An experimental and computational investigation of the flow in an engine representative diffuser system has been performed. Many parametric changes to the system were considered including inlet conditions, pre-diffuser geometry, cowl geometry and pre-diffuser strut position. The focus of this paper is an overview of the results from the twelve configurations that were both predicted and experimentally tested.
It was shown that the CFD predictions contained a high degree of numerical error and that to reduce this to a low level would be currently impractical for a parametric study of this type. However, by carefully maintaining approximately the same degree of numerical error in the predictions, it was shown that a valid parametric study could be performed to the extent that the same conclusions about the parametric changes can be drawn equally from the predictions or measurements with one small exception. It was recognised that the level of agreement between the predicted and measured losses was to a degree fortuitous and that the k-ε turbulence model performs poorly for this type of flow.
