RISK ANALYSIS OF AN HYDROGEN REFUELLING STATION: UNCERTAINTY IN CONSEQUENCE MODELLING
A. Carpignano, E. Ponte, F. Ganci
Politecnico di Torino
Turin, Italy
E-mail , ,
SESSION: 1.3 - Hydrogen Risk Assessment and Management – Consequences
Abstract
Uncertainty associated to the risk evaluation of technological systems is of difficult quantification since, even the evaluation of risk itself is not a precisely defined procedure but rather a complicated process of several parts, each part is characterised by its own uncertainty.
All phases of risk analysis are subjected to inaccuracies closely related to the complexity of the examined system.
Some European Benchmarking Exercises (Benchmark Exercise on Major Hazard Analysis 1992, ASSURANCE 2000) have demonstrated that different groups of analysts, facing the safety analysis of a same plant, obtain results very different from each others by orders of magnitude in the point value of risk. These studies highlighted as the tasks which mainly contribute to the overall uncertainty of a risk analysis are the consequence assessment and the vulnerability analysis.
In the case of risk analysis about an hydrogen refuelling station, uncertainty of analysis is higher than in other systems because of newness level of these technologies.
Also in this case, consequence analysis introduces several uncertainties, particularly due to phenomena modelling, because: parametric and simplified models usually applied cannot model the behaviour of a light gas like hydrogen, in particular when an outdoor release is considered; more complex models, as CFD models, that require more time and knowledge by analyst, usually are not validated with reference to hydrogen dispersion in not confined area; finally, available experimental data principally concern confined release but very rarely release in open space.
Beside, it is important to point out that hydrogen behaviour is differently described in literature, especially if the possibility of an outdoor explosion is analysed.
Activity performed and described in the paper is aimed to analyse the capability grade of models usually applied in these studies and compare the results obtained with experimental data and application of more complex models.
In particular, the activity has been articulated in the following phases:
· individuation of accidental scenarios that can occur in case of hydrogen release in a refuelling station;
· modelling of such scenarios (dispersion, jet fire, flash fire, explosion) by different parametric and simplified models (in particular, application of the following software: Effects – TNO and Phast – DNV);
· comparison between results obtained and experimental data performed by a collaborating working group (Pisa University);
· comparison of the previous results with the ones obtained in modelling phenomena by Lagrangian model.
The work described is founded by the Italian Ministry of Research and University (PRIN 2005). This activity belongs to national project performed in collaboration with Polytechnic of Milan and Universities of Pisa, Rome and Palermo.