Variable guide vanes are airfoils, normally in the front stages of multistage compressors, that can be restaggered to extend the compressor’s operational range. To allow the variable guide vanes to rotate the design will inevitably include a gap or cavity between the vane’s rotating mounting feature (Penny) and the stationary inner and outer sidewalls. These penny cavities cause additional leakages which impact losses, and airfoil turning, so reducing a compressor’s efficiency and stability. A compressor model which accurately simulates the penny cavity leakage is central to improving the design. This paper presents a study looking into how to accurately include penny cavity leakage effects during the design of multistage compressors.
Multiple blade-row RANS-based flow simulations are the current state-of-the-art standard during the design of multistage compressors. However, it is unlikely that such models have the numerical accuracy to simulate the penny cavity leakage effects in detail: firstly the RANS-turbulence model cannot accurately recreate the turbulent mixing which takes place between the leakage and the main flow, and secondly because a typical multiple blade-row mesh is too coarse to resolve the details of the much smaller penny cavity. To compensate for this, the numerical modelling of penny cavities in the design of a compressor would need to be adjusted. On the other hand, a dedicated hybrid LES-model can more accurately simulate the secondary flows with its significant turbulent mixing, at the cost of computational capacity. In this paper, high resolution hybrid LES-simulations have been used as a benchmark to adjust RANS-calculations typical for the design of a multistage compressor.
The paper presents the following steps:
Using a standard Jet-in-Crossflow test case, a high resolution model was evaluated using both RANS and a hybrid LES model, and compared against measurements. The flow structures were analyzed and compared to measurements of this test case available from literature. These show that the hybrid LES-model performed significantly better than the RANS-model, in being able to predict the jet impact and flow structures.
For a second model consisting of a generic compressor variable vane with penny cavity. RANS and hybrid LES simulations were performed with a highly refined mesh in the region of the penny cavity. The modelling is described in detail and the resulting penny cavity effects compared.
Finally, the vane with standard mesh and penny cavity was run using RANS-turbulence CFD and compared to the above. From this conclusions were drawn on how to transfer experience from the higher-fidelity turbulence model to a more industry-standard RANS model, which could for instance be used during the design phase of a multistage compressor.