Adaptation of three atmospheric dispersion models for jack rabbit ii case study and comparisons with observations : lessons for predictive modelling around industrial sites

Chlorine release trials were conducted at the Dugway Proving Ground in Utah, during the Jack Rabbit II trials. An interesting database of emissions and dispersion of pressurized liquefied chlorine was set up. This database allows to make comparisons between atmospheric dispersion modelling and large-scale observations about toxic massive releases (up to 10 ton). In this work, three test cases (2 trials free of obstacles and an obstructed environment) were studied on purpose for a model inter-comparison study. INERIS modelled these cases with three atmospheric dispersion models: the open source SLAB code, the commonly used PHAST software (commercial licence) and the 3D code FDS (Fire Dynamic Simulator). This paper focuses on the adaptation of the models and comparisons with observations in the near-field for CFD (Computational Fluid Dynamics) modelling and large field (up to 10 km) for other models. Discussions about harmonization of practices between models and practices for toxic consequences assessment around industrial sites are also proposed. Indeed, a massive release of pressurized liquefied material could generate complex phenomena around the release location: jet under pressure, rain-out, cooling, etc. For each of these atmospheric dispersion models, the source term can be set up in various ways depending on its level of complexity. A similar discussion is proposed about the unsteady meteorological flow that cannot be assessed by the simplest models