Hello, everyone. I'm now validating the Reynolds Stress Turbulence Model with experimental values of 2-D Channel. I got similar results in the velocity profile comparison to the experiment. But, in the wall-law, viscous sublayer is not properly predicted. This means that the range of the linear part(u+ = y+) is so small such as from 1 to 4. Typically, this linear part expands from 1 to nearly 8. I need some suggestions and advises to fix this one.

cf) 1. The y+ of the first grid from the wall is equal to near 1. 2. The first-order upwind scheme is used for inviscid flux terms. 3. The wall damping function(Not wall function) is used in the pressure-strain correlation. 4. The reynolds stress term is added by the source term in RANS equations.

Thank you in advance.

Is this a low Reynolds number RST model? Is the test case fully developed channel flow at what Reynolds number?

(1) y+ < 1 is more usual but y+ = 1 is likely to be OK.

(2) Not relevant for fully developed flow because zero in the cross stream direction and constant in streamwise direction.

(3) There are several damping functions. Changing the type or the exponent to get a better match is quite a common trick.

(4) Not relevant to the answer you get unless instability prevents this.

Generally, RST models have lots of parameters that can be tuned to improve the fit to a particular experiment. Unfortunately, problems tend to come when one tries to find a general set of parameters.

If you are not getting the expected behaviour it may be wise to compare against the original predictions of the author of your particular RST model before comparing against experiments. If the original predictions show the same behaviour then you can have confidence in your implementation and concentrate on modifying the parameters to fit your purposes.

Thanks andy for your comments!!! I will try it. ^^