open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,

We are working on a case involving wave modelling in a lake using both Tomawac and Artemis, chained together. We are relatively new to those modules, especially Artemis. We use version 8.1.2.

We run two scenarios corresponding to two different lake surface elevations and wind speeds, both in Tomawac and Artemis (bathymetry is expressed relatively to a fictive water surface at 0 in all models). For one of the scenarios, the Artemis results are OK but for the second scenario, the results are not OK, showing what looks like boundary effects propagating into the mesh (see first image). What is strange is that the animation created for this very similar scenario is showing OK results (see second image). Does someone know what is happening? It is strange to get OK results in the animation but not in the result file itself?!

Are we missing something? I attach the steering and boundary file. Both scenarios use the same boundary file (.cli) and the same set-up.

Any help is greatly appreciated!
Thank you in advance!

Kind regards,
PL

1st image



2nd image

Hello PL,
Its hard to troubleshoot only with the information you provided. A legend and/or scale to the results would help. Also, can you post the "results as expected" model to compare with this one and see which parameters are causing the issue.
Best Regards,
Costas

Hello Costas, thanks for having a look.
We cannot share the model but I post below a couple of images to illustrate the difference between the frist scenario (V = 30 m/s, WAVE HEIGHT + Animation, both output seem to be relevant with respect to each other) and the second scenario (V = 20 m/s, WAVE HEIGHT + Animation, Animation looks reasonable w.r.t output from Scenario 1 but not w.r.t WAVE HEIGHT...).

The thing that is strange for us is why the Scenario 2 outputs are not relevant w.r.t each other.


Scenario 1 - WAVE HEIGHT



Scenario 1 - Animation



Scenario 2 - WAVE HEIGHT



Scenario 2 - Animation




Thanks in advance!

Kind regards
PL

Hello PL,
I think both your models exhibit similar behavior, but the patterns are more clear in the second plot. From the look of it, you have reflections from the northern and southern solid boundaries (assuming incident waves from the west. Reflection coefficients in the CLI file for solid boundaries (222) are set to 1.0 (7th column), so you should expect some, except reflection coefficients are otherwise (BOHR subroutine?).

I would advise to set reflection coefficients to 0.0 to begin with, so that ARTEMIS results would be comparable to TOMAWAC (diffraction apart). If you have some overlap, you could cross check.

Best regards,
Costas

Hello again,

Here comes a new cli-file (I am told that the one I posted in the first message is not the latest - sorry for this) together with an image showing the extent of each BC type.

Kind regards
PL

Hi PL,
If reflection patterns are not the concern, the next thing that needs investigation is spectrum discretization and definition. To do that, I would suggest to cancel TOMAWAC chaining and define ARTEMIS boundary conditions manually, as illustrated by the TOMAWAC 2D results. If the results improve, you 'll need to have a second look at the chaining setup.
As a general note for the computational mesh, make sure you have at least 7 nodes per wave length, as this tends to smoothen 'choppy' results. It creates more elements, but on the other side the solution converges faster.
Costas

Hi,

Thanks for your advices, we'll do some more checks.
Yes, I think we have around 10 nodes / wave length.

Kind regards
PL

Hi Costas,

I just wanted to update you regarding this topic.

We did some checks and indeed found one thing that influenced the results of the scenario we were not so happy with. The cause was than in that geometry, some nodes had a zero elevation (we modelled different water levels in our different scenarios). Assigning a minimum depth to 0.1 m solved the issue.

The final model now performs fine including the chaining with Tomawac.

Kind regards
PL

Hi PL,
There is a warning message that pops in the output when a minimum depth is reached (0.01m) that should raise concern. Clipping small depths has been standard procedure here for so long that I can't recall the particular reasons.
Nice to hear you got it sorted and thanks for the feedback.
Best Regards,
Costas