The authors’ preceding analysis on centrifugal vaneless diffusers is used to examine the influences of diffuser geometries and of flow inlet conditions on the critical flow angle for reverse flow, and the results are presented in graphs. The diffuser width to radius ratio, the inlet Mach number, and the distortion of the inlet velocity distribution have significant influences on the critical flow angle, while the Reynolds number and the boundary layer thickness at the inlet have minor influences.

Based on the theory in Reference (3), the flow behavior is predicted in four vaneless diffusers with different geometries at the flow rates just before rotating stall and just before stall, which have been experimentally determined. Since the predicted velocity distributions at the critical conditions are physically reasonable for rotating stall and for stall are tentatively set so that the critical conditions are predictable using the theory. Prediction is made for two cases for backward leaning blade impellers and three cases for radial blade impellers in the literature in addition to the authors’ experiments. Satisfactory results are obtained except one example.

The limit of rotating stall was experimentally determined for three very small specific speed centrifugal blowers. The impellers were specially designed for stall-free at very small flow rates, so that the cause of rotating stall could be attributed to the vaneless diffusers. Experimental results demonstrated that the blowers did not stall until the flow coefficient was reduced to very small values, which had never been reported in the literature. The critical flow coefficient for rotating stall agreed very well with the prediction based on a flow analysis and a criterion for rotating stall in vaneless diffusers developed by the authors.

No abstract

An impeller of a supersonic centrifugal compressor was tested in a casing without a diffuser so that the flow range was not limited by the diffuser. Regarding the impeller, emphasis was placed on critical conditions such as inducer stall and surge. Experimental data were examined based on a one-dimensional analysis and a quasi-three-dimensional analysis. Furthermore, the variation of shroud pressure with respect to time at many locations was utilized to guess the details of flow behavior between impeller blades near the shroud, and the contour of isobars was compared with that predicted by a quasi-three-dimensional analysis. When the inlet relative velocity was supersonic, a detached shock wave and a shock wave in a blade channel were recognized, but the compressor operated efficiently, although such condition existed only in a narrow flow range limited by surge and choke.