Towards Resolving Natural Shock Oscillation and Mitigation of Altitude Excitation in a Transonic Fan

  • Nel, Philipp (Rolls-Royce Deutschland)
  • Pirozzoli, Sergio (University of Rome)
  • Swoboda, Marius (Rolls-Royce Deutschland)
  • Weiss, Julien (TU Berlin)
  • Grothe, Patrick (Rolls-Royce Deutschland)

Please login to view abstract download link

A laminar boundary layer persists on the suction side of transonic fan at high altitudes, leading to laminar shock-boundary layer interaction, shock oscillation and the consequent excitation of structural blade modes. Structural excitation of a problematic mode may be mitigated by tripping the suction side boundary layer. The oscillation mechanism however, is not understood: it is not understood whether when mitigated, the (aerodynamic) oscillation amplitude reduces, or whether a shift toward off-resonant frequencies occurs. Furthermore, the dimensionality of the oscillation mechanism is not clear (2D vs. 3D). This also means that the requirements for resolving the oscillation are unknown. It is known experimentally that the structural excitation occurs for a specific range operating conditions at high blade loading. Therefore, a representative operating condition within this range is investigated in both full span and Q2D configuration, using U/RANS and LES. In the case of U/RANS, an uncertainty study is performed in both Fine/Turbo and the in-house RollsRoyce Hydra CFD codes. The results suggest that U/RANS does not capture the physics required to resolve the oscillation, since only numerical sources of unsteadiness are found. Although RANS and URANS initially showed different results, the uncertainty study leads to reconciliation in showing no physical sources of oscillation. Numerical sources of oscillation are discussed, and include multi-grid sweeps and sensitivity to CFL, as well as the mesh itself which may cause oscillations in URANS although RANS is sufficiently mesh independent. In the case of URANS, numerical sources of oscillation are damped out by the inertial term. In the case of RANS, it may be suggested that numerical sources of oscillation show the region where the physics are sensitive (convergence problem). Another problem in the case of URANS is the inability to capture interactions between stochastic and deterministic unsteadiness which are possibly essential for resolving such an oscillation mechanism. In the case of LES, significant physical shock oscillations are observed, contrary to U/RANS. A sensitivity study is performed on the resolved oscillation in order to investigate the oscillation mechanism. A transition strip is used to investigate the effect of suction-side boundary layer transition on the oscillation. The oscillation mechanism is analysed and it is concluded that the mechanism is of a 2D nature.