Model and mechanism of the impact of water immersion process on the minimum ignition temperature of coal dust

Abstract

Coal dust explosions pose a major threat to the safety of industrial processes involving coal handling and utilization (e.g., coal mining, coal processing, and coal-fired power generation). The Minimum Ignition Temperature (MIT), as a core parameter for evaluating the risk of coal dust explosions in industrial process safety management, is influenced by multiple coupled factors including water immersion time, coalification degree, volatile matter content, and particle size distribution. This study systematically investigates the mechanism by which coal dust characteristics affect the MIT in the context of industrial water-related coal handling processes and builds a multi-factor predictive model using experimental testing and machine learning methods—with the goal of providing a tool for process safety risk mitigation. The Godbert-Greenwald furnace was employed to measure the MIT of coal dust clouds under various water immersion conditions. Key influencing factors were identified through Pearson and Spearman correlation analyses, with a focus on their relevance to process parameter optimization. The XG-Boost algorithm was utilized to develop a predictive model with features such as water immersion time, volatile matter content, active functional group content, median particle size, dust cloud concentration, and wettability. The results indicate that the volatile matter content (Pearson coefficient −0.78, p < 0.001) and active functional group content (Spearman coefficient −0.71, p < 0.001) are strongly negatively correlated with MIT, serving as key determinants influencing MIT in coal-related industrial processes. Water immersion time shows a moderate negative correlation with MIT (Spearman coefficient −0.50, p < 0.001), with prolonged immersion reducing MIT by 60°C—this elucidates how moisture (a controllable process factor) changes the hydroxyl content and pore structure of coal dust surfaces, thereby lowering the activation energy of oxidation and increasing process safety risks. The XG-Boost model ranks feature importance as follows: volatile matter content > active functional group content > water immersion time > wettability > dust cloud concentration > median particle size—providing clear guidance for prioritizing process parameter monitoring. The determination coefficients (R²) for the model training and testing datasets are 0.9999 and 0.9512, with average absolute errors (MAE) of 1.470 × 10⁻⁴ and 1.647, demonstrating a high level of predictive accuracy for supporting real-time process safety decision-making. This study offers a theoretical foundation for the dynamic assessment of coal dust explosion risks in industrial processes with variable coal quality and controllable process parameters. It is advised that in industrial process safety practice, emphasis should be placed on monitoring volatile matter and active functional group content of coal (pre-process quality control), while optimizing process safety prevention and control strategies through the regulation of water immersion time (process parameter) and wettability (process control index).