Abstract

In open-pit mining, off-the-road truck haulage plays a crucial role in facilitating the movement of bulk materials (e.g., ore, waste, and overburden) from mining faces to designated locations within mine sites. The productivity of truck haulage (or truck productivity), also known as tonne per net operating hour (TPNOH), is a measure of the amount of ores (unit: tonnes) that can be moved by a mining truck in a given period of time (unit: hours, consisting of wait and idle time, haul time, and empty time), which is a critical determinant of the overall productivity and cost-effectiveness of mining operations. To assess truck productivity, researchers have initiated data-driven modeling to build the relationship between truck productivity and vital influencing parameters. Real-site weather conditions are often involved as input variables in building prediction models to predict truck productivity. However, previous studies have not considered the temporal-resolution effect of weather conditions, which may cause unreliable prediction results. This study built prediction models of truck productivity incorporating real-site weather conditions with varying temporal resolutions (hourly, daily, weekly, and monthly) for the first time. After that, the best models were combined with SHapley Additive exPlanations (SHAP) to offer quantitative and qualitative analysis for each input variable’s effect on the model outputs. The results showed that the nonlinear relationship between input variables and truck productivity progressively diminished with decreasing temporal resolutions (i.e., from hourly to monthly). Mining engineers can make more accurate predictions of truck productivity at the weekly resolution compared with other resolutions. Regardless of the temporal resolutions, the three most influential input parameters were haul distance, empty speed, and ambient temperature. Extreme weather, such as strong wind speed, heavy precipitation, and extreme relative humidity, had a certain effect on truck-shovel allocation at mine sites. Meanwhile, a unified graphical user interface was designed and developed to predict hourly, daily, weekly, and monthly truck productivity in open-pit mining. This study contributes solutions for predicting truck productivity at varying temporal resolutions more easily and quickly and provides new insights into mine planning and decision-making for mining engineers.