UNCERTAINTY EVALUATION OF CFD SIMULATION USING OPTIMAL STATISTICAL ESTIMATOR

摘要

Uncertainty methods are widely applied to evaluate the uncertainty of thermalhydraulic system codes, being the main safety analysis tool in past decades. Rapid increase of computational power along with continuous development of local mechanistic models opens the space for detailed Computational Fluid Dynamics (CFD) codes. Like system codes, CFD predictions can also be affected by uncertainties. Uncertainty evaluation methods for CFD codes are still under development. The main purpose of this study was the uncertainty evaluation of NEPTUNE_CFD results for Generic Mixing experiment (GEMIX). A method called Optimal Statistical Estimator (OSE) has been proposed for response surface generation of CFD predictions, which replace code calculations when using Monte Carlo method to randomly sample the input parameters to quantify the uncertainty. The turbulent mixing experiment GEMIX performed at Paul Scherrer Institute was used as a benchmark case. In the GEMIX experiment two turbulent horizontal channel flows with the same fluid properties and inlet velocities have been mixed together to form a mixing layer flow. Two uncertain input parameters were considered as uncertainty variables: inlet velocity profile and inlet turbulence intensity. The calculational matrix consisted of 30 CFD calculations. The parameter turbulence intensity was varied in equidistant steps of 1 % and the parameter alpha (0 ≤ α ≤ 1) was varied in steps of 0.2. The results of the uncertainty analysis are presented. It has been demonstrated that OSE is a very efficient and accurate method for uncertainty evaluation of CFD calculations when a few parameters are being varied.