Statistical analysis and uncertainty risk assessment of large-scale fatal gas explosions based on Chinese coal mine accidents.

Abstract

Large-scale fatal gas explosions (LSFGEs) are primary accidents in coalmines owing to complicated formation and coupling paths of the causative factors. Therefore, space-time distributions of 282 LSFGEs from 2000 to 2022 in Chinese coalmines were analyzed. A cause-chain inference model of gas explosion was proposed using the Delphi method and Bayesian network (BN). The results showed that working-excavating faces and winter were the main sites and seasons of gas explosions respectively, accounting for 67% and 32.3%. These statistics suggested the ignition sources of 14.5% of LSFGEs were fuzzy, including coal spontaneous combustion and friction sparks. Moreover, instantaneous gas overruns were attributed to periodic roof caving and coal-rock dynamic disasters, accounting for 11.0%. The change in roof pressures and concealment of the enclosed areas were key risk factors of a gas explosion. A comprehensive analysis of cause-chains with expert knowledge and probabilistic inference reduced the epistemic uncertainty of fuzzy causes. The sensitivity analysis highlighted electric sparks and the absence of wind or breeze as the primary causes of gas explosion accidents and its percentage of exceeded 20%. This study aimed to elucidate key causative factors and their interrelationships in gas explosions, providing valuable guidance for daily risk management and accident investigations.