Webinar: Top 5 Tips for Mineral Resource Estimation

Thank you for joining us at this live stream. I'm Anya, the Business Development Manager at K-MINE. Today, we're going to delve into a highly significant topic for geologists and mining engineers. Our focus will be on the top five tips for mineral resource estimation. With us is Cameron Geologist Tatiana, who will answer your questions at the end of the stream.

If you're interested in studying common mistakes to avoid and getting insights into effective techniques for the mineral resource estimation process, this webinar is for you.

We've been on the market for nearly 30 years, offering a standalone application that can be tailored to fit your company's specific needs. Our solution is built around flexibility thanks to its modular design - a perfect fit for businesses of any size.

Our application is powered by our very own patented graphic core, which enables fast information processing. Our goal is to cater to a wide range of operational needs including 3D modeling, resource estimation, and mine planning.

We currently have 12 modules specifically designed for companies involved in exploration, development, and production stages. These modules are highly configurable, adaptable, and can be enhanced to suit each deposit - whether it's open pit or underground mining.

Each of these 12 modules can function as a standalone application or seamlessly integrate with one another. This eliminates the hassle of creating SharePoints, dealing with compatibility issues, wasting time on import/export processes, or tedious data adjustments. It provides different departments within a company with a unified platform to exchange information.

Our team includes engineers, geologists, planners, surveyors, and other industry professionals. We can provide a wide range of services from resource modeling and mine engineering to auditing. Our engineers not only implement our solutions but also set up your entire company's infrastructure and support you at any stage of your project.

When it comes to geology and mine engineering, professionals often face a range of questions and challenges. Today we aim to address several of them:

• Block size: when creating a block model, what factors should one rely on when making this decision?
• Most prevalent mistakes in the allocation of mineral resource categories
• The complex procedure of delineating ore bodies within polymetallic deposits
• Various approaches for dealing with individual samples that exhibit anomalous content
• The importance of accurately determining the volumetric weight of ores and different interpolation methods

The parent block size stands as a paramount parameter that influences the accuracy of grade estimates within a resource model. When determining the block size for a block model, various factors come into play, including mineralization controls, geometry, mining method, spatial distribution of drill holes and samples, and the anticipated grade control method.

If you choose a block size that is too small, it leads to excessive smoothing of grade estimates, ultimately yielding an inaccurate grade-tonnage ratio. On the other hand, if you select a block size that is too large, it becomes less useful for pit optimization and mine planning. In K-MINE's mine planning model (the Scheduling model), block sizes are set to match the time intervals indicated in the mine plan.

To illustrate the impact: a block model with a 120-meter block size showed an average grade of 3.14% and a maximum block grade of approximately 9%. A model with a 30-meter block size exhibited an average grade of 4.13% and a maximum block grade of about 14%. With a cutoff grade limit of 3%, the difference in tonnage calculation was 20%.

Experts commonly recommend setting the block size at approximately one-third to one-half of the distance between drill holes. Smaller block sizes can introduce artificial smoothing - when neighboring small blocks are evaluated using the same samples, they tend to yield similar grades. If the block size is too large compared to drill hole spacing, it fails to fully utilize the density of the exploration network.

A good rule of thumb: consider a range of at least one-fourth of the average density of the exploration network.

Another crucial factor influencing block size decisions is the estimation of recoverable resources, which encompasses the concept of the Selective Mining Unit (SMU).

The SMU refers to the smallest volume of material used for determining ore or waste classification, derived from geostatistical estimation. It represents the minimum unit that can be selectively mined. However, the smallest mining unit is usually much smaller than the sampling grid dimensions where we can collect information - particularly during exploration and feasibility stages.

When we try to estimate values for these tiny blocks directly, we often end up with low precision. Direct interpolation of small blocks also distorts the grade-tonnage curves - predictions of attribute content above a certain cutoff can end up far from reality.

There are two approaches to solve this issue:

The first approach is to set the contact boundaries separately from the grade estimate. We start by determining the size of the elementary blocks based on the sampling network, then fill the sub-blocks with estimates from the parent blocks.

