Dense gas dispersion model development and testing for the Jack Rabbit II phase 1 chlorine release experiments

Abstract

In 2015 and 2016, the US Army conducted a series of large-scale chlorine releases at the Dugway Proving Ground in Utah, known as the Jack Rabbit II trials. The purpose of these experiments was to improve our understanding of pressure-liquefied chlorine releases and atmospheric dispersion, and provide useful practical knowledge for emergency responders. The tests conducted in 2015 featured a grid of Conex shipping containers around the release point to simulate a mock urban array of buildings, while the tests in 2016 studied different release orientations and culminated in the full discharge of a 20-ton chlorine road tanker. Before, during and after these experiments, a group of dispersion modelling experts from around the world collaborated in simulating the tests to help configure the experiments and evaluate the performance of models. This paper presents the progress made in that modelling activity by two of the groups involved: the UK Health and Safety Executive (HSE) and the US National Center for Atmospheric Research (NCAR). The models tested by HSE and NCAR range in complexity from integral models that can be run quickly for emergency response to Computational Fluid Dynamics (CFD) models that require days of computing time. This paper summarizes these groups' analysis of the 2015 experimental data and initial model findings, and outlines future research directions. The discharge model predictions by HSE show that meta-stable models tended to over-predict the measured release rate from the chlorine tank, whilst flashing models under-predicted the release rate. CFD simulations of the near-field flow behavior show that the chlorine cloud was initially directed laterally, out of the sides of the mock urban array, due to the alignment of Conex containers. The two integral models tested by HSE (DRIFT and PHAST) provide best agreement with the downwind concentration data when they take into account the rainout of liquid from the impinging two-phase jet. The two models tend to over-predict concentrations slightly, but many of the measurements may have under-recorded the peak concentrations, due to sensors saturating and the clouds bypassing sensors. The newly-developed NCAR integral model accounts for the transition from momentum to buoyancy-dominated behavior and shows promising agreement with the CFD results and measured concentrations. In addition, it reproduces the observed −5/3 power law decay in concentration at large distances. Efforts are continuing on analysis of the Jack Rabbit II experiments and a collaborative international model inter-comparison exercise is currently underway. Additional papers by the coordinators of that exercise and other experts should be forthcoming. The Jack Rabbit II dataset will no doubt be used for decades to come as a seminal test case for validating flashing jet and dense gas dispersion models.