Computational Study of Upstream-Propagating Potential Disturbances in an Axial Flow Compressor

One of the most common causes of aircraft engine failure is the high cycle fatigue of rotor and stator blades. High cycle fatigue is often the result of neighboring blade row interactions that cause an aerodynamic excitation to be resonant with a mechanical natural frequency. Historically, the primary source of the aerodynamic excitations has been limited to viscous wakes (and later the inclusion of shocks), even though several different types of blade row interactions are known to occur. Advances in the design of turbomachinery have resulted in more closely spaced and highly-loaded blade rows. These changes have caused designers to revisit potential disturbances, once thought to be unimportant.

As part of this rethinking, a single stage, axial fan of an F109 engine was tested in a previous experimental investigation at the Air Force Academy, which obtained unsteady velocity and pressure measurements upstream of the fan rotor. The experiments showed that, for some operating conditions, the magnitude of the potential fluctuations were surprisingly large, contrary to the notion that potential disturbances are negligible compared to the other interaction types. However, several questions were left unanswered regarding such potential disturbance interactions and in recent years, both experimental and numerical blade-row interaction investigations have shifted their focus to transonic stages.

The purpose of the present study was to revisit the former F109 potential disturbance investigation by performing a two-dimensional, time-accurate, RANS simulation on the fan stage using the commercial code, FLUENT. As in the experimental investigation, upstream measurements were taken in both a rotor-only and inlet guide vane (IGV) test configuration. In the rotor-only configuration, the unsteady velocity magnitude and phase from the numerical study agreed with the experimental data providing validation for the computational setup. In the IGV configuration, unsteady pressure measurements were extracted from the IGV surfaces, as done in the experiments. While the agreement of the numerical and experimental data was not as close as that for the rotor-only configuration, the pressure response was similar in amplitude, and much of the character of that observed experimentally. Further, data analysis was performed using the Proper Orthogonal Decomposition (POD) method, highlighting the method's ability to efficiently approximate an initial data set using only a limited number of modes.