The selection of a turbulence model for a problem is not trivial and has to be done systematically after comparison of various models with experimental data. It is a well known fact that there is no such turbulence model which fits all problems (, ). The flow in S-duct diffuser is a very complex one where both separation and secondary flow coexist. Previous work by the author on CFD analysis of S-duct diffuser was done using k-ε-Standard model , however it has been seen that choosing other turbulence model may result in better capturing of the physics in such a problem. Also flow control, to reduce energy losses, is achieved using a technique called Zero Net Mass Flow (ZNMF), in which suction and vortex generation jets (VGJ) are combined and positioned at optimum location. A proper turbulence model has to be chosen for capturing these phenomena effectively. Extensive experimental data is available on this problem and ZNMF technique from previous work done by one of the authors which is used for validating the CFD results. Here the focus is on choosing the best turbulence model for the S-duct diffuser.
Numerical (CFD) analysis is carried out using Ansys Fluent 13.0 with six turbulence models for the geometry with (ZNMF) and without (Bare duct) flow control and then compared with the experimental results. The turbulence models used are Spalart-Allmaras, three variants of k-ε – Standard, RNG and Realizable and two variants of k-ω – Standard and SST model. All the parameters of comparison are non-dimensionalized using the free stream properties, so that the results are applicable to a wider range of problems. This work is limited to incompressible flow analysis, as the experimental data is only available for low Mach number flows. Comparison of all these models clearly shows that results obtained using k-ω-SST model are very comparable to the experimental results for the bare duct (without flow control) and flow controlled duct both in terms of distribution of properties and aggregate results. Compressible flow analysis can be attempted to achieve reliable results in future with ZNMF using the best turbulence model based on this study.