Short-Term Schedule Optimization with Nonlinear Blending Models for Improved Metallurgical Recovery in Mining

In mining and mineral processing operations, ore blending is an inherent practice, as individual ore blocks are seldom processed in isolation but rather in combination with other ore blocks. The common assumption that estimated values within these blocks will be realized implicitly presupposes a linear relationship governing the blending process. While this assumption holds for linear variables such as grades, it falls short of accurately representing metallurgical recovery, which does not exhibit linear averaging behavior. This paper introduces nonlinear blending models capturing the intricate behavior of metallurgical recovery. Specifically, we present two selected nonlinear metallurgical recovery models and evaluate their impact on a mine schedule compared to a linear model. Also, nonlinearity allows optimization. We develop and implement a simulated annealing algorithm for short-term scheduling to maximize total metal recovery. The results of our study demonstrate that the optimized schedule, guided by the nonlinear blending model, exhibits improvements in addressing synergistic and antagonistic blending behaviors when contrasted with conventional linear-based scheduling plans. These findings underscore the significance of adopting nonlinear models in mining and mineral processing operations for enhanced accuracy in metallurgical recovery predictions and overall operational efficiency.