Modeling of multisystem coupling evolution and identification of key control factors for liquefied petroleum gas leakage accidents

Liquefied petroleum gas (LPG) leakage accidents feature sudden onset and complex evolution. Existing research lacks analysis of dynamic coupling and quantitative intervention strategies, this study employs system dynamics and AnyLogic simulation to model five subsystems: personnel, equipment, environment, emergency response, and management. Results reveal a four-stage evolutionary pattern: latent accumulation, risk manifestation, controlled decline, and sustained stability. The equipment subsystem has the greatest influence, with a 50 % initial risk increase raising the peak risk by 17.55 %. The most sensitive factors within these subsystems are lack of regular maintenance, inadequate safety education, and natural wind speed, with risk change rates of 29.71 %, 28 %, and 25.25 %. Crucially, the emergency trigger thresholds exhibit the highest risk change rate of 219.33 %, significantly surpassing other emergency factors and underscoring their pivotal role in risk control. The management subsystem demonstrated sustained effectiveness, with the cost of violation at 32.55 % initial change rate. In summary, personnel and equipment factors are the primary drivers of risk accumulation. Emergency trigger thresholds represent the critical juncture for risk convergence, while management subsystems provide long-term constraints. This study reveals the dynamic evolution mechanism of LPG leakage accidents from a multisystem coupling perspective, offering quantitative evidence for phased risk management strategies.