Benchmark exercise on risk assessment methods applied to a virtual hydrogen refuelling station

A benchmarking exercise on quantitative risk assessment (QRA) methodologies for hydrogen safety has been conducted within the project HyQRA, under the framework of the European Network of Excellence (NoE), HySafe. The aim of the exercise was twofold: (i) to identify the differences and similarities in approaches in a QRA and their results for a hydrogen installation and (ii) to identify knowledge gaps in the various steps and parameters underlying the risk quantification of hydrogen safety. First, a reference case was defined for the benchmark: a virtual hydrogen refuelling station (HRS) in virtual surroundings comprising housing, school, shops and other vulnerable objects. For the study, a two phase approach was followed. In phase 1, all nine partners were requested to conduct a QRA according to their usual approach and experience. Basically, participants were free to define representative release cases, to apply models and frequency assessments according their own methodology, and to present risk according to their usual format. To enable inter-comparison, a required set of results data was prescribed, like distances to specific thermal radiation levels from fires and distances to specific overpressure levels. Moreover, complete documentation of assumptions, base data and references was to be reported. It was not surprising that a wide range of results was obtained, both in the applied approaches as well as in the quantitative outcomes and conclusions. This made it difficult to identify exactly which assumptions and parameters were responsible for the differences in results. These results provided the basis for a more guided QRA, the second phase. This phase 2 was defined in which the QRA was determined by a more limited number of release cases (scenarios). The partners in the project agreed to assess specific scenarios in order to identify the differences in consequence assessment approaches. The results of this phase provide a better understanding of the influence of modelling assumptions and limitations on the eventual conclusions with regard to risk to on-site people and to the off-site public. (c) 2010 Published by Elsevier Ltd on behalf of Professor T. Nejat Veziroglu.