Modal Analysis of High-Fidelity Simulations in Turbomachinery

Abstract

We revisit recently published high-fidelity implicit large eddy simulation data sets obtained with a high-order discontinuous Galerkin spectral element method and analyse them using Proper Orthogonal Decomposition (POD) as well as Spectral Proper Orthogonal Decomposition (SPOD). The first configuration is the MTU T161 low-pressure turbine cascade with resolved end wall boundary layers in a clean version and one with incoming turbulent wakes. We focus on the behaviour of the laminar separation bubble and the secondary flow system and how these phenomena are affected by incoming wakes. The second configuration is a transonic compressor cascade in which we analyse the unsteady behaviour of the shock wave boundary layer interaction. Throughout the paper, we try to discuss not only the flow physics but also insights into how the modal decomposition techniques can help facilitate understanding and where their limitations are. POD can, e.g. lead to misconceptions due to its missing frequency resolution capability. These can be avoided by using SPOD, which, on the other hand, does need more data to achieve a similar signal-to-noise ratio in the computed modes. We conclude that these techniques are most effective, when used in combination with other unsteady analyses.