The second approach is to bring percentage calculations into the picture when assessing volume - using block models that allow for percentage calculations rather than assigning parent block scores to sub-blocks.

In K-MINE, both approaches are available. For a cubic block model, we use a blocking system where users can select a block model with parent blocks and generate a new model based on it. All attributes are automatically loaded, requiring only the specification of a new block size that is a multiple of the original size.

When reducing block size, the grade of the parent block is assigned to the sub-blocks. When increasing block size, attribute values can be recalculated using one of three formulas: weighted average by volume, weighted average by tonnage, or maximum value assignment (assigning the value of the original sub-block that occupies the largest volume within the new larger block).

For a cubic sub-block model, K-MINE utilizes an algorithm to calculate the percentage of ore volumes that fall within each block. Depending on the ratio of ore volumes within a block, other parameters such as content and specific gravity are also weighted.

The process starts with figuring out the percentage of material volume in each block. Once that's done, we estimate grades for each geological unit, then calculate the final average block grade as a weighted average for all geological units found within that block. This approach not only reduces the number of stored blocks but also gives more variables to work with in each block.

Some software also offers mathematical analysis methods to determine the optimal block size. One such method is Kriging Neighbourhood Analysis (KNA, also known as QKNA).

KNA generates statistics that measure conditional bias - the extent of over-smoothing in block estimates compared to the theoretical variogram of grades at that block support. The goal is to identify parameters like block size, number of samples, search radius, and discretization that minimize conditional bias while avoiding excessive negative weights.

Two key statistics are used for optimization:

Kriging Efficiency tells us how well the kriging estimate accurately reflects the local block grade. It's computed by comparing the kriging variance of the block with the theoretical variance of the block. Lower kriging efficiency indicates a high degree of over-smoothing; higher kriging efficiency indicates a low degree of over-smoothing.

Slope of Regression (Conditional Bias Slope) summarizes the degree of over-smoothing for high and low grades. It represents the regression slope of the estimated block grades against the corresponding true (yet unknown) grades. Ideally, the optimal result is a slope of 1 and kriging efficiency of 100%. In practice, common values typically fall within 80-90%.

By leveraging KNA, software users can analyze and justify the optimal block size by considering statistical measures of conditional bias and ensuring a balance between over-smoothing and accuracy in grade estimation.

When evaluating mineral resources, a solid understanding of ore-controlling structures and precise definition of mineralization zones plays a vital role in ensuring accurate estimation and effective resource management.

Despite the benefits of computer technology in mineral resource modeling, there is a drawback in the widespread use of software products. Users often rely solely on mathematical calculations embedded in the software, neglecting the analysis of algorithms and actual utilization of all available geological information. This can lead to inadequately classified mineral resource estimates - commonly known as the "Spotted Dog" effect.

During resource estimation, we typically use statistical analysis to determine the radii of the anisotropy ellipsoid, then apply one of the kriging methods to estimate blocks. In regions with many samples or closely spaced drill holes, the program determines a high level of reliability for the Measured category. In remote areas where samples are limited, we switch to the Indicated category. As a result, isolated spots or ring structures of the Measured category appear, with spaces between them filled by the Indicated category.

This approach sometimes overlooks crucial factors such as lithology or fault systems that control mineralization. While values like the number of samples, kriging variance, regression slope, and proximity to nearest drill hole are helpful in assessing the confidence of content estimation, they should be considered only as an initial step in the mineral resource classification process.

It's vital to take broader aspects into account: confidence in geological interpretation, data quantity, distribution, and quality.

Practical solutions exist to address these challenges, such as the development of resource classification frameworks that can be modified based on kriging variance. Configuration smoothers through computer functions help achieve smoother boundaries for resource categories.

The Competent Person should also consider additional criteria: reliability of drilling data, structural interpretation as a database, feasibility of economic extraction, metallurgical test work, technical data, geotechnical considerations, social accessibility, legal and land tenure factors.

To ensure compliance with reporting standards such as JORC Code, CIM Standards, NI 43-101, and SK-1300, it's essential to consider the requirements for continuity. Continuity refers to the correlation between drill holes - not contouring around individual drill holes.

