Hi,

I made a "simple" steady state simulation with the standard k-Epsilon model and after that exactly the same with the SST model. The results with k-Epsilon were normal and the SST results are similar except of the "Wall Heat Transfer Coefficients" which are 10 times bigger than those of the k-Epsilon model and are of the same quantity all over the domain (which is in the k-Epsilon simulation widely spread...). Convergence is very good so I dont know what the problem is... Maybe some can help! Thanks a lot!

marco

Hi,

In this case, please check your y+ values at walls, when using SST. They're should be <= 1. In addition, you might need about 10-20 nodes inside the boundary layer.

Re-submit the case and post here.

When using k-eps, y+ should be higher than 30.

Good luck!

cfdguy

Hi,

thank You for Your reply. Yplus is in the range 0.5 - 2... not so fine mesh but quite slow air flow. Is Yplus important when using "automatic near-wall treatment"?? (Of course finer mesh does better results...) To me it seems more a problem of calculation of the "Wall Heat Transfer Values". All other quantities (flow, temperature, wall adjacent temp, heat flux..) look similar to those of the k-epsilon-simulation. btw how is "wall heat transfer coefficient" calculated? Is it evaluated from the heat flux or is heat flux evaluated from the coefficient?

Thanks again for Your replies!

Marco

Hi,

Make it fine near the walls. In the bulk region, you can leave as a big mesh size, if that region is not important.

The HTC is calculated based on the wall temperature (that you specify, or the solver calculates) and the adjacent node.

"Is Yplus important when using automatic near-wall treatment"? I would say that y+ is important in all turbulent flows using RANS. They are even more important when you use a low-Re model, such as SST.

Cheers!