The Current Gap Between Design Optimization and Experiments for Transonic Compressor Blades

Authors

Edwin Joseph Munoz Lopez  - Deutsches Zentrum für Luft- und Raumfahrt, Germany
Alexander Hergt - Deutsches Zentrum für Luft- und Raumfahrt, Germany
Manfred Beversdorff - Deutsches Zentrum für Luft- und Raumfahrt, Germany
Sebastian Grund - Deutsches Zentrum für Luft- und Raumfahrt, Germany
Wolfgang Steinert - Deutsches Zentrum für Luft- und Raumfahrt, Germany
Volker Gümmer - Technical University of Munich

Abstract

The design of compressor blades for transonic flow conditions is heavily supported today by numerical optimization algorithms and advanced CFD solvers. This allows designers to thoroughly search a given design space to find the optimized member that best satisfies the objective functions. However, a successful design process will always rely on the accuracy of the flow solver employed. In fact, it is notoriously difficult to capture the performance of these blades with the RANS solvers typically employed due to the complex aerodynamic phenomena present. Therefore, in order to assess the current gap between the numerics and experiment within the context of optimization design processes, the Transonic Cascade TEAMAero (TCTA) was tested at the Transonic Cascade Wind Tunnel (TGK) at the DLR in Cologne. This cascade was recently designed at the department through an optimization process supported by the DLR’s RANS solver, TRACE. The cascade was tested at its design Mach number of 1.2 and over its working range with inflow angles between 145° and 147°. The size of the sidewall boundary layer was controlled with suction slots placed in the passage of the cascade. Measurements were made at the inlet, over the blade span, and at the outlet with different measurement techniques. Schlieren visualization of the shock structure was also performed at different operating points. The results indicated generally satisfactory agreement with the overall performance expected from the design process, and with the additional set of CFD simulations prepared for the experiments. Notably, the cascade was found to choke at similar inflow angles, and it was shown to provide a high level of loading and flow turning with relatively small losses around 10% in terms of total pressure loss coefficient. However, some key differences were still observed in the characteristics of the wake, as well as a stronger dependence of the area contraction and the lowest inflow angle achieved. This comparison with experimental results validates the process followed to design the TCTA by demonstrating its generally good performance over the expected working range. However, it also demonstrates the limitations of the RANS solvers, which fail to capture some of the details that affect the cascade's performance and must be eventually taken into account with undesirably large design margins or compromised working ranges.