Uncertainty assessment in mineral resource estimation using geostatistics and Monte Carlo simulation

Purpose. The accurate estimation of mineral reserves is a critical step in the planning and development of mining operations, particularly for phosphate deposits. This study aims to enhance reserve estimation for the Bled El-Hadba phosphate deposit (eastern Algeria) by combining geostatistical and probabilistic approaches. Methodology. The proposed method integrates ordinary kriging (OK) with Monte Carlo simulation (MCS) to better evaluate the reserves of the main mineralized layer. Ordinary kriging was applied on a 5  5 m grid to generate estimation maps of P2O5 grades and layer thicknesses, serving as a spatial model of the deposit. The kriging outputs (mean and variance) were then used as inputs for a Monte Carlo simulation involving 20 million iterations, preserving the statistical characteristics of the original data. Findings. The combined use of OK and MCS reduced the fluctuations typically observed in kriging-based estimates and led to more stable and robust reserve values. The application of P90 and P50 probability categories contributed to more conservative and reliable classifications of mineral reserves, improving the assessment of proven and probable categories essential for economic decision-making. Originality. This study introduces a novel integration of kriging with large-scale Monte Carlo simulation to manage uncertainty in phosphate reserve estimation. The approach enables a more precise characterization of spatial variability and supports probabilistic interpretations of resource quantities. Practical value. The methodology offers a valuable tool for mine planners and decision-makers by improving confidence in reserve estimates and supporting sound economic evaluations. Its application to the Bled El-Hadba deposit demonstrates its effectiveness and potential for broader use in phosphate mining projects.