open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I'm modeling an estuary and I'm comparing the model against CTD data. The problem is that the model is too diffusive and there is almost no stratification, see attached figure.
I already tried to modify different parameters, such as:
1.- Different turbulence models. The best was K-E which is the one that now I'm using.
2.- Coefficient for horizontal (and vertical) diffusion: I tried both increasing and decreasing the values. However, no better result was achieved.
2.- Different schemes for advection of K-E were tried. The best one was the explicit scheme + MURD PSI scheme.
3.- Based on comments from the forum, I read that numerical diffusion is mainly related to the advection of the velocities. Thus I tried a different scheme for the advection of velocities. The one from the results attached corresponds to the Explicit scheme + MURD.
4.- Also based on the forum comments, I tried to use a different boundary value for the salinity, but the results were not improved.

Does anyone have any advice on how to increase the stratification? I've also attached the steering file.
Thanks

Hello,

It is very difficult to give an advice about vertical mixing - it is mostly an art of trial and error :) In my case of modelling small coastal estuaries I use a Smagorinsky model in horizontal direction and one dimensional k-eps GOTM model in vertical direction. It also seems to be helpful to use a hybrid vertical coordinate: several sigma layers over z-layers. It works for me.

Good luck,

Hello,

You seem to use an old release. From December 1st 2023, the latest release is v8p5r0. My first advice is to use the latest release of TELEMAC-3D. Many new features and bug fixes have been available for years and the release you seem to use.

I would add a few comments:
- if using the k-epsilon turbulence model, do not try to tune any COEFFICIENT FOR... DIFFUSION OF... (velocities or tracers),
- you can try the LIPS advection scheme which is the recommended in particular if using tracers which is less diffusive in 3D (SCHEME FOR ADVECTION OF... = 5 + SCHEME OPTION FOR ADVECTION OF... = 4),
- for velocities or k-epsilon, you can try the method of caracteristics (= 1) if LIPS or PSI are not OK,
- increase the number of horizontal levels,
- if you have stratification, as Sokolov has suggested, you can use z-layers or mixed z-layers and sigma layers, in particular refine where you have gradients over the vertical.

Hope this helps,

Chi-Tuan

Hi.

Thank you both for your answers.
pham wrote:
Hi,

After several months of troubleshooting the installation of V8P4, I'm still not able to have that version running in parallel mode. Only works on serial mode but took just too long to be useful.


About your comments:
- I'll try LIPS since I've tried already with PSI, and method of characteristics.
- What should I do to use the mixed z-layers and sigma layers? I know about their existence but I don't know how to use them in this particular case.

Thanks!

Hello,

Which version do you use? If it is too old, you may have bad results with LIPS as there were bug fixes a few years ago.

Use z-layers planes where you expect to get stratification and refine locally around this expected stratification. You can then use sigma layers above the last z layer.
Do not hesitate to read the TELEMAC-3D user manual (the latest one, not an old one).

Hope this helps,

Chi-Tuan