Evaluating Ignition Probability of Hydrogen Releases in Alkaline Electrolyzers

SUMMARY & CONCLUSIONSAs global energy transition advances, hydrogen is emerging as a clean and versatile fuel. However, its high flammability poses significant safety challenges. Accurately estimating ignition probability is essential for effective risk management, particularly in environments where ignition sources cannot be fully eliminated. Traditional methods often rely on methane-based models which may not precisely reflect hydrogen’s behavior. While computational fluid dynamics (CFD) simulations provide detailed insight, their high computational cost limits their use in multi-scenario assessments. The current study presents a semi-quantitative Bayesian network-based model for hydrogen ignition probability estimation that bridges the gap between simplified surrogate approaches and complex CFD tools. By incorporating key ignition factors and explicitly modeling their causal relationships, it addresses both immediate and delayed ignition scenarios. Applied to a 10 MW alkaline electrolyzer, the model estimated the probability of immediate ignition as 0.049 for a leak and 0.071 for a rupture, while the probability of delayed ignition was 0.047 for a leak and 0.095 for a rupture. This approach provides a practical, scenario-based tool for improving hydrogen safety assessments and supporting its safe, large-scale deployment.