Simulating laminar-turbulent transition with a low reynolds number k - epsilon turbulence model in a navier-stokes flow solver

摘要

A low Reynolds number (LRN) k -epsilon turbulence model has been implemented in a 3D Navier-Stokes flow solver such that laminar-turbulent boundary layer transition can be simulated. This was done in an attempt to improve the loss prediction of the code which previously used wall functions and assumed fully turbulent flow, leading to overprediction of the loss for some turbomachinery blade rows containing significant regions of transitional flow. The paper detail the problems associated with implementing a LRN k - epsilon model in an explicit Navier-Stokes flow solver and presents results from testing the model on zero pressure gradient flat plate flow at various freestream turbulence intensities and length scales. The results are compared to experimental data and a 2D boundary layer code which utilizes the same turbulence model. The addition of the LRN turbulence model improved the loss prediction capability of the Navier-Stokes code, and the chosen length scale had a large effect on the predicted loss. The sensitivity to the near wall grid, convergence and stability problems, and excessive run times make the LRN k - epsilon model in a Navier-Stokes code impractical for design applications at this time.