Abstract
The accurate prediction of transitional flows is crucial for the industrial turbomachinery design process. While a Reynolds-averaged Navier–Stokes approach inherently brings conceptual weaknesses, large-eddy simulation will still be too expensive in the near future to affordably analyze complex turbomachinery configurations. We introduce a transitional delayed detached-eddy simulation (DDES) model, namely, DDES-γ, and analyze the numerical results of the compressor cascade V103. A comparison with the fully
turbulent DDES approach emphasizes the benefit of coupling DDES with a transition model. Issues with undesired decay of modeled turbulent kinetic energy in the freestream are improved when running DDES-γ
in combination with the synthetic turbulence generator method. The best results for DDES-γ are obtained when changing the inviscid flux solver blending from dynamic to constant mode. We show that DDES-γ is capable of
predicting the transitional flow through a linear compressor cascade, but we also critically discuss the general concept and results.
turbulent DDES approach emphasizes the benefit of coupling DDES with a transition model. Issues with undesired decay of modeled turbulent kinetic energy in the freestream are improved when running DDES-γ
in combination with the synthetic turbulence generator method. The best results for DDES-γ are obtained when changing the inviscid flux solver blending from dynamic to constant mode. We show that DDES-γ is capable of
predicting the transitional flow through a linear compressor cascade, but we also critically discuss the general concept and results.
| Original language | English |
|---|---|
| Journal | Journal of Propulsion and Power |
| Early online date | 3 Feb 2024 |
| DOIs | |
| Publication status | E-pub ahead of print - 3 Feb 2024 |
Bibliographical note
AcknowledgmentsThis research did not receive any specific grant from funding agencies in the public, commercial, or non-for-profit sectors. All simulations have been carried out on DLR’s supercomputer CARA (https://www.top500.org/system/179779/) within the internal DLR project ADaMant (https://www.dlr.de/as/en/desktopdefault.aspx/tabid-17581/27883_read-72278). Eventually, we gratefully thank our colleagues at the Department, especially Michael Bergmann, Marcel Matha, and Pierre Sivel, who spent a lot of time for discussions on this test case and the presented model.