The Influence of Transient Inlet Distortions on the Instability Inception of a Low-Pressure Compressor in a Turbofan Engine

Abstract

While studies on compressor flow instabilities under the presence of inlet distortions have been carried out with steady distortions in the past, the investigation presented here focuses on the influence of transient inlet distortions as generated by variable geometry engine intakes of super- and hypersonic aircraft on the characteristic and the nature of the instability inception of a LPC. The flow patterns (total pressure distortion with a superimposed co- or counterrotating swirl) of the distortions are adopted from a hypersonic concept aircraft. A LARZAC 04 twin-spool turbofan was operated with transient inlet distortions, generated by a moving delta wing, and steady total pressure distortions starting close to the LPC’s stability limit until it stalled. High-frequency pressure signals are recorded at different engine power settings. Instabilities are investigated with regard to the inception process and the early detection of stall precursors for providing data for a future stability control device. It turned out that the transient distortion does not have an influence on the surge margin of the LPC compared to the steady distortion, but that it changes the nature of stall inception. The pressure traces are analyzed in the time and frequency domain and also with tools like Spatial FFT, Power Spectral Density, and Traveling Wave Energy. A Wavelet Transformation algorithm is applied as well. While in the case of clean inlet flow, the compressor exhibits different types of stall inception depending on the engine speed, stall is always initiated by spike-type disturbances under the presence of steady or transient distortions. Modal disturbances are present in the mid-speed range that do not grow into stall, but rather interact with the inlet flow and produce short length scale disturbances. The obtained early warning times prior to stall are adversely affected by transient distortions in some cases. The problem of appropriate thresholding becomes evident. The best warning times have been acquired using a statistical evaluation of the Wavelet coefficients, which might be promising to apply in a staged active control system. This system could include different phases of detection and actuation depending on the current precursor.