Axial compressor performance modelling with a quasi-one-dimensional approach

Abstract

This paper describes a technique for the numerical modelling of axial compressors using the simplest form of representation—a mean-line one-dimensional method. Although this type of flow analysis is a gross simplification of a complex three-dimensional system, which can now be modelled more accurately by many of today's computational fluid dynamics (CFD) packages, it does offer the advantages of simple input requirements and fast convergence times. These factors were major considerations when choosing a prediction method to be employed within an optimization program developed at Cranfield for the optimization of stator vane stagger angles, where detailed stage data were not required. The objective was to develop a technique capable of simply modelling the flow in multistage compressors to provide performance data including overall pressure ratio, efficiency and flow range at various speeds, as well as providing interblade data so that restaggering of blades could be accounted for. The salient issues presented here deal firstly with the construction of the compressor model and its incorporation into a FORTRAN program. A discussion of the results then follows, where four very different compressors have been modelled. Finally, the stability limit (or surge point) prediction methods are reviewed. Results have shown that consistent agreement can be obtained with such a performance prediction model, providing the loss and deviation correlations used are applicable to the blade profiles. It was also apparent that the two simplistic methods of surge prediction, which relied upon either stage or overall characteristic gradients, gave better agreement than the more rigorous discrete stage-to-stage model. This was thought to be a result of the continuous nature of the empirical rules utilized, which predict a smooth stage characteristic and hence falsely delay the estimated instability point.