Effects of Impeller-Diffuser Interaction on Centrifugal Compressor Performance

This research program focuses on characterizing the effect of impeller-diffuser interactions in a centrifugal compressor stage on its performance using unsteady threedimensional Reynolds-averaged Navier-Stokes simulations. The computed results show that the interaction between the downstream diffuser pressure field and the impeller tip clearance flow can account for performance changes in the impeller. The magnitude of performance change due to this interaction was examined for an impeller with varying tip clearance followed by a vaned or vaneless diffuser. The impact of unsteady impeller-diffuser interaction, primarily through the impeller tip clearance flow, is reflected through a time-averaged change in impeller loss, blockage and slip. The results show that there exists a tip clearance where the beneficial effect of the impeller-diffuser interaction on the impeller performance is at a maximum. A flow feature that consists of tip flow back leakage was shown to occur at design speed for the centrifugal compressor stage. This flow phenomenon is described as tip flow that originates in one passage, flows downstream of the impeller trailing edge and then returns to upstream of the impeller trailing edge of a neighboring passage. Such a flow feature is a source of loss in the impeller. A hypothesis is put forth to show that changing the diffuser vane count and changing impeller-diffuser gap has an analogous effect on the impeller performance. The centrifugal compressor stage was analyzed using diffusers of different vane counts, producing an impeller performance trend similar to that when the impeller-diffuser gap was varied, thus supporting the hypothesis made. This has the implication that the effect impeller performance associated with changing the impeller-diffuser gap and changing diffuser vane count can be described by the non-dimensional ratio of impeller-diffuser gap to diffuser vane pitch. A procedure is proposed and developed for isolating impeller passage blockage change without the need to define the region of blockage generation (which may incur a certain degree of arbitrariness). This method has been assessed for its applicability and utility.