Toward reliable green hydrogen production: dynamic risk assessing of PEMEC via molecular and system-level modeling

Abstract

This study presents an integrated risk assessment framework for PEMEC systems that advances existing methodologies by incorporating a Weibull distribution-Markov chain integration to refine the modeling of progressive degradation events. The framework combines molecular dynamics (MD) simulation with equation-oriented modeling of PEMEC degradation scenarios to generate high-fidelity datasets, and employs fuzzy dynamic Bayesian networks (fuzzy DBNs) to reveal the dynamic risk evolution of PEMEC systems. Molecular-level insights into membrane chemical degradation, obtained from MD simulations, are quantitatively linked to macro-scale performance loss through equation-based modeling of electrochemical reactions and transport phenomena. Multi-condition operational datasets are generated from Aspen Custom Modeler and used to support the fuzzy DBN for risk evolution, The risk analysis framework not only consider degradation but accounts for multi-source failures: equipment, human, environmental, and management for comparation. All root events from real-world PEMEC factory and applying Leaky Noisy-OR logic to address unmodeled risk factors. The risk model reveals a nonlinear accumulation of system failure probability, increasing from 16.01% to 31.12%. Key risk factors include overpressure, abnormal voltage and current, thermal runaway, and manufacturing defects, mainly from equipment. Although degradation impacts PEMEC performance, its influence on safety is limited. The assessment framework effectively identifies critical failure pathways and can be extended to other renewable energy technologies such as batteries and fuel cells. Copyright © 2026. Published by Elsevier Ltd.