Xuedong Zhou, Dr. Mitch Wolff, Wright State University, USA

Turbomachinery flutter/ forced response is a physical phenomena strongly related to the unsteady aerodynamics of the jet engine which can result in a high cycle fatigue structural failure. To accurately simulate the physical phenomena, a fully coupled structural and fluid analysis is required to predict the blade behavior.

As an initial step toward this goal, research has been accomplished utilizing a prescribed motion of the blades. STAR-CD is employed for the numerical analysis of the NASA Glenn transonic oscillating linear cascade of advanced design blades. The blades have negative chamber starting from leading edge, which provides a very challenging configuration because of significant separation on the suction surface. The center blade oscillates 0.6° about the mid-cord. The numerical simulation is conducted for a frequency range from 200~500 Hz.

A hybrid grid, with 1.2 million cells per passage, which utilizes a structured O-grid around the airfoil and an unstructured grid elsewhere, is employed.
The Spalart-Allmaras (S-A), one equation low Reynolds turbulence model, is employed for both steady and unsteady analysis.

Figure 2 shows the steady velocity vectors and pressure contours distribution on suction surface. The separation starts from the leading edge and reattaches at close to 50% chord at mid-span. The CFD prediction also clearly captures the passage vortexes due to the end wall effect at both the tip and hub sides. The predicted solution matches very well with oil trace flow visualization on the suction surface from NASA.

The prescribed oscillating motion is modeled by an enhanced algebra method, implemented via a STAR-CD user subroutine. A series of CFD simulations with different oscillation frequencies and inlet Mach numbers was conducted with all nine blades and end wall modeled by a hybrid mesh. The suction surface pressure time history adjacent to the center oscillating blade is post-processed by Fast Fourier Transformation (FFT) to obtain the first harmonic of pressure coefficient which is compared with the experimental data. Figure 3 shows the comparison at Mach 0.8 and 300Hz oscillation frequency. The first harmonic magnitude matches very well with the experimental data while there is some phase error.

Overall, current model provides an excellent agreement with the experimental data. Thus, the modeling method is a good candidate for fluid-structure interaction simulations for turbomachinery.