Enhanced Assessment of Compressor Rotor Stability: Using Computational Fluid Dynamics to Predict Impeller Destabilizing Forces, 18-R9491

Principal Investigator
Jeff Moore

Inclusive Dates:  07/01/04 – 07/01/06

Background - This project will apply Computational Fluid Dynamics (CFD) software to enhance SwRI's state-of-the-art capabilities for evaluating rotordynamic stability of high-performance centrifugal pumps and compressors. Although SwRI has been a leader in establishing effective methods for predicting rotor instability, its treatment of impeller forces now suffers from clear limitations, as do those of others. There is an unmet need for enhanced understanding and methodology.

Approach - During the last ten years, computational methods and computer speeds have reached levels that make it feasible to efficiently and accurately model the flow in a centrifugal compressor. Within the project proposed herein, a CFD model will be prepared for a real compressor, which has exhibited subsynchronous, aerodynamically-excited vibrations. A complete model of this impeller was generated with a proven CFD program—and the output results will be integrated to obtain rotordynamic coefficients. These coefficients will then be compared to the laboratory data collected by a turbine manufacturer.

Accomplishments - The methodology that was used was developed by Dr. Moore for his Ph.D. dissertation. It consisted of a geometric perturbation of the centrifugal impeller and solving of the flow field in the whirling frame of reference. Successive computations at varying whirl frequencies resulted in rotordynamic impedances that produced the stiffness and damping coefficients compatible with most commercial rotordynamics programs. This method was already benchmarked against measurements for pump impellers and labyrinth seals with good results.

The CFD models predicted both the primary and secondary flow passages around the centrifugal impeller. The resulting impedances from a geometric perturbation of the impeller yielded direct and cross-coupled stiffness, damping, and inertia and showed good correlation to the measurements. The results compare favorably by predicting the instability of a full-scale compressor test provided by the turbine manufacturer. By performing parametric CFD runs, critical operating parameters were identified, and an alternative closed form expression is presented. While further validation with other known test cases is needed, the new closed form expression (termed the Moore-Ransom expression) has the potential for improving the accuracy of predicting the aerodynamic cross-coupling forces even when CFD is not employed.

2006 Program Home