Benchmark analysis of accident modelling software applied to hydrogen storage

Global greenhouse gas emissions are reaching unprecedented levels, driving a strong interest in decarbonization. Hydrogen, with applications in industry, transport, and power sectors, offers a CO2- free energy vector when produced renewably. However, the transition to hydrogen as a replacement for fossil fuels presents significant technical challenges including sustainable production and, crucially, safe operation, transportation and storage. Quantitative Risk Analysis (QRA) encompasses a comprehensive methodology to characterize risks widely used in risk management. Since risk depends on the probability and consequences of failure, accurately estimating consequences is essential for precise risk assessment. Various software tools assist in consequence analysis, and comparing those helps identify their strengths and limitations, improving risk management decisions. In this paper we present a benchmark analysis aimed at evaluating the capabilities of Phast and HyRAM+ software tools in modelling dispersion and fire incidents of gaseous hydrogen from losses of containment in pressurized tanks. Real-world experimental results have been used as a ground truth for comparing simulation outcomes, ensuring accurate and reliable assessments. Three different experimental studies (Ekoto et al., 2012, Han et al., 2014; Carboni et al., 2022;) were chosen to assess concentration levels of hydrogen clouds, and flame length and radiation exposure from horizontal hydrogen jet fires. Simulated experiments investigated storage pressures ranging from 60 to 400 bar, with release hole diameters spanning from 0.5 mm to 52.5 mm for large-scale hydrogen jets, defined here as flames exceeding 15 meters in length. A comparison of the results obtained for the dispersion assessment reveals that both software tools generally underestimate hydrogen concentrations, with Phast showing less pronounced underestimation than HyRAM+. On the other side, both software tools generally overpredict flame length, except for slight underpredictions in simulations with the smallest release diameter. For large-scale hydrogen jet fires, Phast tends to overpredict radiation, while HyRAM+ tends to underpredict it. In a QRA framework of hydrogen jets from pressurized tanks, this study suggests that using Phast is preferable due to its more accurate results. However, further validation of the software for large-scale jets is necessary.