Webinar: Mineral Resource Validation with K-MINE

K-MINE offers 12 modules for open-pit and underground mining, each built on real-world use cases. These modules can operate as standalone solutions or work together, allowing each company to assemble a workflow that fits its specific needs - from geological modeling and resource estimation through to mine design and production scheduling.

K-MINE engineers support clients at every stage: building 3D models, planning schedules, estimating resources and reserves. The platform also supports custom add-ons for integration with dispatch systems, fleet management, drilling machines, and other hardware that captures real-time operational data. ERP integration allows combining management and design tools within a single environment.

Herbert Hoover was one of the first to propose dividing mineral resources by the degree of geological reliability, back in 1909. His approach became the basis for modern classification systems. According to Hoover, resources were divided into proved, probable, and possible categories.

Independently, in 1910, a mining congress proposed designating resource categories using letters A, B, and C. Six categories were eventually defined, each with established maximum permissible errors in the quantity of mineral resources - for example, A1 at plus/minus 10%, A2 at plus/minus 20%, B1 at plus/minus 30%, and B2 at plus/minus 50%.

Meanwhile, in the United States, Australia, and European countries, the Hoover approach continued to evolve. Resource and reserve categories communicate reliability of estimation and associated risks.

A turning point came when the industry worked to unify the concept of resource and reserve categories across different national classification systems and develop bridging documents. Since 1994, the CRIRSCO organization has worked to create a set of international standards for reporting mineral resources and reserves, building on the evolving JORC Code definitions.

Notably, JORC does not specify numerical confidence levels but gives clear definitions of resource and reserve categories. Some other professional organizations rely on project stages and offer quantitative expressions of estimation accuracy at each stage.

Polygonal methods for resource estimation are still widely used in many countries for compiling mineral resource and reserve reports. However, attempts to distribute resource categories by reliability using these methods proved problematic.

The reliability of initial data - sampling and chemical analysis - does not change across different resource categories. Therefore, when estimating resources, it is nearly impossible to assign a predetermined error value for individual categories.

As a result, the industry adopted the practice of assigning resources to categories depending on the density of the exploration grid. For each group of deposits, the distances between exploration intersections were established for various resource categories, based on the complexity of the geological structure and the morphological types of ore bodies.

Geologists generally use three methods for quantitative evaluation and validation of resource estimates.

In this approach, resources defined by the densest exploration grid are taken as a baseline. The grid is then progressively thinned, and at each density level, the error in measured resources is determined by comparing them with the estimates from the denser grid.

The grid thinning method simulates the impact of estimation errors caused specifically by grid density, without accounting for other factors such as reservoir morphology, testing errors, methods of interpreting exploration data, or the resource estimation methodology itself. Similar to the measured resource validation principle - if you drill more holes and get the same result, this confirms geological and grade continuity.

Resources are estimated using several different methods - for example, the method of horizontal sections and vertical sections - and then the results are compared.

The reliability of estimated resources depends on the number of ore intersections involved in the estimation. With polygonal methods (without creating block models), the results of resource estimation using geological blocks or vertical sections are directly dependent on the determination of average thickness values and quality indicators for ore intersections.

Errors in the estimation of average block grades relate to individual counting blocks and depend on the variability of mineral quality, the number of exploration sections, the density and layout of the exploration grid, the geometry of the samples, and the geometry of the counting blocks.

Geometrization errors depend on the morphological features of the deposit, the density of the exploration grid, and the thickness and structural discontinuity of the deposit. These errors are not functionally related to errors in average block grade estimation.

By creating a digital model of the deposit using GIS technologies - even without a formal block model, using vectorized geological plans and sections - geometrization errors can be minimized through visualization and spatial referencing.

The third validation method compares resource estimation results with actual operational (production) data. The downside is that this is only possible at the operational stage. However, it allows quantifying the calculation error in specific areas of the deposit and can create a benchmark for evaluation.

The approach involves estimating resources based solely on exploration data, then comparing those estimates against a model based on operational data.

When estimating resources without applying mathematical modeling, it is difficult to quantify the reliability of geological interpretation. Programs for statistical processing of exploration data often work only with tabular data, without coordinate reference or visualization of results.

K-MINE integrates all deposit data into a single spatial environment for detailed analysis, improving the accuracy of geological interpretation - especially in the areas between exploration intersections. Using geostatistical analysis in K-MINE, you can calculate quantitative and qualitative indicators using various mathematical algorithms.

