators, although they are rarely shared externally, as formal definitions have yet to be accepted. This paper presents a structural extension to the reconciliation process, wherein resource ratios (R Series factors) are used to quantify the long-term impact of the information effect, together with dilution and ore loss. A case study from the El Limón Guajes (ELG) Mine Complex in Mexico is used to demonstrate how an R Factor Series can be implemented to facilitate operation optimisation, over and above conventional F Series reconciliation treatments. This approach is applicable to any mining operation.

Mar 22, 2022

By N Pranoto, E Lawell, T Dean, M Grant

The Goonyella Riverside Mine is located 30 km north of Moranbah in Central Queensland, Australia. Recent mining activities have exposed the north–south striking Kenya and Zimbabwe thrust Faults. These faults vertically offset coal seams by up to 25 m and locally thicken coal packages through imbrication of coal seams. The presence of back-thrusts add to the geotechnical complexity of maintaining highwall stability. The Kenya and Zimbabwe Faults have provided multiple challenges to the mining operation, with an unplanned additional 240 kt of coal tonnes being mined from imbrication of coals seams or from the refinement to local scale geological models. These additional coal tonnes have been either identified during the mining of the coal seam or just prior to overburden removal above the coal seam. These faults can also negatively impacted highwall stability with highwall failures, lowering the overall slope angle for pit designs and increasing stand-off distances for hazardous highwall conditions. Unmanned Aerial Vehicles are used to scan exposed highwalls and enable geological mapping via high-quality 3D images. Measure while drill data from drill and blast rigs can provide local-scale intersections of the faults and coal seams. Mine Geologists utilise these and other data sets to create localised geological models refined position of the fault planes and coal seams. These models are utilised by relevant mine site teams to ensure the safe and efficient mining of coal around these faults.

Mar 22, 2022

By R J. Goldie

The Monty Copper Gold Deposit is an ultra-high grade mine in Sandfire Resource’s Doolgunna region which was completely drilled out and developed within two years of commencing mine development. During this period, a total of 986 grade control diamond drill holes were completed, assayed, and validated, 15 levels developed, and 3136 faces mapped. The aggressive schedule meant a high flow of geological data was produced, validated, and incorporated into regular geological model updates. To process this data at a rate which allowed for timely use, more robust data processing methodologies needed to be generated. A new validation system was implemented which required two levels of validation being applied to any diamond drilling update. Step one was the validation of initial physical data including survey pick-ups and logging. This was done prior to new drill data being added to the geological model. Step two was the validation of assay data which was required prior to resource estimation. The implementation of Seequent’s Leapfrog Geo modelling software also allowed for a more seamless and efficient modelling workflow. The use of the Leapfrog Geo vein systems allowed for more rapid generation of geological models which then afforded more time for stringent assessment of the mineralisation system. An example of this was the introduction of cross-shift peer reviews in which opposing rosters of geologists would rigorously evaluate the previous rosters work prior to reporting. This additional level of validation resulted in more robust geological models and the identification of new ore trends. This presentation will detail the evolution of the Monty Geological model from a simple tabular indicated resource into a complex highly refined measured resource. Themes will include how the model was produced using workflows developed and optimised to meet key deliverables.

Mar 22, 2022

By N Anderson

With traditional grade control processes, geological models are created by geologists at the completion of the data collection stage and wireframe construction is closed off at a set date to allow for time to complete the estimation process. Any issues identified at earlier stages or the addition of new data requires the estimation process to be started again. Alternative interpretations are explored only if time allows and often in a limited fashion. An alternative workflow has been developed at the Waihi Underground Gold Mine utilising Leapfrog Edge and Seequent Central. Geological and grade control models are created, dynamically linked, and saved as separate branches. A project geologist is able to work on the grade control model and a mine geologist on the geological model simultaneously. This allows for a non-linear modelling workflow with continual review and feedback. The introduction of new data can occur right up to the model due date, with data collection, validation, alternate interpretation, and review happening incrementally. Models are categorised and stored in Seequent Central according to project status. An approved version of the geological model and grade control estimate are uploaded at the time of release and transparently capture all data used at a point in time, as well as the review process. Dynamically updating models for grade control has reduced turnaround time while increasing accuracy and quality. Estimates of individual domains are vastly improved from being able to rapidly evaluate the effects of model parameter changes in real time and 3D. Development and stope design directly benefit from these streamlined processes, with higher quality, timely production occurring as a result.

