open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi All,

We are doing some comparison work between the CGWAVE and ARTEMIS model..the objective being to move away from CGWAVE and use ARTEMIS in future projects.

A first step was to convert an existing CGWAVE grid to a selafin grid. CGWAVE uses semi-circular open boundary and we were planning to change that to linear segments as more commonly used in ARTEMIS (at least from what I have seen so far).
We still tried to run with semi-circular boundary just to see how it would look like and..a bit surpringly, this seemed to work pretty well..We used some existing fortran code (from examples) to specify the correct phases along the boundary and got nice , uniform crests along the boundary with correct timings etc...

Would be keen to have comments/feedbacks from more experienced users on that ? is this ok to use or is it maybe violating some assumptions ? I guess a semi circular boundary is really many linear segments so it sort of make sense that it works..

Any insights/comments greatly appreciated would be very much appreciated!

Hi again modellers !
Anyone with some lights on this ?

thanks

Hi,

I think you methodology is correct. As soon as you defined the phase correctly using fortran example, it should work.
The only limitation would be in the case you have a variable bathymetry on the boundary, as the wave number would change leading to problems in phase definition.

best regards,

Christophe

Thanks Christophe,
We do have a variable bathy along the boundary but I think this is taken care of within the fortran code (adapted from the BWTI example)...thanks again and more testing underway !

Hi all,

I'm working on this topic together with Simon, we got results using this methodology (see this illustration of the free surface)




Comparing it to the results from CG Waves though, the boundary conditions are computed very differently. In ARTEMIS the wave height and direction are the same along the whole boundary (as declared in the subroutine) while CG Waves uses a parabolic approximation (the reason the boundary has to be a (semi) circle). This seems to adjust the wave phase, wave height and wave direction along the boundary depending (as well) on the wave number. (See illustation)



Now this seems to be quite a bit more realistic (seems to correct for refraction as well). The below plot of the wave height along the intersection (the red line shown in the first illustration), shows that this difference in boundary conditions has a big influence on the results.




Does anyone have an idea how to get more realistic results in ARTEMIS? (the CG Waves model has been validated against physical results already)

Thanks a lot in advance!

(added a zipped folder as well, just to make sure you get the illustrations)

Hi,

In ARTEMIS you are free to impose the direction, phase and wave amplitude on each node of the boundary using the BORH subroutine (as you did for phase) : direction is TETAB and amplitude is HB.
Maybe I'm missing something but if you know what the other code imposes on the boundary (parabolic aprox.) then you should be able to impose it on the ARTEMIS boundary as well using ALFAP, TETAB and HB.

If you have an experimental reference and if you can share data, I can have a look to your case and maybe integrate it in the validation database (if results are better).

Regards,

Christophe

Cool. Thanks for that. The other model uses an approximation for the boundary conditions, but how it does that for every node is all in the 'black box' of the model (an other advantage of ARTEMIS) so not easy to impose these conditions on ARTEMIS. I reckon though for instance using TOMAWAC would be an option as well to get more realistic results.
I'm keen to send you the example once it's further developed so it can be used in the database. I'll let you know!

Hi,

I am a new user of ARTEMIS and also have a semi-circular domain (beach bounday is linear and ocean boundary is circular). My problem is that I am getting strong boundary effects (not realistic) near the ocean boundary (in part of it), when plotting the wave height output (for lower wave heights at incident wave boundary this problem is not verified). How can I eliminate/reduce those boundary effects? I presume that extending the domain might solve the problem, but I would prefere to apply other options.
Thanks,

Amy

Sorry missed that ..Do you have correct phases along your boundaries ?
How far/ deep your model extends ?
there are some fortran examples in the "examples" provided with the model.

Thanks for your reply. I was able to solve the problem by defining an adequate TETAP in BORH subroutine.
Regards,

Amy