[ZviH]

Hello,

I'm trying to validate an LES code using DNS data by Moser for the fully-turbulent channel case.

I have seen some posts about it here before, but I can't find how the pressure gradient is determined to reproduce the DNS Re_tau values. This is not specified anywhere as far as I know and I'd like to keep the trail-and-error steps to a minimum...

For example, the DNS data shows that Re_Tau of ~550 for Lref and Uref of 1 is achieved with a bulk Re of 10,000 - but no pressure is prescribed.

Just to point out something:
I've seen posts like this, where the pressure gradient is modified on-the-fly (although I'm not clear exactly how to use the difference of mean flow rates) - however shouldn't the flow be driven by a constant gradient?
If I do need to get the pressure gradient corrected - I need to base that on the mean-velocity given by the DNS data, where the mean velocity in the LES case take a while to converge, won't it cause stability issues in the beginning (rapid change of the pressure gradient)?

Help?
Zvi.

[sbaffini]

Long story short:

1) Most DNS use a fixed mass flow rate (instead of pressure gradient), that' s why the pressure gradient is not specified.

2) If your channel has height H=2, rho=1, mu=1/Re_tau then dp/dx=-1 will work for any given value of Re_tau. Note that this is only true in DNS, where the exact channel flow equations (should) hold.
LES is almost there (depending on the resolution, SGS model, etc.) and you can double-check by actually computing u_tau from your mean velocity profile. This should be 1 with these parameters and no errors (i.e., DNS).

[FMDenaro]

I am quite sure this issue has been already discussed in other posts in cfd-online forum....