Advancing model evaluation protocol for gas release risk in energy transition

Abstract

The coupling of fossil fuel-based hydrogen production with Carbon Capture, Utilization, and Storage (CCUS) technology represents an effective low-carbon emission approach for hydrogen production. The use of predictive models can more efficiently capture potential hydrogen and CO₂ leakage risks. However, the validation procedure within the current Model Evaluation Protocol (MEP) remains unclear. This study aims to improve the validation procedure by proposing a refined sampling-based method for evaluating gas dispersion models. A point-wise concentration sampling framework is introduced to enhance the consistency and reliability of Statistical Performance Measure (SPM) calculations. The method is applied to evaluate the predicted performance of five models under hydrogen and CO₂ dispersion. The results show that the PHAST model exhibits broader applicability to both gases, while the C-SAT model performs best for hydrogen. In contrast, other models all exhibit significant deviations. Statistical analysis indicates that this method leads to SPM values more closely approaching the ideal range. The results demonstrate that the proposed sampling method can refine or standardize the existing MEP, facilitating the correct evaluation of the gas dispersion models.