On the Extrapolation of Rim Sealing Performance From Test Bench to Real Engine: A Numerical Survey

The main annulus hot gas ingress into turbine wheel-spaces is still one of the most challenging problem designers face. During the decades, several experimental test benches were developed worldwide to improve the knowledge associated to the rim seal flow physics. Even if in some cases quite complex and advanced rig configurations were proposed, limitations in the operating conditions and in the reproduction of the real engine geometries/characteristics into the rig are present. In this paper, validated CFD computations are used to explore the impact of some experimental rigs design choices/limitations on the sealing effectiveness prediction and their ability to mimic the real engine configuration behaviour. Attention is paid on several test rig related aspects such as operating conditions, flow path configuration (blade and vane count) and accuracy in the real engine rim seal geometry reconstruction applied to the rig. From the computations it emerges that a scaled geometry operated at lab conditions is able to mimic pretty well the real engine sealing performance when rig and engine experience the same flow path ΔCp. The ability of the rig to match the engine data is not affected by the differences in main annulus Mach number between test bench and engine. A further result that emerges from the computation regards the fact that the Φ0 - ΔCp0.5 curve is not linear, proving that the linear extrapolation of rim sealing performance from test bench to real engine when rig and engine are characterized by different ΔCp0.5 values is not of general application and an alternative approach is given. Finally, it is found that the impact of vane count on the rim sealing effectiveness is significant, making the extrapolation of data from rig to engine difficult.