open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I'm running a very simple 3D model which is a flume 2000m long and 180m wide. It has 10 plains (9 layers) in vertical but with a very thin layer at the bottom (2mm). As I showed before, the k and epsilon profiles are not correct near the bottom but the velocity seems ok.





Then we calculated the analytical solution of the velocity and we found that the velocity at the bottom was underestimate and this might cause the overestimate of k and the bottom.



How does Telemac-3D calculate the velocity at the bottom?
Thanks!

Best regards,
Qilong

Hello,

The important point at the bottom is to compute the friction velocity, it is done assuming a logarithmic law, and then the friction term is deduced. You can find this in subroutine TFOND in Telemac-3D. In version 6.2 we have changed the implementation and introduced a variant, to use logarithmic derivatives, which gives better results, so I would suggest that you wait for the next release (end of the month) and try again.

With best regards,

Jean-Michel Hervouet

Tahnks Jean-Michel.

Best regards,
Qilong

Hello,

I checked the subroutine TFOND and I found:

LOGARITHMIC LAW FOR COMPONENT ON THE BOTTOM: NU*DU/DN = AUBORF*U + BUBORF

I guess NU is the viscosity and DU/DN is the velocity gradient.
Could you please tell me some details about this because I can't link it with the log law?

Hello,

Yes, NU*DU/DN is the stress to apply on the bottom, but above we compute UETCAR, which is U* squared, the friction velocity squared, and you see the logarithmic law in the case with Nikuradse friction. In other cases this is not useful as the 2D theory of Strickler and Chézy coefficients readily gives the stress, regardless of the velocity profile. So in the Nikuradse case, the velocity profile will give the U* and it will go in the stress.

Regards,

JMH

Tahnks Jean-Michel.

And what are the meannings of AUBORF and BUBORF? AUBORF is the function of UETCAR and VNORM but what does the BUBORF come from?

Best regards,
Qilong

Hello,

AUBOR and BUBOR is our general form of stress, including an implicit term AUBOR and an explicit term BUBOR. For friction we keep only the implicit term to avoid the fact that a friction that would be too large would reverse a velocity. With an infinite implicit term the velocity will be only cancelled.

With best regards,

Jean-Michel Hervouet

Thanks for your explaination.

I tried to use another value of AUBOR to correct the velocity at the bottom but I failed. I used the theoretical value of bottom velocity as the VNORM and then I used the same equation AUBOR(N)= - UETCAR(N)/MAX(1.D-4,VNORM).



I got the bottom velocity which is very close to the analytical solution but the whole profile has a slightly shift to the right. So I will leave this problem there and wait for the new release and I'll test it again with the new version.

Best regards,
Qilong

Hello, Qilong:

I'm modeling with Telemac-3D and when I check the vertical velocity profile, I find that velocities on the 1st layer (bottom) are not zero. But in your case, velocity vertical profile is quite OK. So I'm wondering where I'm doing wrong ?

thanks a lot.

with best regards.

Hello Robinson,

The last point towards the bottom is not the real bottom node. You expect to have another horizontal plane beneath on which all the nodes have zero velocity (theoretically). That's the real bottom. The thickness of this real bottom layer is determined by Telemac3D. If I remember correctlly, the thickness of this real bottom layer is half of the layer thickness just above it.

For my case, I used a very thin layer (4mm) near the bottom. so the velocity profile I got starts from a very small value near the bottom.

Kind regards,
Qilong