I'm running periodic flow through the channel with one large amplitude wavy wall and the other - smooth. Using the enhanced wall treatment for both RANS and KE gives pretty different results for the pressure gradient (for the same mass flux). Any suggestions? BTW the same is true for even the smooth pipe flow. Thank you.

RANS == RSM ??

RSM.

Do you make a comparison with some empiric correlation for the smooth pipe flow pressure gradient? Re=? Y+=? Grid dependence? Do you sure that result must be the same for k-e and RSM for your flow?

For the wavy channel - the source is DNS and PIV+LDV; I do not expect RSM and k-e to behave the same way. For the smooth pipe - Re>3*10^4, I do expect both k-e and RSM to behave closely the same, and not within 15% as it is for now. Concerning the experiment/correlations for the smooth pipes, say, Blasius. BTW, grid dependence and Y+ are checked and are OK.

I was thinking that difference is mach more then 15%. I do not feel to discuss this 'fine' differences.

I have read that the values of the standard k-e model constants represent a compromise chosen (with subjective judgment) to give the 'best' performance for a range of flows. For any particular flow it is likely that the accuracy of the model calculations can be improved by adjusting the constants.

Maybe the standard values of the constants is not a 'best' values for the smooth pipe flow with Re=3e+4 ?