Performance testing and detailed flow measurements were made in an axial compressor rotor with various tip clearances. The experiments were conducted on the condition of the same incidence angle at midspan. Thus, the effect of tip clearance distinguished from that of incidence angle was investigated on the overall performance, work-done factor, blockage factor, and increases in displacement, momentum, and blade-force-deficit thicknesses of the casing wall boundary layer, The phase-locked flow patterns obtained by the multisampling technique show clear evidence of a leakage vortex core behind the rotor. Behavior of the leakage vortex was clarified for various tip clearances by examining loci of the vortex center, decay characteristics of the vorticity at the center, and the total amount of vorticity shed from the blade tip. These results were compared with the leakage vortex model presented by Lakshminarayana.

Ensemble-averaged and phase-locked flow patterns in various tip clearances of two axial compressor rotors were obtained by aperiodic multisampling technique with a hot wire in the clearance and with a high-response pressure sensor on the casing wall. A leakage flow region distinct from a throughflow region exists at every clearance. In the case of a small tip clearance, the leakage jet flow interacts violently with the throughflow near the leading edge, and a rolling-up leakage vortex decays downstream. As the clearance increases, a stronger leakage vortex comes into existence at a more downstream location, and a reverse flow due to the vortex grows noticeably. A scraping vortex is recognized at the pressure side near the trailing edge only for the small clearance. A horseshoe vortex appears in the upstream half of the through flow region for every tip clearance. The solidity does not affect the flow pattern substantially except for the interaction of the leakage vortex with the adjacent blade and wake.

Evolution and structure of multiple stall cells with short-length-scale in an axial compressor rotor have been investigated experimentally. In a low-speed research compressor rotor tested, a short-length-scale stall cell appeared at first, but did not grow rapidly in size, unlike a so-called “spike-type stall inception” observed in many multistage compressors. Alternatively, the number of cells increased to a certain stable state (a mild stall state) under a fixed throttle condition. In the mild stall state the multiple stall cells, the size of which was on the same order of the inception cell (a few blade spacings), were rotating at 72 percent of rotor speed and at intervals of 4.8 blade spacings. With further throttling, a long-length-scale wave appeared overlapping the multiple short-length-scale waves, then developed to a deep stall state with a large cell. In order to capture the short-length-scale cells in the mild stall state, a so-called “double phase-locked averaging technique” has been developed, by which the flow field can be measured phase locked to both the rotor and the stall cell rotation. Then, time-dependent ensemble averages of the three-dimensional velocity components upstream and downstream of the rotor have been obtained with a slanted hot-wire, and the pressure distributions on the casing wall with high-response pressure transducers. By a physically plausible explanation for the experimental results, a model for the flow mechanism of the short-length-scale stall cell has been presented. The distinctive feature of the stall cell structure is on the separation vortex bubble with a leg traveling ahead of the rotor, with changing the blade in turn on which the vortex leg stands. [S0889-504X(00)00701-7]

The transient processes of rotating stall evolution have been investigated experimentally in a low-speed axial compressor stage with three stator-rotor gaps. The pressure traces at eight circumferential locations on the casing wall near the rotor leading edge have been analyzed by the wavelet transforms. With the appropriate mother wavelets, the evolution of short and long length-scale disturbances leading to the stall can be captured clearly. Behavior of these disturbances is different depending on the stator-rotor gap. For the large and middle gap, the stall inception is detected by a spiky short length-scale disturbance, and the number of spiky waves increases to generate the high frequency waves. They become the short length-scale part-span stall cells at the mild stall for the large gap, while they turn into a big stall cell with growth of a long length-scale disturbance for the middle gap. In the latter case, therefore, the stalling process was identified with “high-frequency stall inception.” For the small stator-rotor gap, the stalling process is identified with “long wavelength stall inception” and supported the recent computational model for the short wavelength stall inception by showing that closing the rotor-stator gaps suppressed the growth of short length-scale disturbances. From the measurement of the pressure field traces on the casing wall, a hypothesis has been developed that the short length-scale disturbance should result from a separation vortex from a blade surface to reduce circulation. The processes of the stall evolution are discussed on this hypothesis.

Statistical characteristics of pressure fluctuation on the casing wall of two axial flow compressor rotors have been investigated experimentally to find a precursor of rotating stall. Near stall, the casing wall pressure across a flow passage near the leading edge is characterized by a highly unsteady region where low-momentum fluid accumulates. The periodicity of the pressure fluctuation with blade spacing disappears and an alternative phenomenon comes into existence, which supports the disturbance propagating at a different speed from the rotor revolution. The precursor of rotating stall can be detected by monitoring collapse of the periodicity in the pressure fluctuation. To represent the periodicity qualitatively, a practical detection parameter has been proposed, which is easily obtained from signals of a single pressure sensor installed at an appropriate position on the casing wall during operation of a compressor.