Advanced methods for assessing flow physics of the TU Darmstadt compressor stage: Uncertainty quantification of RANS turbulence modeling

Abstract

In this paper we quantify the turbulence modeling uncertainty for the transonic TU Darmstadt (TUDa) compressor. The present work applies the Eigenspace Perturbation Framework (EPF), as it is the only published physics-based framework capable of addressing the model-form uncertainty in turbulence closure modeling. To sample from the possible solution space and obtain the modeling uncertainty, we perform simulations perturbing the eigenvalues of the Reynolds stress tensor in addition to simulations using an unperturbed turbulence model. We show, that the shape of the Reynolds stress tensor ellipsoid has significant impact on the evolution of turbulence, flow separation, vortex systems, shock-boundary layer interaction and finally the overall performance of the compressor. We compare the estimated uncertainties with available measurements and transitional Delayed Detached-Eddy Simulations (DDES). This allows us to assess the confidence of the chosen turbulence model and to evaluate the sharpness and coverage of the resulting uncertainty bounds. Thus, the EPF is comprehensively validated and suggestions for its future applicability with respect to turbomachinery components are made.