Dilution can be estimated prior to mining by using advanced methods that incorporate geological, geotechnical, and mine design data to predict the amount of waste material likely to be extracted along with the ore.
One practical approach uses machine learning models, such as back-propagation neural networks, decision trees, and random forests. These models incorporate multiple input variables including rock stability, drilling accuracy, powder factor, and geometry to predict unplanned dilution with high accuracy. For example, a neural network model trained on 120 datasets achieved a strong correlation (R² = 0.9761) between predicted and actual dilution values, outperforming traditional empirical methods(Rodrigues et al., 2025).
Another method involves creating mathematical prediction models based on geological, geotechnical, and mining design attributes for individual stopes. These models can be integrated early in mine planning to provide granular dilution forecasts per stope, aiding in production schedule optimization. They deliberately omit subjective human performance variables, ensuring the model generalizes well for strategic planning and provides earlier, cost-saving insights(Chimunhu et al., 2025).
More so, dilution can be estimated using numerical simulations, empirical graphs, and spatial analysis that consider unit size, ore-waste contacts, and equipment precision. Advanced methods also adjust block grades by analyzing ore-waste relationships and potential waste inclusion. These predictive methods enable engineers to manage dilution risk, optimize mine design, and enhance metal recovery before mining begins(Chimunhu et al., 2025).
Reference:
-
Chimunhu, P., Faradonbeh, R. S., Topal, E., Asad, M. W. A., & Ajak, A. D. (2025). Dilution prediction in underground open stope mining using gene expression programming and backpropagation artificial neural network algorithms. Mining Technology, 134(2), 105–120. https://doi.org/10.1177/25726668251348707
-
Rodrigues, C. O., Matos, J. M. V., Santos, T. B. D., & Santos, A. E. M. (2025). A new approach to dilution prediction of underground mine gold using computing techniques. Anais Da Academia Brasileira de Ciências, 97, e20240426. https://doi.org/10.1590/0001-376520252024042