Mar 22, 2022

By I Minniakhmetov, J Ashford

Orebodies constitute naturally occurring, commonly highly variable concentrations of minerals in the earth’s crust formed through varying and highly unpredictable processes. Orebody knowledge is a core foundation of a mining company’s asset base underpinning all decisions about Mineral Resource and Ore Reserves, future production, development and ultimately the profitability and competitiveness of a mining enterprise. The decision to invest in a mineral project is usually based upon a techno-economic evaluation of the project and having clear approval pathways to gain local community support and address social and environmental impacts. The main objective of the techno-economic evaluation is to determine the optimum development sequence for the particular orebody in question and to estimate the economic viability of the project and the future return on investment to its stakeholders, balancing the inherent value of the mineral in the ground against the costs of development and market pricing. With the exception of market-driven commodity prices, virtually all revenue and cost estimates fundamentally stem from the understanding and subsequent modelling of the orebody. A reliable estimate of the ore deposit grade and other geological value-drivers is the foundation upon which the business case for future mine development and operation is intricately dependent. Orebody knowledge is, therefore, the fundamental driver of the mineral project cycle, influencing the progression of studies and consequently, the resource company’s ability to declare Exploration Results, Mineral Resources or Ore Reserves which govern the source and cost of funding available to the project. Drilling provides a ‘vector’ to the true nature of the orebody and effectively buys time to manage the technical risk associated with the orebody. Despite its perceived high cost, in both time and money terms, drilling is the only available technology for acquiring the necessary orebody knowledge to support technical and economic studies and to minimise orebody-driven cash flow uncertainty and volatility. However, like all costs, pressure to minimise drilling budgets is ubiquitous across the minerals industry. Paired with inappropriate key performance indicators, drilling is often regarded as a cost to be minimised, rather than an investment to be optimised. Furthermore, as future deposits get deeper and more complex, the collection of orebody knowledge is increasingly on the critical path for project development. The increasing cost of data collection pertaining to these deposits increases the temptation to minimise orebody knowledge gathering at the cost of long-term value as the reduction or postponement of drilling results in greater orebody knowledge risk at all stages of the value chain. Indeed the true cost of the decision to drill less is rarely known until the mine is in operation, well after the mine project evaluation team has moved on. This paper aims to promote orebody knowledge as the fundamental source of endogenous project financial risk and, consequently, its critical importance in project financial evaluation. Furthermore, the report documents experimental research into the development of an econometric model for valuing orebody knowledge. Using the density of drilling data as a proxy measure of orebody knowledge, the proposed approach integrates well-established stochastic grade modelling techniques with the mathematical mechanics of Decoupled Net Present Value (DNPV) which, combined with stochastic grade modelling (conditional simulation), lends itself to the financial International Mining Geology Conference 2022 | Brisbane, Australia and Online | 22–23 March 2022 3 assessment of orebody knowledge risk. Based on this approach, a metric is formulated, representing the minimum value of the orebody knowledge added by additional drilling

Mar 22, 2022

By K L. Silversides, A Chlingaryan, A Melkumyan

Multiple data sources can be fused to improve boundary modelling in stratified deposits such as the banded iron formation (BIF) hosted deposits in the Pilbara Region of Western Australia. Accurate boundaries are essential to produce high quality geological models. Initial models are based on the widely spaced exploration hole data. This creates a high level of uncertainty in areas between the holes, with different interpretations of the boundary equally probable. Including directional information collected in these holes in the models can reduce the number of possibilities. Including data collected during mining, such as blastholes and bench measurements, can further guide the model. This research demonstrates a Gaussian Processes based method of fusing these disparate data sources to better inform the boundaries. Our results demonstrate that when only exploration points are used the model creates a smooth boundary with minimal variation between the holes. Each additional data source refined the model and added additional detail to the boundary. For the blast data, the new detail projects down to the bench below the current bench to allow for a better prediction before mining commences on that bench.