In JORC Code, the definition of a Measured Mineral Resource states that geological and grade continuity between points of observation must be confirmed through outcrops, trenches, pits, workings, and drill holes.

The definition of an Indicated Mineral Resource emphasizes that there must be significant evidence to assume geological and grade continuity between points of observation.

Practitioners can deliver precise and dependable evaluations by making sure that geological and grade continuity is convincingly demonstrated throughout the range of data collection methods.

It's crucial to understand the distinction between Indicated and Inferred resources, since only Measured and Indicated resources can be converted to Ore Reserves.

Indicated mineral resources should exhibit sufficient confidence to allow the application of Modifying Factors supporting mine planning and economic evaluation. Inferred mineral resources lack the necessary confidence for detailed planning and should be approached with caution in technical and economic studies.

There are two main approaches employed in mine design and reserve estimation:

In the initial stages (scoping studies), we often apply inflated coefficients for dilution and extraction factors due to limited information available. We often resort to using analogy methods, drawing insights from similar deposits and production methods.

In later stages (pre-feasibility to final feasibility study), mine designs are crafted for each individual stope, taking into account dilution, mine recovery, recovery factors, backfill methods, geological structures, equipment capabilities, and corporate requirements. Stopes that meet economic criteria - along with metallurgical, marketing, legal, environmental, social, and government factors - have potential to be classified as Ore Reserves.

When engineers use block models for resource estimation, they encounter a challenging situation with stopes containing a combination of mineral resource categories, unclassified material, and waste. There are two available solutions:

The first approach is to calculate weighted average resource classification and assign the corresponding reserve category. However, this may lead to excluding Indicated resource blocks from being converted into reserves, resulting in a significant number of resource blocks not being converted.

The second approach is to determine the tonnes and content of the resource categories in the stope and convert individual resources to the appropriate reserve categories. This ensures accurate conversion, though reporting Inferred resources and unclassified material present in the mine (which may be treated as internal dilution or waste) remains a challenge.

Determining what values are considered outliers is a subjective process. The Competent Person typically decides whether to exclude outliers from the sample or replace them with a different value. Outliers are extremely high values - many grade distributions exhibit positive skewness, and a small number of very high values can significantly impact summary statistics such as the mean, variance, correlation coefficient, and measures of spatial continuity.

When dealing with data, first remove any erroneous values from the dataset. For valid extreme values, we can either classify them as a separate statistical population and give them special treatment, or employ robust statistics that are less affected by extreme values.

The golden rule: avoid modifying data unless you know for sure that there are errors present. However, even valid extreme values may have limited influence on the spatial predictive model.

Outliers need to be identified and decisions made before compositing, working with the source database. Several methods can be employed for identification:

Visualization methods include box plots (outliers appear as individual points outside the expected range) and scatter plots (outliers appear as points that deviate significantly from the main data cluster).

Statistical measures include the Z-score method (measuring the number of standard deviations by which an observation deviates from the mean, with a typical threshold of plus/minus three) and the IQR method (Interquartile Range - the difference between the 25th and 75th percentiles). The lower limit is Q1 minus 1.5 times the IQR; the upper limit is Q3 plus 1.5 times the IQR. Any data point outside this range is considered an outlier.

The 68-95-99.7 rule (empirical rule) provides a way to understand data spread within a normal distribution: approximately 68% of data falls within one standard deviation, 95% within two, and 99.7% within three standard deviations.

Estimated tonnage of the deposit relies heavily on the tonnage factor or density applied to model volumes. In a study conducted by Abzalov examining 50 technical reports filed on SEDAR, the findings showed:

• Only 20% of reports utilized a density-specific dataset, independently estimating density for each block in the block model
• 18% of reports didn't discuss density at all
• Around 58% of reports simply reported a single average density value without considering data distribution

The more appropriate term now is dry bulk density (rather than specific gravity) - the dry mass of a rock per unit of actual in-situ rock volume while accounting for porosity.

Sample density is often correlated with sample grade due to the higher specific gravity of most metals compared to host rock. Deposits such as uranium, massive sulfides, iron, and some high-grade deposits typically exhibit a clear correlation between grade and density.