Grade calculation in block models can be performed using any interpolation method: nearest sample, inverse distance weighting (IDW), or kriging. Comparative analysis of results between methods - or against traditional counting results - is straightforward within the platform.

The accuracy and reliability of resource estimation depend primarily on geological knowledge and exploration data, and much less on the counting method itself. However, geological interpretations are always less accurate than mathematical calculations. Since geostatistical analysis studies distribution patterns based on autocorrelation, it helps both to estimate resources correctly and to determine the reliability of the assessment in quantitative terms.

Calculation of resource classes or categories is performed simultaneously with grade interpolation. The classification depends on the dimensional parameters of the search ellipsoid, which are typically determined during semi-variogram analysis, as well as on the number of samples included in the ellipsoid or its sectors.

Resource estimation validation is based on search ellipsoid parameters: as the search radius decreases, the reliability of the calculation and the resource category increases. On the basis of the block model, in addition to general metal estimation, it is possible to predict the amount of reserves that will actually be extracted.

Using K-MINE's Pit Optimizer module, you can create a final pit and perform calculations under various scenarios and cutoff grade options to select the optimal model.

K-MINE supports a comprehensive approach to mineral resource estimation through multivariate analysis. A block model can simultaneously contain information on technological types by processing method, technological grades, enrichment indicators, and industrial grades of ore outlined according to specified conditions.

The best modeling results are achieved when model parameters - radius of influence, anisotropy type, kriging equation coefficients, or other smoothing procedures - are selected by comparing model calculations with operational exploration results from worked areas of deposits.

However, there are situations when geostatistical methods cannot be applied. Resource estimation is a model that relies on large amounts of data and on human judgment and interpretation. The data used to produce a resource estimate is subject to different levels of quality control. In terms of errors directly associated with the production of a 3D block model, it is not always possible to quantify model uncertainty - particularly for models produced using traditional estimation methods.

In this example, a deposit with rare earth zirconium ores is examined. The ore zones are hosted in altered syenites, granites, and greisens. Due to the metasomatic origin of the ore, there is no clear lithological control, so ore body delineation was performed using drill hole and trench sampling results.

Verification of the resource model begins with the database - comparing it with original source materials. The same applies to digitized geological maps and sections: when working with historical paper data, it is essential to ensure no critical errors were made during database creation.

When creating a block model, solids of ore bodies are used, with contouring based on plans and sections compiled using additional data such as natural gamma logging, X-ray diffraction analysis, and spectral analysis.

For validation of global and local estimates, the final resource block model must be consistent with:

• Primary geological and mineralization data
• Wireframe models and structural models
• Topography, excavation surfaces, and volumes
• Analytical data used to prepare estimates of modeled attributes

Key validation methods include:

Visual inspection - comparing block model grade distribution against original drill hole and sample grades in section and plan views. Volume estimates can also be compared between the block model and 3D wireframe models.

Alternative estimation comparison - comparing primary estimated block grades using different estimation methods. Checking for global bias by estimation pass, by domain, by resource category, or by comparing interpolated grades against nearest-neighbor or clustered composite statistics.

Swath plot analysis - comparing block grade distribution against sample grade distribution along coordinate lines and elevation. This method evaluates block grade errors caused by the selected interpolation method in a specified direction.

In the K-MINE demonstration, block grades calculated by the nearest-neighbor and inverse distance methods are compared with composite samples in the north-south direction. The boundaries of zone coordinates can be limited for drift analysis, with the step or number of intervals specified for the chart.

If significant discrepancies appear in the graph, attention should be paid to that sector to understand the cause. This method is convenient when using a block model and allows express analysis of any area or the deposit as a whole, both in extension and depth.

Comparison with previous estimates - comparing the current block model with previous mineral resource estimates, for example comparing a block model from the beginning of the year with the end of the year to analyze losses and dilution.

Global summary statistics - comparing composite grades to block grades.

In mining projects, sensitivity analysis applied to the grade-tonnage relationship is a recognized stage in all existing quality standards. In a technical report, sensitivity analysis refers to the curves or functions that relate average grade and tonnage at different cutoff grades.

