Application of machine learning methods for process safety assessments

Journal article proposing a neural-network surrogate model to accelerate computationally intensive CFD studies used in gas dispersion and gas explosion consequence analysis for oil and gas processing facilities. By learning from large CFD scenario sets, the model predicts flammable cloud sizes and explosion overpressures much faster than running full simulations. For Process Safety Management, this can materially improve hazard identification and quantitative risk analysis by enabling broader scenario coverage (e.g., many leak rates/wind conditions) during design reviews, QRA updates, and emergency planning—without prohibitive simulation time.