Abstract Routine monitoring of methane concentration in underground coal mines yields measurements that remain well below regulatory limits while exhibiting measurable variability influenced by environmental and operational conditions. This study presents a probabilistic modeling framework for examining methane concentration variability and ventilation sensitivity using routine monitoring data from an underground coal mine in the Soma region of Türkiye. Continuous measurements of methane concentration, air temperature, and relative humidity were recorded at five-minute intervals throughout 2023, yielding 105,120 observations. Probability distributions were assigned to each variable using maximum likelihood estimation, with family selection supported by information criteria and Kolmogorov-Smirnov evaluation. A multiple linear regression response model explained 54% of the variance in methane concentration and identified ventilation performance as the dominant predictor (standardized coefficient − 0.621), followed by relative humidity (0.337) and air temperature (0.212). Monte Carlo simulation with 10,000 iterations propagated input uncertainty through the response model across three ventilation scenarios. Reducing ventilation to 80% of nominal raised the mean methane concentration by 9.3%, while improving it to 120% lowered the mean by 9.6%, with modest changes in dispersion. Quantile comparison between simulated and empirical distributions yielded a root mean square error of 1.3 ppm, confirming the internal consistency of the fitted response model and input distributions with the monitoring data. The framework provides an analytically rigorous basis for uncertainty-aware ventilation management and is transferable to other mining environments provided that distributions and the response model are re-estimated from local observations.
Eyuboglu et al. (Thu,) studied this question.