Integrated optimization of underground mine ventilation using hardy cross, Monte Carlo simulations, and machine learning techniques: a step toward sustainable mining

Underground mine ventilation is crucial for ensuring worker safety, air quality, and operational efficiency. Effective optimization of ventilation systems reduces energy consumption, enhances airflow distribution, and minimizes risks associated with harmful gases and fire hazards. This study proposed an integrated framework combining the Hardy Cross (HC) method, Monte Carlo (MC) simulations, and machine learning (ML) techniques to optimize the ventilation system at the Jabal Sayid mine in Saudi Arabia. The HC method was employed to provide deterministic baseline airflow calculations, while the MC simulations accounted for uncertainties in resistance values and environmental conditions, generating probabilistic distributions of key parameters. Among the five ML algorithms tested (ANN, RF, GB, SVM, and LSTM), the LSTM model demonstrated superior predictive accuracy, particularly for dynamic, time-dependent parameters. The hybrid HC-MC-ML model integrated these approaches to achieve comprehensive ventilation optimization. Results indicated a 12.8% reduction in airflow resistance, leading to enhanced fan efficiency and a significant improvement in energy consumption. For instance, fan system efficiency increased by up to 7% in combined operations, while resistance values consistently decreased across all scenarios. Additionally, the hybrid model effectively managed exhaust gas emissions, maintaining pollutant concentrations within permissible limits by dynamically adjusting airflow routes. The findings demonstrate that the HC-MC-ML framework not only improved energy efficiency but also ensured safe and sustainable mine ventilation.