Leakage from the overhead hydrogen piping (OHP) must be contained because of the wide flammability range and low minimum ignition energy of hydrogen; rapid leakage detection and appropriate emergency responses are necessary to ensure safe OHP operation. Consequence analysis after leakage detection is useful for hazardous-area estimation and assists operators in emergency-response planning. However, consequence analysis and leakage-detection methods have not been combined so far, and an integrated method has not yet been developed. This study aimed to develop a method based on machine learning (ML), physical modeling, and consequence analyses for rapid and appropriate decision-making after hydrogen leakage. Initially, an OHP model was constructed using physical modeling; this is a method that models target systems based on fundamental physical equations. Subsequently, an ML model for leakage detection was constructed based on various pressure or flow-rate profiles obtained from the OHP model. Regression models were constructed to estimate the leakage diameter. In addition, consequence analyses were conducted to estimate the effects of hydrogen explosions and fires using the leakage diameters predicted by the regression model. We confirmed that the ML model could distinguish between leakage and non-leakage conditions, and the estimated consequences could be used to visualize the risk level of hazardous areas near the leakage points in the case studies. The results of the case studies demonstrated the effectiveness of the proposed method for rapid decision making when hydrogen leaks from OHP. This method enables improved emergency responses in post-leakage situations.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call