Numerical Analysis of Flutter in Variable Geometry Compressors

Aeroelastic behavior of a transonic rotor in a newly designed 1.5 stage compressor with variable geometry is studied numerically in this paper. The stage is intended to be the front part of one-shafted large frame industrial gas turbine (IGT) compressors. The compressor was designed using open source software MULTALL and numerical computations were performed using the three-dimensional aeroelasticity code AU3D which has been developed for aero-engine components.

In the first step, the aerodynamic damping of the 1st flapwise bending mode along 100% speedline was computed to determine the least stable nodal diameter as well as the contribution from the shock movement. It was found that the blade is stable at all the operating conditions. Secondly, flutter stability at various off-design conditions typically experienced in IGT (varied inlet temperature and inlet guide vane angle) was studied. Although in all the cases the rotor remained stable, clear trend in aerodynamic damping was observed which can be explained by shock position. Moreover, for some operating conditions, a decrease in blade frequency resulted in flutter.

Considering the recent understanding gained on the driving parameters for “flutter-bite” on fan blades in aero-engines, flutter stability at part-speed was investigated in the next phase. It was concluded that, part-speed flutter is unlikely to occur for the present rotor as the blade vibration mode frequency is never in the critical range, where the acoustic wave is cut-off downstream and cut-on upstream. However, instability can be observed if the blade frequency was artificially set to be in the critical range, which supports the findings in previous work.

In the last phase, two variants of rotor geometry were generated and studied. In the first trial, the rotor was made forward-swept using the same sectional airfoils stacked to the curved line. The swept rotor showed improved stability due to displaced mean shock location as well as the enhanced efficiency and stall margin. In the second trial, enlarged tip gap and large incidence at hub were introduced to the datum rotor. Unsteady computations without blade motion showed a typical tip rotating instability with 11 cells rotating at 84% of the mechanical speed of the rotor in the stationary frame. The frequency of this unsteadiness was very close to one of blade’s natural frequencies, hence large amplitude displacement caused by lock-in was observed in the fluid-structure coupled simulation. Even when the blade stiffness was artificially varied around actual value (by -35% to +50%) the blade was still excited by the lock-in of the rotating instability, where it adjusts the cell count to maintain the same excitation frequency as the blade vibration frequency while keeping the propagation speed approximately constant.