Quantitative risk model and safety assessment of flammable working fluid leakage and explosion in a limited space

Abstract

At present, to ensure the secure utilization of flammable working fluids, quantitative risk assessment has become a critical issue. In this study, information diffusion, Markov, and Monte Carlo simulation methodologies are employed to develop a probability-based risk model based on combined forecasting and modifying techniques for quantitative risk assessment, focusing on scenarios involving the leakage and explosion of flammable working fluids within limited spaces. This model comprises three distinct components, designed to evaluate leakage and explosion accidents through information diffusion processing of existing accident data, probability proportion calculation based on Markov time series forecasts, and Monte Carlo simulation of leakage conditions and environmental parameters. The case study of the refrigeration systems indicates that, despite the comparatively higher likelihood of igniting for larger leakage scales, the danger of explosion under small leaks is higher due to the high order of magnitude of leakage. As the degree of leakage increases, the risk of explosion escalates considerably, potentially exceeding the risk tolerance range, necessitating the adoption of appropriate risk control measures. The risk results associated with flammable working fluids are compared under identical leakage conditions, revealing that the risk of mixed working fluids in a limited space is largely dependent on the ratio of leaked components. Although the addition of flame retardants to flammable working fluids proves to be an effective risk mitigation measure, it needs to be integrated with consideration of the charge mass in the refrigeration system. Furthermore, the consequences of an explosion can be evaluated using the overpressure model to assess the probability of damage to personnel and buildings. On this basis, the model offers the ability to provide quantitative risk mitigation strategies, and the validation of its calculation outcomes through comparison highlights its applicability across diverse scenarios, particularly in limited space.