A UQ based calibration for the CFD modeling of the gas dispersion from an LNG pool

The modeling with Computational Fluid Dynamics (CFD) of gas dispersion from a liquified natural gas (LNG) pool is investigated in detail to better elucidate the sources of uncertainties and the influence of physical phenomena, such as convection and diffusion, just above the pool. Indeed, a better comprehension of these topics can improve gas dispersion analysis and aid the implementation of mitigation measures. However, the literature shows a lack of knowledge on this matter, since the LNG pool inlet conditions have not been precisely analyzed so far. To this purpose, the present work proposes, for the first time in this field, the application of an Uncertainty Quantification (UQ) technique to calibrate the inlet conditions of a CFD model for cloud dispersion from a LNG pool. More specifically, the Burro test series is used to validate numerical simulations based on the solution of Unsteady Reynolds-averaged Navier-Stokes (URANS) equations. As the LNG is released into a water pool, the real LNG pool radius is unknown. Moreover, the gas release is also unknown as it is not equal to the LNG spill rate. The generalized Polynomial Chaos (gPC) expansion is therefore used to estimate these uncertain parameters, by minimizing the errors between CFD and available measurements. The optimization performed on the Lower Flammable Limit (LFL) concentration maps shows how this procedure can give a very good agreement with the experimental observations, extending the accuracy of CFD simulations within risk assessment studies. Besides, this approach highlights how the influence of convection and diffusion on the simulation results strongly depends on the wind conditions. In this manner, the present work can help modelers to better setup CFD simulations with the purpose to aid the decision making in the process safety framework.