Uncertainty Quantification method for CFD validated for turbulent mixing experiments from GEMIX

Abstract

In Computational Fluid Dynamics (CFD), input parameters, numerical methods, and physical models, will introduce uncertainty in the results. In order to assess the reliability of the computations, it is necessary to quantify this uncertainty.The OECD initiated the first international Uncertainty Quantification (UQ) benchmark for CFD in 2015, based on turbulent mixing experiments from the GEMIX facility, where two water layers flow into a rectangular pipe and mix. NRG participated in this benchmark with a UQ method based on the ASME Verification and Validation (V&V) standard for UQ in CFD. The propagation of the uncertainty in the input parameters was sampled with the Latin Hypercube Sampling. The numerical uncertainty was determined by means of Richardson extrapolation.A source of uncertainty which we did not evaluate explicitly was that of the turbulence model input parameters, which depend heavily on the flow conditions. In order to investigate this contribution, we consider the turbulence model parameters as uncertain input variables. The results of this so called Extended ASME methodology are presented and discussed. It is found that the uncertainty in the turbulence model input parameters has a significant contribution to the overall uncertainty in the simulation results for the considered GEMIX mixing experiment.