Active control of rotating stall in a three-stage axial compressor

Abstract

Stall inception in a three-stage axial compressor has been suppressed over a range of previously unstable operating points through the feedback of velocity perturbations to the inlet flow field. Perturbations were generated using 12 individually actuated guide vanes at the compressor inlet. The operating range was extended by 7.8% to a slope of 0.9 on the pressure rise characteristic. Over this range data describing the compressor's pressure rise and torque were collected. Flow field measurements upstream of the compressor revealed the excitation of spatial harmonics in the annular flow field before stall inception. With or without feedback, a spatial mode was observed to grow into a stall cell without a discontinuity of amplitude or position. The decrease in the stall inception mass flow as a result of damping the critical spatial mode indicated the importance of spatial modes in the stall inception process in this compressor.

Destabilization was originally caused by the first mode. After stabilizing an under-damped mode, the flow range was extended until the next sequential mode became unstable. The independent behavior of the modes before stall inception is described by four versions of a small disturbance model of compressor dynamics adapted from that of Moore and Greitzer. The two more sophisticated versions are presented here for the first time. The models require a description of the compressor's geometry, knowledge of its pressure rise characteristic, and some versions require a lag parameter characterizing the pressure rise response lag. The response lag parameter which enabled the models to most accurately predict the open-loop compressor dynamics was consistent with published values. The open-loop modal dynamics were determined experimentally and could be accurately described by the dynamics represented in the simplest model.

The open-loop dynamics were measured for the first three modes over a range of stable and formerly unstable operating points. The more elaborate models gave accurate predictions of the closed-loop compressor performance, and the most accurate one predicted the flow range extension with less than 1.5% error.