Uncertainty Analysis of Phast's Atmospheric Dispersion Model for Two Industrial Use Cases

We have undertaken an uncertainty analysis of the dispersion model of a widely used tool for consequence assessment, comparing the level of output variability observed for an accident investigation use-case (where input variables concerning the release conditions are uncertain) and a risk prevention use-case (where the effect of uncertainty in internal model parameters is evaluated). As expected, for the two flammable and two toxic materials studied, uncertainty for the risk prevention use-case is significantly lower than that for accident investigation. We have identified the release conditions which lead to the highest level of variability in model outputs.