Multivariate regression (MVR) mining project cost estimator for future mineral supply scenario modeling

Abstract

Growing demand for various communication, transportation and energy technologies will require a wide range of minerals at quantities that are significantly greater than historical levels. To model future mineral supplies to meet this demand, we developed a new, globally representative, multivariate regression-based (MVR) mining project cost model that can be used to estimate capital expenditures (CAPEX) and operational expenses (OPEX) for 28 mineral commodities that cover the agricultural, energy, infrastructure, and manufacturing industries. These equations utilize common mining project parameters and do not require detailed knowledge of a specific project to estimate costs.

An uneven distribution in reported CAPEX and OPEX data led to the generation of two sets of MVR equations: commodity-specific MVR equations for commodities with a large sample size (e.g., Au, Cu, Zn), and data-aggregated, non-commodity-specific equations for commodities with minimal reported data (e.g., Nb, Ta, and V). Commodity-specific CAPEX and/or OPEX equations were developed for the following 19 minerals: Ag, Au, Co, Cu, Fe, graphite, Li, lanthanides, Mo, Ni, Nb, Pb, phosphate, potash, Pt, Sn, U, W, and Zn, with most equations yielding R² > 0.7. Using the data aggregated approach, non-commodity-specific regressions were developed to estimate bauxite, Nb, Ta, ilmenite, rutile, Mn, Pd, Ti, V, and Zr project costs. Furthermore, we demonstrate the diverse utility of our equations through case studies coupled with Net Present Value (NPV) estimates, highlighting (1) the impact of price volatility on fertilizer projects, (2) the dependency of co-production in platinum group metal projects for increasing the added value and viability of the project, and (3) the capability of various copper projects in meeting growing global demand.