Date of Award
Master of Science (MS)
In this thesis, a non-intrusive uncertainty quantification (UQ) method is used to improve the accuracy of a wall-modeled large eddy simulation (WMLES) model. Detailed UQ studies focusing on the closure coefficients of two LES models are performed. A non-intrusive polynomial chaos model is used to evaluate output sensitivities and uncertainties in the entire flow domain. The proposed UQ method allows for the investigation of specific flow features and phenomena within the domain. The results of the UQ analyses are then used to identify which turbulence model closure coefficients most influence the flow features of interest. Sobol indices of closure coefficients ( in Smagorinsky model and in WMLES model) are obtained. Based on the magnitudes of Sobol indices, refinements are then made to the closure coefficients of interest to improve the accuracy of the turbulence models. OpenFOAM is used as the flow solver and the UQ analyses are conducted with DAKOTA. The UQ method is applied to the channel flow at different Reynolds numbers. The refined LES turbulence models with modified closure coefficients show improvement in the prediction of the skin-friction coefficient in the channel flow.
Dave Peters Swami Karunamoorthy