Turbulence models for the numerical prediction of transitional flows with RANSE

Thesis: S.M. in Naval Architecture and Marine Engineering; and S.M. in Mechanical Engineering , Massachusetts Institute of Technology, Department of Mechanical Engineering 2015Cataloged from PDF version of thesis.Includes bibliographical references (pages 111-112).Research on turbulence modeling in naval architecture has extensively increased in importance over the years and it is now considered one of the most important ways to accurately compute high Reynolds number flows with Reynolds Averaged Navier-Stokes Equations (RANSE) solvers. In naval architecture, turbulence models are necessary to solve typical hydrodynamic problems both in model scale, where Re=O(106), and in full scale, typically Re=O(108), since direct numerical simulations are not possible in these cases. This thesis aims to study the performance of different turbulence models to predict the laminar-turbulent transitional flow in the boundary layer of streamlined bodies. Starting with a systematic study on a flat plate and arriving to transitional flow airfoils like the NACA 651-213 a=0.5. The RANSE solver is built on the libraries of OpenFOAM(Open Field Operation and Manipulation) which is a free, open source CFD program which enables a large group of users to solve broad range of problems. Turbulence models considered range from one equation models such Spalart-Allmaras, two-equation models such as k-epsilon, k-omegaSST, three-equation model kkl-omega as RANS solvers, LES solvers and DES Solvers. The validation of OpenFoam based solver and the different turbulence models is made on the prediction of the friction and pressure drag components as well as lift predictions. In particular, the capability of the turbulent models to capture the transition between laminar and turbulent regime plays a vital role in engineering applications. Four different turbulence models are used in this scope: k-epsilon, k-omegaSST, Spalart- Allmaras and kkl-omega in conjunction with different wall functions. The flat plate case was simulated with all of these turbulence models by using the pimpleFoam transient solver and the hydrofoil case was tested with the kkl-omega and kOmegaSST models by using simpleFoam steady-state solver. The kkl-omega t.m. is one of the newest transition models and it was developed to superior to the other models since it provides the transition region information. Its current implementation in OpenFOAM significantly underestimates the skin friction and the onset of the transition point. We propose a series of modifications which we implemented on model equations and empirical parameters. These changes improve the prediction accuracy of the frictional drag component in transitional flows.by Mert Gokdepe.S.M