Enhancing mine risk assessment through more accurate reproduction of correlations and interactions between uncertain variables.

Risk is a significant phenomenon in mineral industries due to several associated social, environmental, technical, and financial uncertainties. Risk assessment is a standard procedure that evaluates the effects of uncertainties on a mining project. To deal with technical and financial uncertainties, the most well-known risk assessment technique is the Monte Carlo simulation (MCS), which requires reproducing correlations between uncertain variables. Correlation does not imply causation, but it does provide information regarding how uncertain variables interact. Given that samples generated in MCS are used in a transfer function (e.g., to produce net present value), transfer function values may mislead risk assessors if the interactions are not reproduced. This study uses historical reference data to compare MCS outcomes based on Pearson and copula correlations with regard to their ability to reproduce interactions. Furthermore, results from a case study on a gold mining project—including gold price, production cost, grade, and recovery as well as interest rate as uncertain parameters—show that if the associations between the variables are non-linear, copulas capture interactions and correlations more accurately than Pearson. [ABSTRACT FROM AUTHOR]