The grade-tonnage curve approximation to reality depends on natural parameters such as geology and geochemistry, as well as the geometric distribution of the deposit. Generally, the more variable the grades and the more complex the geometry, the less reliable the curve becomes.

Selecting the optimum cutoff grade by reference to the grade-tonnage curve is critical, as it will guide the mining project throughout its operation. Where relevant, mineral resource estimation may include material below the selected cutoff grade to ensure the resource comprises bodies of mineralization of adequate size and continuity.

The grade-tonnage curve calculated from block kriging estimates sometimes differs from the actual one. Kriging should not be used to estimate small block grades from sparse data.

K-MINE supports cross-validation with two diagnostic diagrams: value estimation and difference value. Users can analyze statistical parameters of the kriging results, including number of samples, estimation and kriging variance, regression, and standard error.

K-MINE provides reports on volume, tonnage, and average grades calculated by different methods - weighted average by volume, weighted average by weight, or arithmetic mean. Users can group results by any attribute (resource category, rock type) and export to PDF, Excel, CSV, and XML formats.

Block model filtering by any property allows calculating volume, tonnage, and average grade for specific zones. Operational block models can be compared with previous mineral resource estimates.

The system implements several methods of volume calculation, including the nearest-area method and polygonal methods (Thiessen polygons). Simple grade-tonnage calculations using polygonal outlines allow constructing influence zones and calculating area, volume, and tonnage reports.

When creating a resource model, the full set of exploration data must account for reasonable prospects for eventual economic extraction. This assessment includes approximate mining parameters - whether the deposit is likely to be mined by open-pit or underground methods, and the likely treatment process.

Mineral resource estimation is not an inventory of all material drilled and sampled. It is a realistic inventory of mineralization under assumed and justifiable technical, economic, and development conditions. Material that is marginally economic today may become extractable in the future.

Question: What criteria can be used to determine that the block model has been validated accurately after reconciliation?

There are multiple techniques for reconciliation - resource model with grade control, reserves with grade control, and production reconciliation. Some experts suggest that a normal range of ore accuracy is about 10%. A significant advantage of reconciling grade control and mineral resource models is the ability to produce a total mass balance comparison, including ore, waste, and low-grade material. Ultimately, it is not the percentage variation between models that matters most, but the analysis of the reasons behind the differences.

Question: Should a mine engineer simply re-block the resource model for planning purposes?

The main difference between ore reserve predictions and subsequent grade control reconciliation stems from an incorrect understanding of how geostatistical concepts of block size, panel size, and selective mining unit (SMU) relate to actual mining practice.

In mining applications, we often need to map the distribution of mineral attributes on block support rather than sample support. The selective mining unit refers to the minimum support upon which ore-waste allocation decisions are made, particularly in open-pit situations. The SMU is usually smaller than the sampling grid dimensions, so direct linear estimation of such small blocks has very little precision.

Three methods address this correctly: reporting at the parent block level using a sub-cell model for proportions; reporting at the elementary level using estimates created from parent blocks; non-linear geostatistics (multiple indicator kriging, disjunctive kriging, residual indicator kriging); and conditional simulation.

Question: How often should reconciliation between the block model and actual data be performed?

Reconciliation is recommended at least once a month based on operational results, depending on productivity and production targets. An indicator for unscheduled verification can be a material change between the predicted and actual values from the block model. Reconciliation can be carried out on a monthly, quarterly, half-yearly, or yearly basis, with information accumulated to refine both the resource and operational block models.

Question: Should the resource block model be updated after reconciliation?

Ore reserves are the economically minable part of measured and indicated mineral resources, including diluting materials and allowances for losses. Confidence in inferred resources is usually not sufficient for detailed planning, so there is no direct link from inferred resources to ore reserves.

However, inferred resources sometimes make sense to include in a mine plan and schedule. For example, the JORC Code does not prevent using inferred mineral resources in mine planning, but it does guide how they should be used for estimating and reporting ore reserves.

K-MINE provides a single digital workspace where geologists, mine engineers, and surveyors work together, exchanging information and receiving real-time data from the site. The platform allows specialists to update their calculations when conditions change on the mine, while managers and directors receive up-to-date summaries at any moment to support faster decision-making.

K-MINE is used by companies of different sizes for open-pit and underground mining, with over 5,000 surveyors, geologists, mine engineers, designers, and technicians leveraging the platform daily.