Reynolds number effects in low-pressure compressors

Abstract

The performance of a compressor measured in flight differs from that measured on the ground. One of the main reasons for these differences is associated to the inlet Reynolds number variation. In view of the fact that in-flight testing involves a risk to the equipment, the flight instrumentation is limited to what is strictly necessary to determine compressor performance while the equipment for ground engine tests is extensive. Given the great abundance of data on the ground, the CFD calculations are therefore calibrated on it. The aim of this work is therefore to identify and understand these Reynolds effects, then to be able to quantify them so that corrections on data measured on the ground (at high Reynolds) can be applied to predict the compressor performance in flight (at low Reynolds).

The understanding of Reynolds effects is approached step by step. As a first step, the Reynolds number effects on performance of a two-dimensional compressor cascade are investigated. It results that the variation in performance due to Reynolds number is closely linked to the boundary layer development and the transitional phenomenon. Therefore, empirical methods are applied on published data to attempt to quantify these effects. As a second step, a description of the three-dimensional flows that are encountered in compressor cascade and their influence on cascade sensitivity to Reynolds number variation are included. In the last step, the Reynolds number effects in multistage axial-flow compressor are physically described on compressor characteristic operating at design rotation speed. The variation in performance when the Reynolds number changes might be interpreted in terms of viscous losses, aerodynamic blockage level and mismatching within the compressor. Finally, two Reynolds number corrections available in the literature are physically described with the effects previously identified. To illustrate how they are used, they are applied to experimental data and the results encourage the reuse of these corrections on other data sets.