Time-resolved measurements with fast-response probes and laser Doppler velocimetry at the impeller exit of a centrifugal compressor: a comparison of two measurement techniques

Abstract

The goal of this work was the investigation of the unsteady flow field existing at the exit of the impeller of a centrifugal compressor and the critical comparison of velocity measurements with two systems based on totally different measurement principles. The paper focuses on the measurement concepts, the accuracy and the error quantification. A one-sensor fast-response probe (FRAP® system) and a commercially available laser Doppler velocimetry (LDV) system were used to perform wall-to-wall traverses and to measure the two main time-resolved velocity components. The test rig was run at the best point of the operating line, at a tip Mach number of 0.75. The mean levels and fluctuations of the velocity vector components were compared. Over the middle, 70 per cent of the channel width, the FRAP® and LDV measurements typically agree within 2–5 per cent of the momentary local flow velocity in terms of both the radial and the circumferential velocity components. The time-mean distributions agree within typically 0–4 per cent. Attention was focused on the near-hub and near-shroud layers where both measurements are affected by wall proximity errors. Largest discrepancies (of 10 per cent) in the time-mean velocity occurred at the shroud where velocity component values could differ by as much as 20 per cent momentarily. These discrepancies should be viewed in relation to the peak-to-peak amplitude of the ensemble-averaged velocity fluctuations which was of the order of 25 per cent in the present compressor.