Prediction Performance Of Several Turbulence Models In Flows Pertinent To Turbomachinery

The research of turbulence models and transition models is of essential importance in understanding, predicting and controlling flows and then in engineering designs. In many fields such as turbomachinery and aircraft research, much effort has been paid to turbulence models and transition models. However, the physical and numerical aspects of the prediction performance of turbulence models and transition models have not yet met the needs of numerical computation for complex flows. As a result, further research on these models is needed so as to combine with experimental measurement and theoretical analysis to gain more insight into flows and then to design advanced devices such as turbomachinery and aircraft.The present thesis studies the prediction performance of several turbulence models in flows pertinent to turbomachinery, the main work and its conclusions include:Constructing a research line base on a basic framework which consists of ten aspects including physical mechanism, mathematical model, boundary condition, initial condition, geometry grid, numerical method, program design, computation implementation, data processing and result analysis to lead the work of the present thesis.Developing a new classification of methods for turbulent and transitional flow prediction in order to unify different classifications in different documents; analyzing the work of various research groups in the research field of turbulence models and transition models in order to infer a perspective of the state-of-the-art research of turbulence models and transition models which is a category of methods for turbulent and transitional flow prediction.Computing three flows pertinent to turbomachinery in order to assess the prediction performance of four turbulence models. RANS equations and turbulence models equations are solved using four-order implicit scheme. The results of the computation demonstrate that: In turbulent flat-plate boundary layer flow, SA model, SST model, k ?ωmodel and q ?ωmodel can all predict mean streamwise velocity, Reynolds shear stress and other physical variables accurately. In turbulent backward-facing-step flow, the predictions of skin-friction coefficient and other physical variables by SA model, SST model and q ?ωmodel are not as accurate as those in the turbulent flat-plate boundary layer flow, while these models can give rather accurate predictions of reattachment length and other physical variables and small corner eddy and other physical phenomenon in turbulent backward-facing-step flow. In turbulent turbine rotor plane cascade flow, the prediction performance of SA model, SST model and q ?ωmodel further demonstrates the limitations and importance of turbulence models, that is: none of turbulence models can predict all complex flows accurately, and good turbulence models can improve the accuracy of flow predictions to a large extent.Developing a modification method for a turbulence model and verifying it in numerical computations of two flows. The results of the computation demonstrate that: the value of the closure constantβ1 of SST model has an impact on the prediction performance of SST model. The impact of changingβ1 from 0.075 to 0.05 is exactly the opposite of the impact of changingβ1 from 0.075 to0.1, 0.125 and 0.2 , and the impact of changingβ1 from 0.075 to 0.1 or to 0.125 or to 0.2 is the same. The improvement gained from changingβ1 from 0.075 to 0.2 is the most remarkable.