Hybrid RANS/LES modelling of OGV/prediffuser flow

In the gas turbine engine, the OGV/prediffuser combination is key to achieve a good design for combustor external aerodynamics. Since the flow includes 3D turbulent wakes and boundary layers in adverse pressure gradients with the possibility of flow separation, the OGV/prediffuser combination offers significant turbulence modelling challenges for CFD. In order to understand the optimum approach for modelling turbulence in this important sub-component of compressor/combustor interaction, a comparison is reported in this thesis with available experimental data for both a conventional and an advanced OGV/prediffuser combination using (i) both high Re and low Re RANS CFD, (ii) LES CFD, and (iii) hybrid RANS/LES CFD. In the hybrid RANS/LES CFD, a new method based on the use of an Algebraic Stress Model and a modified Recycling and Rescaling method has been developed to generate a spatially and temporally correlated unsteady velocity field for the LES inlet conditions from the time-averaged RANS solution at OGV exit. The results show that: 1)Both high Re and low Re RANS solutions show good agreement with the experimental data for the OGV wake prediction, but high Re RANS provides better predictions of overall pressure loss and is certainly more cost effective considering computing costs. 2) The LES solution shows partial flow separation of the OGV suction side boundary layer prediction which was not noted in the experiment. This is probably caused by the presence of relaminarisation and subsequent transition of the suction side OGV boundary layer. This places high demands on the LES near wall mesh required, as well as providing an extreme challenge for the LES sub grid scale model. 3) The Hybrid RANS/LES approach is able to provide a good balance of predictive capability, matching RANS predictions on global performance (pressure rise/loss) and improving the prediction of velocity distribution at prediffuser exit, and it thus offers an optimum approach for OGV/prediffuser flow simulation considering both accuracy and cost.