Fuel optimization in mining trucks using machine learning

Terpstra V.J.; Lara-Yejas O.; Mokhtari K.; Cruz J.F.H.S.; de Mattos M.P. · IISE Annual Conference and Expo 2021 (2021)

Resumen

Haul trucks are one of the most important pieces of equipment in the mining operations, having a major impact in overall operational costs. In fact, fuel consumption for an entire fleet of trucks could escalate to millions of dollars annually. This paper presents a combined data exploratory analysis with machine learning models to evaluate trucks' fuel efficiencies and estimate potential savings. The prediction of fuel consumption is based on tracked truck dispatch cycles and re-fueling events. In addition, comprehensive statistical analysis has been carried out to identify the effective parameters influencing fuel consumption, considering the inter-correlation of the independent variables. Different models including XGBoost, SVM, and Neural Networks were evaluated. Our analysis indicates four key areas for fuel savings: truck maintenance, operator driving style, truck stops, and operational decision optimization, totaling up to 10% ($12M) cost reduction. Therefore, prediction of fuel consumption for mining trucks proves to be a viable tool for significant operational costs reduction.