For deposits with low metal content and simple mineralogy, calculating the average of all bulk density measurements may be sufficient. However, for deposits with complex mineralogy and a relationship between density and grades, simply averaging bulk density within each geological domain can lead to errors. A better approach is to apply similar interpolation parameters used for grade estimates.

The number of dry bulk density samples typically ranges from less than 200 to several thousand depending on deposit size and type. Geostatistically optimized grids for measuring rock densities are usually positioned between the chemical assay grids needed for estimating Proved and Probable Reserves.

When dealing with polymetallic deposits, it can be helpful to present results in terms of the metal that contributes the most value. The primary mineral chosen for reporting on a metal equivalent basis is the one that holds the highest value in the deposit. Secondary minerals are multiplied by their respective current mineral prices and divided by the current price of the primary mineral.

The calculation of equivalent grade relies on two factors: metal price of each commodity and the process recovery for each commodity.

For JORC Code compliance (Clause 50), when reporting metal equivalents, the following details must be included:

• Individual grades for all metals involved in the metal equivalent calculation
• Assumed commodity prices for all metals (simply referring to "spot price" without disclosing the actual price is not sufficient)
• Assumed metallurgical recoveries for all metals, with a discussion of how these recoveries were derived
• A clear statement that all elements have a reasonable potential to be recovered and sold
• The calculation formula for metal equivalents

If there is no reliable estimation of metallurgical recovery information available, reporting metal equivalents would not be suitable. For many projects at the exploration results stage, metallurgical recovery information may not be available or able to be estimated with reasonable confidence.

Q: How can a resource category be delineated in the early stages of exploration when there are large distances between drill holes?

A: In the early stage of exploration, the generated data are insufficient for estimation of mineral resources and ore reserves. However, it might be of use to investors, therefore it's reported as exploration results - including results of outcrop sampling, geochemical exploration data, and geophysical survey results. If a company reports exploration results, then estimates of tonnage and average grade usually are not assigned to the mineralization. If the company chooses to discuss the exploration results in terms of exploration target size, it should be quoted as a range of tonnes and range of grade.

Q: Can mineral resources and reserves be evaluated outside the final pit contour?

A: Resources that fall outside of the pit shells are not to be converted to reserves because they are not economically valuable given the economic and technical parameters used for constructing pit models. However, these can still be reported as mineral resources if they pass the requirement for reasonable prospects for eventual economic extraction.

The same applies to Inferred resources which cannot be converted to Ore Reserves because of low confidence in their tonnage and grade. The JORC Code does not stop us from using Inferred resources in a mine plan, but it does guide how they should be used for estimating and reporting Ore Reserves.

If there are no reasonable prospects for eventual economic extraction, then regardless of the geological and sampling data available, a concentration of solid material of economic interest cannot be defined as a mineral resource. Examples include: a deposit at a depth too great to be economically mined for the deposit grade, a deposit within a national park or heritage site, and a deposit in a country that has political restrictions or war.

Q: What is more important when evaluating metal equivalent - price or recovery?

A: Both are equally important. The calculation of metal equivalent depends on assay data, mineral pricing, and metallurgical recoveries. Clause 50 from the JORC Code specifies reporting requirements including individual grades, assumed commodity prices, assumed metallurgical recoveries (and how they were derived), a clear statement that all elements have reasonable potential to be sold, and the calculation formula.

Some companies use every element possible to include in their equivalent calculations, sometimes with unrealistic metal ratios, high prices for by-product metals, and low prices for the metals they are calculating equivalent values for. The most common issue is providing recovery percentages for some but not all of the metals, using 100% recovery for metals without a stated value. Reporting on the basis of metal equivalents is not appropriate if metallurgical recovery information is not available or cannot be estimated with reasonable confidence.

Comment: The IQR method sometimes cuts too much data.

A: Yes, it depends on the type of mineralization, the size of the deposit, and the data distribution.

Thank you for joining us. If you're interested in trying K-MINE, we are currently offering free proof of concept. If you want to see how your workflow can be done in K-MINE and how we can optimize and streamline your processes within one interface that includes all modules, feel free to reach out.