INVESTIGATIONS OF THREE-DIMENSIONAL FLOW FIELD DEVELOPMENT IN AN AXIAL COMPRESSOR CASCADE

This paper presents a complete study about three-dimensional (3-D) flow field development in a linear compressor cascade where flow field in the blade passage has been studied experimentally as well as numerically. In the experimental work, a linear compressor cascade test section was installed in an open loop wind tunnel. The experimental data was acquired for a Reynolds number of 2.98x105 based on the blade chord and the inlet flow conditions. The flow field characteristics in blade passage including 3-D flow velocity and velocity magnitude have been measured by using calibrated five and seven-hole pressure probes connected to ATX sensor module data acquisition system (DAQ). To investigate flow development in the blade passage, velocity coefficient through streamwise planes has been calculated from the measured data. The computational fluid dynamics (CFD) study of the flow field was performed to gain a better understanding of the flow features. Present computational study was first validated with previous experimental and numerical work to check mesh accuracy and give confidence for computational results. Then, two turbulence models, Spalart-Allmaras (S-A) and shear stress transport SST (k-omega) were used for the present work. From both parts of study, the flow field development through the cascade have been investigated and compared. Moreover, the received data demonstrated a good agreement between the experimental and computational results. The predicted flow streamlines by numerical calculations showed regions characterized by flow separation and recirculation zones such as corner separation that could be used to enhance the understanding of the loss mechanism in compressor cascades. All measurements taken by the two probes, 5 and 7-hole pressure probes, have been analyzed and compared. The 5-hole pressure probe measurements have showed more agreements with computational results than 7-hole probe. Furthermore S-A turbulence model calculations showed more consistencies with experimental results than SST (k-omega) model.