Metallurgical variable re-expression for geostatistics

Abstract

One of the key challenges in the geostatistical modeling of metallurgical variables is their nonlinearity; estimation techniques such as kriging require that the variable of interest average linearly. By quantifying and modeling the metallurgical properties of blends, these metallurgical variables can be spatially modeled and scaled. Atheoretical framework for the re-expression of these nonlinear metallurgical variables to linear variables using experimental data on the nature of blending is developed and demonstrated. The framework is developed using the power transformation family. Potential applications for this framework include mine plan optimization using high resolution spatial models and optimal decision making for processing multiple ore types.