Enhancing blasting efficiency: A smart predictive model for cost optimization and risk reduction

Abstract

Mineral extraction involves distinct stages, including drilling, blasting, loading, transporting, and processing minerals at a designated facility. The initial phase is drilling and blasting, crucial for controlled dimensions of crushed stone suitable for the processing plant. Incorrect blasting can lead to unsuitable stone grading and destructive outcomes like ground vibrations, stone projection, air blasts, and recoil. Predicting and optimizing blasting costs (BC) is essential to achieve desired particle size reduction while mitigating adverse blasting consequences. BC varies with rock hardness, blasting techniques, and patterns. This study presents a BC prediction model using data from 6 Iranian limestone mines, employing firefly (FF) and gray wolf optimization (GWO) algorithms. With 146 data points and parameters like hole diameter (D), ANFO (AN), sub-drilling (J), uniaxial compressive strength (σ_c), burden (B), hole number (N), umolite (EM),spacing (S), specific gravity (γr), stemming (T), hole length (H), rock hardness (HA), and electric detonators (Det), the data was split into 80% for model construction and 20% for validation. Using statistical indicators, the model showed good performance, offering engineers, researchers, and mining professionals high accuracy. The @RISK software conducted sensitivity analysis, revealing T parameter as the most influential input factor, where minor T changes significantly affected BC. Lastly, the @RISK software was employed to conduct a sensitivity analysis on the model's outputs. The analyses demonstrated that, among the input factors, the T parameter had the most pronounced effect on the model's output. Even small changes in the value of T led to considerable fluctuations in the predicted BC.