Abstract
The objectives of this research are to develop a high fidelity simulation methodology for turbomachinery aeromechanical problems and to investigate the mechanism of non-synchronous vibration (NSV) of an aircraft engine axial compressor. A fully conservative rotor/stator sliding technique is developed to accurately capture the unsteadiness and interaction between adjacent blade rows. Phase lag boundary conditions (BC) based on the time shift (direct store) method and the Fourier series phase lag BC are implemented to take into account the effect of phase difference for a sector of annulus simulation. To resolve the nonlinear interaction between flow and vibrating blade structure, a fully coupled fluid-structure interaction (FSI) procedure that solves the structural modal equations and time accurate Navier-Stokes equations simultaneously is adopted. An advanced mesh deformation method that generates the blade tip block mesh moving with the blade displacement is developed to ensure the mesh quality. An efficient and low diffusion E-CUSP (LDE) scheme as a Riemann solver designed to minimize numerical dissipation is used with an improved hybrid RANS/LES turbulence strategy, delayed detached eddy simulation (DDES). High order accuracy (3rd and 5th order) weighted essentially non-oscillatory (WENO) schemes for inviscid flux and a conservative 2nd and 4th order viscous flux differencing are employed. Extensive validations are conducted to demonstrate high accuracy and robustness of the high fidelity FSI simulation methodology. The validated cases include: 1) DDES of NACA 0012 airfoil at high angle of attack with massive separation. The DDES accurately predicts the drag whereas the URANS model significantly over predicts the drag. 2) The AGARD Wing 445.6 flutter boundary is accurately predicted including the point at supersonic incoming flow. 3) NASA Rotor 67 validation for steady state speed line and radial profiles at peak efficiency point and near stall point. The calculated results agree excellently with the experiment. 4) NASA Stage 35 speed line and radial profiles to validate the steady state mixing plane BC for multistage computation. Excellent agreement is obtained between the computation and experiment. 5) NASA Rotor 67 full annulus and single passage FSI simulation at near peak condition to validate phase lag BC. The time shifted phase lag BC accurately predicts blade vibration responses that agrees better with the full annulus FSI simulation. The DDES methodology is used to investigate the stall inception of NASA Rotor 67. The stall process begins with spike inception and develops to full stall. The whole process is simulated with full annulus of the rotor. The fully coupled FSI is then used to simulate the stall flutter of NASA Rotor 67. The multistage simulations of a GE aircraft engine high pressure compressor (HPC) reveal for the first time that the travelling tornado vortex formed on the rotor blade tip region is the root cause for the NSV of the compressor. The rotor blades under NSV have large torsional vibration due to the tornado vortex propagation in the opposite to the rotor rotation. The tornado vortex frequency passing the suction surface of each blade in the tip region agrees with the NSV frequency. The predicted NSV frequency based on URANS model with rigid blades agrees very well with the experimental measurement with only 3.3% under-predicted. The NSV prediction using FSI with vibrating blades also obtain the same frequency as the rigid blades. This is because that the NSV is primarily caused by the flow vortex instability and the no resonance occurs. The blade structures respond passively and the small amplitudes of the blade vibration do not have significant effect on the flow. The predicted frequency using DDES with rigid blades is more deviated from the experiment and is 14.7% lower. The reason is that the DDES tends to predict the rotor stall earlier than the URANS and the NSV can be achieved only at higher mass flow rate, which generates a lower frequency. The possible reason for the DDES to predict the rotor stall early may be because DDES is more sensitive to wave reflection and a non-reflective boundary condition may be necessary. Overall, the high fidelity FSI methodology developed in this thesis for aircraft engine fan/compressor aeromechanics simulation is demonstrated to be very successful and has advanced the forefront of the state of the art. Future work to continue to improve the accuracy and efficiency is discussed at the end of the thesis.