Prediction of transient turbulent dispersion by CFD–statistical hybrid modeling method

The transient turbulent dispersion of a gaseous contaminant due to accidental leakage or similar critical situation can be studied by computational fluid dynamics (CFD). However, the transient CFD calculation is extremely computationally expensive because highly refined spatial and temporal resolutions are required for accurate results. In this investigation, a CFD–statistical hybrid modeling method was developed to solve the transient turbulent dispersion problem with a substantial computation time-saving. The computational domain of this modeling problem consists of a point source emitting gaseous contaminant and a fixed-point receptor susceptible to the contaminant dispersed in a steady-state turbulent flow field. In the modeling method, CFD is utilized to obtain the transport behavior of an emitted ultra-fine particle that conceptually represents a “microscopic” molecule of a vapor emission. In turn, the principles of statistics are employed to determine the most probable “macroscopic” time-dependent contaminant concentration at the receptor for a given time-varying point-source emission profile. In a case study, the results predicted by the CFD–statistical model agreed reasonably well with published k–ε modeling results and experimental data. The computation time was successfully reduced by one order of magnitude.