Comparison of k-ε models in gaseous release and dispersion simulations using the CFD code FLACS

Several model validation studies on gas dispersion scenarios have been conducted in the past on the Reynolds averaged Navier Stokes (RANS) based eddy viscosity turbulence models. However, many of these studies are based on a limited number of validation cases involving simple geometries and conformal mesh. In the area of safety engineering, the application of RANS-based CFD for consequence analysis is a widely used methodology. Best practice on use of CFD in this context, as the document developed in the COST Action 732 (Franke et al., 2007), focus primarily on validation and verification aspects as well as simulation setup and definition of input data. Guidelines on turbulence models also exist, among which the ERCOFTAC CFD Best Practice Guidelines, and the works of Moloney et al. (2016) and Mcencle et al. (2001). However, there is no unique recommended model for dispersion simulations. The objective of the present study is to assess the three well-known RANS eddy viscosity models, namely, Standard k-epsilon, Re-Normalization group (RNG) k-epsilon and Realizable k-epsilon, in a representative range of gas dispersion cases by comparing models' behavior with experimental data. The current validation cases include dense CO2 release in a cross-wind, impinging hydrogen jet, and a dense chlorine jet release in an industrial site. All the simulations were conducted using the commercial CFD code FLACS. Turbulence models were assessed based on the ability to reproduce experimental concentrations, required computational-time and numerical-stability. Overall, Standard k-epsilon and RNG k-epsilon models were found to be reasonably good in all cases. Nevertheless, Realizable k-epsilon model shows promise in yielding good results in cases involving complex-geometries and dense-phase gas-releases. These results may also be explained with the interplay between the Porosity/Distributed Resistance subgrid models used in FLACS and turbulence models. (C) 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.