[vertnik]

Hy!

I have problems with Launder and Sharma (LS) model when calculating backward-facing step (BFS) problem.

I am not using CFD software, but the whole model is coded in Fortran. With LS model I already solved 2D channel flow at different Reynolds numbers (low and high) and everything seems to work fine (when comparing with experiments and DNS results).

For BFS, the geometry and flow conditions are taken from experiment by
Jović and Driver, where the Re, based on the step height, is 5000.
The calculated reattachment position is agreed very well with the
experiment. But velocities near the bottom wall are wrong, and also the wall shear stress. Here is the link of the picture with wall shear stress along the bottom wall:
"http://www.p-ng.si/~vertnik/wall shear stress.jpg".
If there is someone, or was having similar problems, to help me out
finding the error?

Similar BFS problem was calculated for laminar flow at Re=800. The wall shear stress is perfect when compared with other authors.

So there must be problem with turbulence model? Can someone explain me, how this extra source terms in LS turbulent model, D in k-equation and E in e-equation, are calculated?


Thank for you answers,

Robert

[SKK]

Launder-Sharma model is best suited for flat plate zero pressure gradient flow. For backward facing step use other models such as k-w or SST. Following paper is a good reference,

http://ghajar.ceat.okstate.edu/pub_p...IJCFD_2005.pdf