Mar 22, 2022

By D Reid, J McKibben, M Noppe

International mineral disclosure standards and guidelines aligned with the CRIRSCO (Committee for Mineral Reserves International Reporting Standards) template were developed to inform stakeholders about the relative accuracy and confidence levels of publicly reported information and to support investment decisions, particularly when advancing projects through different development stages. The traditional stages range from early to advanced exploration, through pre-development (with Mineral Resources and pre-feasibility and/or feasibility levels of study) before the final development stages, where a decision to proceed with construction or production (or both) is made. The traditional definition of Mineral Resource and Ore/Mineral Reserve classification categories and techno-economic study levels (including Preliminary Economic Assessment (PEA), Scoping Study (SS), Pre-Feasibility Study (PFS) and Feasibility Study (FS)) were defined with pre-development project assessment in mind. As such, these definitions are not necessarily well aligned to the reporting structures required by operating mines which need to consider a range of timelines spanning short to long-term operating conditions. However, the same terminology and definitions are used, or expected to be used, by both project developers and mine operators when reporting resources, reserves and techno-economic study outcomes. This paper discusses how various stakeholders in the minerals industry (including competent/qualified persons, mining company leaders, investors, lenders and regulators), generally interpret the definitions of Mineral Resource and Ore Reserve categories in terms of the relative accuracy of the estimates, their link to techno-economic study levels and their contribution to life-of-mine (LOM) schedules. The paper examines the degree to which mining companies recognise, and apply, enhanced operational knowledge to reflect short-term views of the relative accuracy and confidence levels associated with publicly reported Mineral Resource and Ore Reserve estimates and forecasts. Furthermore, this paper explores how this knowledge is applied to support ongoing Mineral Resource, and particularly Ore Reserve, generation and reporting. This paper is informed by public data and responses to an internal survey prepared by the authors and completed in 2021 by several mining companies.

Mar 22, 2022

By J Taylor, M Smith, A Tunnadine, D Nikolakopoulos

As Olympic Dam rapidly expands into the Southern Mine Area (SMA) the Mine Geology team has concurrently ramped up underground diamond drilling to sustain rates of above 175 km/a. With >10 underground rigs double-shifting, traditional paper sections used for the management of drill programs were no longer optimal due to scale of operations. An opportunity was identified to modernise monitoring of all underground diamond drill programs, through the integration of logged copper sulfide estimates with Olympic Dam’s first-fully integrated deposit-scale geological model in Leapfrog Geo. This paper describes the transition within the Mine Geology team from reliance on 2D paper sections, towards near-real time 3D geological modelling for optimising end-of-hole decisions in the field. This transition was enabled by innovative database solutions informing locally constrained implicit models, limited by a distance buffer around active underground fan drilling. Due to the high volume of historic drilling, there were significant computational challenges with the implementation of daily production style implicit models at Olympic Dam and initial testing of regular updates to implicit gradeshells and sulfide domains required in excess of 12 hrs to run. The enabling solution was to optimise the interfaces between the AcQuire database, Deswik MDM and Leapfrog Geo to create a dynamic, streamlined workflow that was easily completed by Mine Geologists in accordance with daily production duties. Geologists are now building rapid implicit models for a range of variables from an integrated data set of logged sulfide estimates and existing assays, spatially constrained by 150 m buffers around active drilling using a unique drill fan identifier flagged during the planning phase. Once the live-implicit models are updated, they are exported for visualisation on field tablets to be used during logging and core checks at the start and end of every shift. To manage the risk of ‘going digital’ interpretation remains at the forefront of the drill section close-out process, with Mine Geologists critically reviewing the finished sections on a standardised template with new learnings wireframed and communicated to stakeholders. The Live Field Model has enabled a step-change in the geological effectiveness of diamond drilling, reducing over – and under-drilling ore-waste at Olympic Dam continuing to push drilling effectiveness towards 20 per cent of intercepts >2 per cent Cu. In addition, upside or risk identified during the proactive management of drilling execution can be rapidly capitalised or mitigated.

