Distortion reconstruction in S-ducts from wall static pressure measurements

Abstract

Purpose This paper aims to report the attempts for predicting “on-the-fly” flow distortion in the engine entrance plane of a highly curved S-duct from wall static pressure measurements. Such a technology would be indispensable to trigger active flow control devices to mitigate the intense flow separations which occur in specific flight conditions. Design/methodology/approach Evaluation of different reconstruction algorithms is performed on the basis of data extracted from a Zonal Detached Eddy Simulation (ZDES) of a well-documented S-Duct (Garnier et al., AIAA J., 2015). Contrary to RANS methods, such a hybrid approach makes unsteady distortions available, which are necessary information for reconstruction algorithm assessment. Findings The best reconstruction accuracy is obtained with the artificial neural network (ANN) but the improvement compared to the classical linear stochastic estimation (LSE) is minor. The different inlet distortion coefficients are not reconstructed with the same accuracy. KA2 coefficient is finally identified as the more suited for activation of the control device. Originality/value LSE and its second-order variant (quadratic stochastic estimation [QSE]) are applied for reconstructing instantaneous stagnation pressure in the flow field. The potential improvement of an algorithm based on an ANN is also evaluated. The statistical link between the wall sensors and 40-Kulite rake sensors are carefully discussed and the accuracy of the reconstruction of the most used distortion coefficients (DC60, RDI, CDI and KA2) is quantified for each estimation technique.