Different turbulence closures were used to predict the flow interaction between the wakes created by compressor outlet guide vanes (OGVs) and a downstream annular pre-diffuser. Two statistical turbulence models were tested based on the classical Reynolds-averaged Navier-Stokes (RANS) approach. Both high- Re and low- Re (Launder-Sharma) versions of the k-ε model were applied to a selected test problem for OGV wake/diffuser flows. The test problem was specifically chosen because experimentally determined inlet conditions and both profile and performance data were available to validate predictions. A preliminary study was also reported of the more advanced large eddy simulation (LES) approach. The LES sub-grid-scale (SGS) model was the basic Smagorinsky eddy viscosity assumption

with a Van-Driest damping function for improved capture of near-wall viscous behaviour. Comparison between the two RANS models showed little difference in terms of velocity contours at OGV trailing edge and diffuser exit. In terms of overall diffuser performance (static pressure recovery and total pressure loss coefficients)

the high- Re model was shown to agree well with experimental data. The preliminary LES study indicates the highly unsteady character of the OGV wake flow

but requires improved treatment of inlet conditions.