(145c) Uncertainties in Modeling LNG Vapor Dispersion with CFD Codes

Authors: 
Qi, R. - Presenter, Mary Kay O'Connor Process Safety Center
Ng, D. - Presenter, Mary Kay O'Connor Process Safety Center, Texas A&M University System
Waldram, S. P. - Presenter, Texas A&M University at Qatar
Mannan, M. S. - Presenter, Mary Kay O'Connor Process Safety Center


Computational fluid dynamics (CFD) models have been widely proposed and developed to predict LNG vapor dispersion behavior and to determine safe separation distances around LNG production storage and transportation facilities. This is largely due to the ability of CFD codes to describe fully the relevant physical processes in complex geometrical circumstances. However, like other consequence models, CFD models for LNG vapor dispersion entail a number of simplifying assumptions, approximations, empirical relations and constants. Therefore uncertainties in predicted results are inevitable. These uncertainties are generally associated with either the numerical solution methods or an incomplete modeling of the physics involved. In this paper, a review of factors which can impact the results of CFD modeling is provided. A sensitivity analysis was conducted to estimate the scale of uncertainties for key parameters. In addition, some recommended methods are introduced to control or reduce the magnitude of uncertainties to a prescribed confidence level.