(91a) User Beware: When Simple Consequence Models Can Give the Wrong Answers

Over the past few years, the use of computational fluid dynamics (CFD) models for safety analyses of LNG facilities in the United States has been steadily increasing.  However, with the notable exception of offshore terminals, for which the use of CFD models is required by the U.S. Coast Guard, the general opinion within the industry is that CFD models are too expensive and/or time-consuming, and that simpler models can be used in most cases.  As a result, the main application of CFD models so far has been to simulate scenarios – such as LNG spills into long trenches – for which no suitable approximation can be made that would allow the use of simpler integral models, such as DEGADIS or others.

While it is true that integral models can be run more efficiently than CFD, their application to anything but the simplest of geometries (i.e., flat terrain and no obstacles or obstructions) requires major approximations to be made.  These simplifying assumptions are widely believed to result in conservative consequence predictions, however, this is not always the case.

This paper will provide examples of LNG vapor dispersion hazard scenarios in which the use of an integral model would provide non-physical results and may lead to potentially dangerous situations.  Examples will include a vapor cloud dispersion in a realistic plant (with storage tanks, piping etc.), a jetting release through a pipe rack and a vapor cloud ignition in a congested area.