The effects of inlet temperature and turbulence characteristics on the flow development inside a gas turbine exhaust diffuser

The overall industrial gas turbine efficiency is known to be influenced by the pressure recovery in the exhaust system. The design and, subsequently, the performance of an industrial gas turbine exhaust diffuser largely depend on its inflow conditions dictated by the turbine last stage exit flow state and the restraints of the diffuser internal geometry. Recent advances in Computational Fluid Dynamics (CFD) tools and the availability of computer hardware at an affordable cost made the virtual tool a very attractive one for the analysis of fluid flow through devices like a diffuser. In this backdrop, CFD analyses of a typical industrial gas turbine hybrid exhaust diffuser, consisting of an annular diffuser followed by a conical portion, have been carried out with the purpose of improving the performance of these thermal devices using an open-source CFD code "OpenFOAM". The first phase in the research involved the validation of the CFD approach using OpenFOAM by comparing CFD results against published benchmark experimental data. The numerical results closely captured the flow reversal and the separated boundary layer at the shroud wall where a steep velocity gradient has been observed. The standard k --epsilon turbulence model slightly over-predicted the mean velocity profile in the casing boundary layer while slightly under-predicted it in the reversed flow region. A reliable prediction of flow characteristics in this region is very important as the presence of the annular diffuser inclined wall has the most dominant effect on the downstream flow development. The core flow region and the presence of the hub wall have only a minor influence as reported by earlier experimental studies. Additional simulations were carried out in the second phase to test the veracity of other turbulence models; these include RNG k--epsilon, the SST k--o, and the Spalart-Allmaras turbulence models. It was found that a high resolution case with 47.5 million cells using the SST k--o turbulence model produced a mean flow velocity profile at the middle of the annular diffuser portion that had the best overall match with the experiment. The RNG k --epsilon, however, better predicted the diffuser performance along the exhaust diffuser length by means of the pressure recovery coefficient. These results were obtained using uniform inflow conditions and steady-state simulations. As such, the last phase of our investigations involved varying the inflow parameters like the turbulence intensity, the inlet flow temperature, and the flow angularity, which constitute important characteristics of the turbine blade wake, to investigate their impact on the diffuser design and performance. These isothermal CFD simulations revealed that by changing the flow temperature from 15 to 427°C, the pressure recovery coefficient significantly increased. However, it has been shown that the increase of temperature had no effects on the size of the reversed flow region and the thickness of the separated casing boundary layer, although the flow appears to be more turbulent. Furthermore, it has been established that an optimum turbulence intensity of about 4% produced comparable diffuser performance as the experiment. We also found that a velocity angle of about 2.5° at the last turbine stage will ensure a better exhaust diffuser performance.