Advanced turbulence modeling for industrial flows.

摘要

This dissertation deals with the development of an improved two-equation turbulence model and its application to various flows. Six different conventional turbulence models were initially tested for predicting the flow inside of an unbaffled stirred tank reactor (STR), and the results are compared with experimental LDV data. Each of the models use low Reynolds number corrections. Results indicate that the radial velocity component in the impeller discharge region is overpredicted by each of the models. The tangential velocity component in the impeller discharge region is predicted well by the models, but is underpredicted near the shaft. The low Reynolds number k- 3 model is the only model which produces reasonable kinetic energy predictions in the impeller discharge region. The model predictions are generally unsatisfactory and produce varying results, which are largely attributed to the difference in the formulation of the low Reynolds number corrections for each model.; Based on these results a new model has been developed which eliminates the need for low Reynolds number corrections and has been demonstrated to produce improved results for various flows compared with the conventional models. The new model, called the v2f-kw model, is developed based on the elliptic relaxation approach of Durbin and is a variant of Durbin's v2f-k3 model. The new model is shown to be superior to all other models on a number of benchmark problems including the backstep, two-dimensional cavity, coaxial jet and jet in a crossflow. The new model is therefore proposed as a superior alternative from both computational effectiveness and accuracy viewpoints.