open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear All,

I was wondering, if any one has done any kind of comparative study between Tele2D and Tele3D. I was concerned about the free-surface elevation.

I get very good results with Telemac-2D for free-surface variation at a validation point inside the domain. But when I use Telemac3D for the same case the z values (free-surface) at the same point in T3DRes the free-surface variation doesn't match at all.

I have tried bunch of things: switching on the non-hydrostatc version, changing the bottom roughness. I was wondering if any one can point me in the direction of tutorial/example in which both Tele2D/3D have been compared.

Thanks,Sumit

Hi

As far as I know, there is no specific study dedicated to the comparison of T2D and T3D. However, you can find in the validation document and/or in the validation tests the same test cases which have been tackled with T2D and T3D.
You can see, for instance, the cases called "gouttedo", "piledepon" or "Malpasset".
Eventuelly, you should remember that it is natural to find difference between 2D and 3D results; and depending of the behaviour of the flow and its physical characteristics, these results could be in the advantage of the 2D or the 3D code.

with my best regards
Riadh

Hello Riadh,

Thanks a lot for your kind reply. I do understand that both 2D and 3D model are suppose to give different results; the thing that confuses me is that even in 3D model (hydrostatic-version) one uses depth-avergae continiuity equation to solve for water surface elevation; hence the water surface elevation produced by both the model should be the same. Any ideas there???

Best regards,
Sumit

Dear Sumit

I'm not sure if we do resolve the averaged continuity equation in this case (I have to dive into detailed algorithms of T3D) but even if it the case, we resolve the same depth everaged continuity equation but the velocity field is not necessary the same in both cases. The vertical average of the velocity field (in 3D) will not necessary coincid with the averaged values obtained with a 2D Saint Venant system.
Am I wrong?

Kind regards

Dear Riadh,

I do agree with you in regards with the velocity. But, as I said the free-surfrace is computed by the same depth-average continuity equation. Hence that should be same in both the cases. I have been working on some what similar codes, Environmental Fluid Dynamics Code (EFDC), there we have the concept of mode-splitting which is again borrowed from POM (Princeton Ocean Model). In all these model the 3D non-hydrostatic version is solved in tandem with the depth-average form to get the water-surface elevation. I am going through the T3D algorithm to see if the same approach is followed. I was asking this on forum hoping some one has already read all this and can give me a quick answer.

Best regards,
Sumit

Hello,

Indeed, 3D (even hydrostatic) may not give you the same results as the 2D. This is in fact even more so if you use the LAW OF BOTTOM FRICTION = 5. If you wish to have a chance to compare calibrations between 2D and 3D you need to use friction laws that are also applied to the 2D velocity field (Chezy, Strickler ...).

There are also some options that are slightly differently implemented in 2D and 3D (such as the wave equation). This particular discrepency has been solved, by the way, and the correction will be in v6p2.

To answer your other question about non-hydrostatic, TELEMAC-3D does similarly to EFDC and POM and many others, in having a combined pressure term including the non-hydrostatic component. The method of projection is used to set a Poisson equation, which is used to solve for dynamic pressure.

I hope this helps further and complement what Riadh said.

Sébastien.

Hey Sebastien,

Thanks a lot for your kind reply. I was thinking on the similar lines I have always used the hydrostatic version of EFDC. The usage of poisson equation to solve for fully 3D NSE is probably same as used in FV schemes in open-FOAM, Fluenr and solvers like that.

Coming back to my initial question, if I only talk about water surface elevation, and used T3D (hydrostatic-version) and T2D, don't you think for the same boundary condition and same bottom friction I should get similar results for water-surface elevation. This is some thing that I am struggling with.

Once again your inputs are very much appreciated.

Best regards,
Sumit

Sorry I realise I may not have been very clear in my previous answer.

What I meant is that there are physical laws and numerical schemes that can make results different. However, I agree with you, in absolute / theoretical term TELEMAC-2D and TELEMAC-3D hydrostatic should give the same answer

Thus ...

In order to get the same results (having also talked to my collegue Alan J. Cooper), we would recommend first using the Chezy law for the comparison. This law is based on the depth-averaged flow so the bed friction is exactly the same in the 2D and 3D models if the depth averaged current is the same. Other friction laws will not work the same in 2D and 3D and will therefore produce different answers.

There is also a representation of bed slope in the friction law that applies in 2D. It is probably the same in 3D but we are not 100% sure -- Jean-Michel may know whether this also applies in 3D.

Also, remember to use the same diffusion, for example the Elder Law (Turbulence model 2) has no analogue in 3D. So use a constant viscosity value in both (horizontal turbulence model =1).

Further, if the results still differ make sure both models are adequately converged and increase the number of iterations or solver accuracy if necessary.

Finally, you should note that the 3D model uses wave formulation so you must compare the 2D model in wave formulation and not in primitive equation form.

... and (what I meant in my previous post about the correction) you also need to check that the implicitation values for depth and velocity are set the same in both models as they are not the same by default and make sure that the code (subroutine LECDON) does not change these.

Hopefully that will sort it out.

Sébastien

Hello Sebastien,

Thanks a lot; thats a good bit of leads to follow and hope for the best. I will definitely come back on this as I complete some simulations.

Once again thanks a bunch.

Best regards,
Sumit

Hello Sebastien,

Unfortunately I am still not having success in the comparative studty of tele2D and tele3D. I am attaching my steering file for both 2D and 3D case, can you kindly take a look and throw some more ideas towards my side.

Specifically my problem is that; I have a validation point inside the domain the fre surface elevation obatined by Tele2D matches perfectly with the observed data at this point. I have been successful with both manning's n and Chezy constant while using 2D model.

I want to reproduce the same result with Tele3D (hydrostatic-vesrion). In the 3D simulation I am examining the free-surface recorded in the 2D RESULT FILE obtained from the 3D simulation and was hoping that it will be same as one obtained by Tele-2D but not much of luck so far.

Please find my steering file attached for both the cases.

Best regards,
Sumit