441636 Modelling and Validation of Dispersion Following an Instantaneous Vapor or Two-Phase Release from a Pressurized Vessel

Tuesday, April 12, 2016: 2:30 PM
370 (George R. Brown )
Henk W. M. Witlox, DNV GL Software, London, United Kingdom, Mike Harper, DNV GL Software, Manchester, United Kingdom and David Webber, Integral Science and Software Ltd., Warrington, United Kingdom

Hazardous flammable and/or toxic materials are often stored in pressurised vessels. This paper deals with the modelling of the dispersion following the catastrophic rupture of a vessel leading to an instantaneous loss of containment.

The Unified Dispersion Model (UDM) in the hazard assessment software package Phast can account for two-phase jet, heavy and passive dispersion including droplet rainout and pool spreading/evaporation. Pressurised instantaneous releases are modelled by an initial phase of energetic rapid expansion, and a subsequent phase of dispersion where equations are adopted applicable for unpressurised releases. As part of the current work, the UDM sub-model for modelling the initial phase has been replaced by an improved new sub-model.

This paper first summarises the results of a literature review relating to both experimental and theoretical work. Subsequently it outlines the mathematical equations governing the new UDM sub-model, which is based on more sound physical principles than the old model. It allows for both vapour and two-phase releases. In the case of two-phase releases, droplets are assumed to expand radially, potentially leading to time-varying droplet rainout and the formation of a spreading evaporating liquid pool.

For ground-level vapour or two-phase releases the correctness of the numerical predictions are confirmed against an analytical solution.  

The paper finally describes the validation of the model against previously published experimental data for pressurised vapour releases and pressurised two-phase releases with and without rainout. This includes validation against ground-level pressurised releases for nitrogen vapour and flashing liquid propylene, and against elevated flashing liquid releases for Freon 11, Freon 12, propane and butane. Overall the old model tends to under-predict the cloud radius and cloud speed versus time, while the new model more closely agrees with experimental data. Therefore the new model produces smaller concentrations and doses, and is less conservative. For two-phase releases the new model predicts an increased amount of rainout, which again is more in line with the experimental data.

Extended Abstract: File Uploaded