Effects of Inlet Flow Field Conditions on the Performance of Centrifugal Compressor Diffusers: Part 1—Discrete-Passage Diffuser

Abstract

This is Part 1 of a two-part paper considering the performance of radial diffusers for use in a high-performance centrifugal compressor. Part 1 reports on discrete-passage diffusers (shown in Fig. 1) while Part 2 describes a test of a straight-channel diffuser designed for equivalent duty. Two builds of discrete-passage diffuser were tested, with 30 and 38 separate passages. Both the 30 and 38 passage diffusers investigated showed comparable range of unstalled operation and similar level of overall diffuser pressure recovery. The paper concentrates on the influence of inlet flow conditions on the pressure recovery and operating range of radial diffusers for centrifugal compressor stages. The flow conditions examined include diffuser inlet Mach number, flow angle, blockage, and axial flow nonuniformity. The investigation was carried out in a specially built test facility, designed to provide a controlled inlet flow field to the test diffusers. The facility can provide a wide range of diffuser inlet velocity profile distortion and skew with Mach numbers up to unity and flow angles of 63 to 75 deg from the radial direction. The consequences of different averaging methods for the inlet total pressure distributions, which are needed in the definition of diffuser pressure recovery coefficient for nonuniform diffuser inlet conditions, were also assessed. The overall diffuser pressure recovery coefficient, based on suitably averaged inlet total pressure, was found to correlate well with the momentum-averaged flow angle into the diffuser. Furthermore, the pressure recovery coefficient was found to be essentially independent of the axial distortion at diffuser inlet, and the Mach number, over the wide flow range (from maximum flow to the beginning of flow instabilities) investigated. It is thus shown that the generally accepted sensitivity of diffuser pressure recovery performance to inlet flow distortion and boundary layer blockage can be largely attributed to inappropriate quantification of the average dynamic pressure at diffuser inlet. Use of an inlet dynamic pressure based on availability or mass-averaging in combination with definition of inlet flow angle based on mass average of the radial and tangential velocity at diffuser inlet removes this sensitivity. [S0889-504X(00)00101-X]