Prediction of the unsteady compressor performance by using time and frequency domain methods

Abstract This research focuses on the numerical prediction of a 4.5-stage compressor's throttle curve by means of unsteady time and frequency domain methods resolving the unsteady interaction between adjacent rows. The results are compared to predictions generated by an established mixing plane RANS method demonstrating the limitations of such approaches when applied to predict the characteristics of the overall compressor map as the isentropic efficiency, the occurence of surge as well as thermal loads faced by rear stages. The investigated frequency domain approach following a Harmonic Balance algorithm allows as any Proper Orthogonal Decomposition method to vary the degree of resolved frequencies and nonlinear interactions by choice. Thus, this research evaluates prerequisites being necessary to achieve a fair approximation of the unsteady compressor throttle characteristics when relying on model order reduction. The general capability of nonlinear frequency domain methods as Harmonic Balancing to predict the unsteady compressor performance at substantially lower computational hardware requirements and run times is demonstrated while resolving multi-stage interactions, the impact of relative clocking positions of rows in the same frame of reference and the interaction of rotor blades with adjacent casing treatments and cavities.