Mar 22, 2022

By J Ashford, Minniakhmetov

Geostatistical simulations are the best practice for uncertainty quantification in non-linear systems. Simulations are currently undertaken on an ad hoc basis in the mining industry due to high requirements in computing environments and advanced geostatistics. In contrast to the estimation process, multivariate simulations can be performed using different methods, tools, and computational environments, similar to data science. For large companies with multiple commodities it is important to have a standard framework to test, build, and run workflows using different methods, applications and computational environments. This work presents an auditable, modular and scalable framework to create geostatistical simulation workflows where the workflow is abstracted from the implementation. Key aspects: automation – workflows should be able to run automatically given updated data (if hands on modelling is not required); auditability – workflows should document key considerations and parameters; modularisation – workflows should permit swapping out components, for example substituting domaining in Leapfrog software with automated domaining using machine learning scripts; scalability – easy, repeatable, portable deployments on diverse infrastructure (for example, experimenting on a laptop, then moving to an on premise cluster or to the Cloud); reproducibility – parameters should be archived to permit re-running and reproduction of workflows in future years; and sensitivity analysis – any component of a workflow should be able to run with multiple scenarios of parameters and data to enable global uncertainty quantification and sensitivity analysis. The presented framework can be used to create different workflows for geostatistical simulations in multiple assets to quantify uncertainty in the value chain for resource categorisation, reserves uncertainty, value of information and robust mine planning amongst others.

Mar 22, 2022

By S Sullivan

A recent review of resource reports lodged to statutory bodies found geological models with oversmoothed or unrealistic lithological boundaries, many of which do not represent geological appearance in the field, pit or underground drive. The underlying data is often too complex or messy to be incorporated into the resource model and is thus a simplification of reality, to the detriment of the mining operation and its shareholders. Volumes can be over- or underestimated and subsequent grade predictions can be misplaced, combining to cause over- or understatement in resource reporting to stakeholders and statutory authorities. Domain models can be improved with new modelling techniques such as machine learning. However, with poor data or improper use of machine learning techniques, the conclusions can be just as misleading as with current techniques. Machine learning rarely offers a single ‘right’ outcome, rather a range of possible outcomes from the data provided. But importantly it can take advantage of the full richness of data for more informed models. Machine learning also provides confidence in its predictions so that uncertainty in the resultant models can be quantified and passed through for risk assessment during mine planning and operation. This paper will discuss the factors accentuating complexity in deposit modelling: data diversity, structural controls, chemistry, data volumes, process workflows and external non-geological constraints. A case study illustrates the risk of reducing or ignoring complexity, which can result in an overly simplified geological model.

Mar 22, 2022

By E J. Hill, F Huang, A Otto

Geochemical data from drill core samples is used in mineral exploration to enable geologists to understand subsurface geology. However, it is challenging to analyse these data sets rigorously and consistently. Manual data interpretation relies on the knowledge and experience of logging geologists. When multiple geologists are involved, it is possible to introduce inconsistencies in logged lithology type and scale of logging (‘splitters’ versus ‘lumpers’). Machine learning (ML) is a powerful tool for analysing high dimensional data (ie data with many variables) and large data sets (ie data with many samples). However, ML methods applied to drill hole samples do not incorporate spatial information and this can result in high misclassification rates and small-scale ‘noisy’ results. We demonstrate the use of multiscale spatial analysis (wavelet tessellation) as a pre-ML step to analyse spatial information. This step helps reduce misclassification and filter out unwanted small-scale variation in the data. In an experiment, we used wavelet tessellation and supervised ML to predict litho-geochemical rock types from geochemistry data. The data set is from the Valhalla uranium deposit near Mount Isa, northern Queensland. First, geochemical knowledge and statistical analysis was used to select appropriate geochemical elements for litho-geochemical classification. Second, the elements were processed using multiscale spatial analysis. Third, we trained a Random Forest model with the data set and found that the model performs worse in predicting rare rock samples than common ones. We applied over-sampling and down-sampling methods in aim to overcome this issue. As a conclusion, the combination of multiscale spatial analysis (to define rock unit boundaries) and Random Forests (to classify rock units) is a good method for rapid and accurate analysis of drill hole geochemistry.

Mar 22, 2022