Title: A Unified Framework for Compressor Stall Inception 

Authors:
Sam Grimshaw Whittle Laboratory, University of Cambridge
Graham Pullan Whittle Laboratory, University of Cambridge
Edward Greitzer Department of Aeronautics and Astronautics, Massachusetts Institute of Technology
Zoltan Spakovszky Department of Aeronautics and Astronautics, Massachusetts Institute of Technology

Abstract: This paper presents a first-of-a-kind description of compressor rotating stall inception that explains the two different routes to stall reported in the literature over the past four decades. A low-order dynamic system model, an actuator disk analysis, is shown to capture these two transient, stall inception behaviors, namely modes: the growth of small amplitude, long length scale (on order of compressor circumference), sinusoidal perturbations, versus spikes: short length scale (on order of blade pitch) velocity disturbances with fast, non-linear growth. The analysis demonstrates that the two distinctly different processes are on a continuum of compressor dynamic behaviour and can be captured by the same unifying framework. Their features, including the different length and time scales, are revealed without explicit modeling of the blades.

The differentiator for stall inception behaviour is shown to be the slope of the compressor pressure rise characteristic at flows below that corresponding to the peak pressure rise. Small (defined in the paper) positive slopes to the left of the peak lead to growth of modes into fully developed rotating stall cells. Large (defined in the paper) positive slopes, lead to a more rapid evolution of spikes into stall cells. The actuator disk simulations indicate these two types of transient fluid motions bound a range of stall inception behaviors that depend on compressor pressure rise characteristic slope.

Data from compressor experiments and full annulus unsteady Reynolds-averaged Navier-Stokes simulations are used to connect the shapes of radially varying compressor characteristics, in the rotating stall regime, to flow features including blade leading edge and corner separations. The experiments and full annulus computations confirm the results derived from the low-order model and establish that it successfully captures the transient flows and physical mechanisms of rotating stall inception.