: A method to reduce the influence of known but uncontrollable variables (like ore hardness variations over time) on trial results. Response Surface Methodology (RSM)
Statistical Methods for Mineral Engineers is both a critical field of study and the title of the industry-standard textbook by . This review covers the essential methods used in the industry and a breakdown of the primary resource available to professionals. Core Statistical Methods in Mineral Engineering
Accurate data collection is the foundation of any statistical analysis. In mineral processing, Pierre Gy’s Sampling Theory serves as the gold standard for understanding and minimizing sampling errors. The Total Sampling Error (TSE)
Mineral engineers rely on statistics to turn complex data into actionable insights. Key reasons include: Statistical Methods For Mineral Engineers
Sampling is arguably the most critical yet frequently misapplied statistical discipline in mineral engineering. Incorrect sampling introduces structural biases that cannot be corrected by downstream mathematical smoothing. Pierre Gy’s Sampling Theory (TOS) provides the industry standard for minimizing sampling errors. The Total Sampling Error (TSE)
A well-designed QA/QC programme is the first line of defence against unreliable estimates. Such programmes include the systematic insertion of certified reference materials (standards), blanks, and duplicate samples into the analytical stream. Statistical techniques then evaluate whether assays are accurate (free from bias), precise (reproducible), and free from cross-contamination. Analysing coarse duplicate data can help practitioners predict the true coefficient of variation of a dataset – that is, the real variability of the mineralisation after accounting for sampling and analytical error. Modern practice calls for adjusting QA/QC programmes over time as data quality requirements change throughout the project life cycle.
| Pitfall | Consequence | Statistical Remedy | | :--- | :--- | :--- | | | Overestimates plant feed grade | Report P50, P90, and mean. Use geometric mean for lognormal data. | | Ignoring nugget effect in variograms | Underestimates short-scale variability | Perform rigorous variography with lag spacing < 10m. | | Applying t-tests to autocorrelated data | Massive type I error (false positives) | Use time-series control charts or pre-whiten data. | | Overfitting with stepwise regression | Model fails on new data | Use cross-validation or regularization (LASSO, ridge). | | Pseudoreplication in flotation tests | Inflated degrees of freedom | A single cell with 5 assays is not 5 replicates. Average first, then test across true replicates. | : A method to reduce the influence of
Reviewers from SMI-JKMRC and Informit describe it as an essential text that every plant metallurgist should have on their shelf. Learning and Training Opportunities
To help apply these concepts to your specific operation, could you share a bit more context?
Occurs when particles are selectively introduced or excluded from the sample cutter based on their size or density. 2. Materialization Errors (ME) Key reasons include: Sampling is arguably the most
= The sampling constant (incorporating mineral density, liberation characteristics, and shape factors).
These metrics quantify how well a circuit meets operational specifications. A Cpkcap C sub p k end-sub
Kriging is the general name for a family of generalised least-squares regression algorithms that use the variogram to assign weights to neighbouring samples for grade estimation at unsampled locations. Ordinary kriging (the most widely applied form) assumes a constant but unknown mean within each estimation domain. The kriging variance – a measure of estimation uncertainty – is produced alongside each grade estimate, providing a quantitative indicator of confidence.
Before fitting a regression model (e.g., recovery = a·grade + b·grind + error), run a Durbin-Watson test. If the statistic is near 0 or 4 (strong autocorrelation), switch to time-series models like ARIMA or use differencing.
Measurements from highly accurate instruments (like calibrated weightometers) receive low variance values (
