open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I'm coupling Telemac2d with Tomawac. The results of graphic printouts I observe are very strange :

For each model, the TIME STEP is the same and equal to 5.
I have tested few combinations of three parameters :
- PERIOD FOR GRAPHIC PRINTOUTS in Telemac
- COUPLING PERIOD FOR TOMAWAC in Telemac
- PERIOD FOR GRAPHIC PRINTOUTS in Tomawac

For each test, Telemac graphic printouts are those expected.

For Tomawac, I have graphic printouts only if COUPLING PERIOD FOR TOMAWAC=1 or PERIOD FOR GRAPHIC PRINTOUTS=1 in Tomawac. Otherwise I have only the first iteration (time=0).

COUPLING PERIOD FOR TOMAWAC=1 is really time consuming so I would like to work with PERIOD FOR GRAPHIC PRINTOUTS in Tomawac.

Moreover, if COUPLING PERIOD FOR TOMAWAC is different from 1 and PERIOD FOR GRAPHIC PRINTOUTS=1 in Tomawac the results are not convenient to milk. I give an example to be more understandable :
- TIME STEP = 5 in Telemac
- PERIOD FOR GRAPHIC PRINTOUTS = 20 in Telemac
- COUPLING PERIOD FOR TOMAWAC = 10 in Telemac
- TIME STEP = 5 in Tomawac
- PERIOD FOR GRAPHIC PRINTOUTS = 1 in Tomawac
I have results for 0, 5, 55, 105, 155,...,855 in Tomawac.
I think the real period for graphic printouts in Tomawac I observed is (TIME STEP)*(COUPLING PERIOD FOR TOMAWAC)*(PERIOD FOR GRAPHIC PRINTOUTS in Tomawac). What do you think about it ? But it only works if COUPLING PERIOD FOR TOMAWAC=1 or PERIOD FOR GRAPHIC PRINTOUTS=1 as already said. Furthermore, the shift of 5s (one TIME STEP) at the beginning is shifting all the graphic printouts and the last time step is not written : my simulation lasts 900s and the last graphic printout is 855. Also, with this shift, the graphic printouts are not the same in Telemac and Tomawac.

So I have questions on two points :
- Is it normal to have results only with COUPLING PERIOD FOR TOMAWAC=1 or PERIOD FOR GRAPHIC PRINTOUTS=1 ?
- Should it be possible to delete the one time step shift between Telemac and Tomawac ? That will say call Tomawac at t=0, call Tomawac at the first iteration and define the COUPLING PERIOD FOR TOMAWAC from t=0 and not the last call ? In which subroutine could I do it ? If it is in a main program I think it is not possible with a coupling...

Regards

Hello,

We are currently trying to correct this for the next version. I enclose a tentative first correction of wac.f for version 6.3 (not tested, not guaranteed, I just changed all occurences of DT_TEL into DT_TEL*PERCOU_WAC, which should improve the behaviour when the coupling period is not 1); However the problem of time shift is probably another one, as you get it even with the period = 1.

With best regards,

Jean-Michel Hervouet

Thank you for the first correction.
Anyway, it is written in the user manual that main code could not be integrated in the Fortran file with coupling...
I'm going to do with the shift adjusting the time simulation with one more time step in duration.

Regards

Hello Jean-Michel,

After our discussion at the user conference, I did some tests about graphic printouts in case of coupling between Tomawac and Telemac.

I have some problems remaining even with your correction :
If the COUPLING PERIOD FOR TOMAWAC or the PERIOD FOR GRAPHIC PRINTOUTS for Tomawac isn't a divisor of TIME STEP*PERIOD FOR GRAPHIC PRINTOUTS, the time records of graphic printouts isn't the same than in Telemac2D. Sometimes there is one more time record, sometimes there is one record less, sometimes the time record doesn't match compared to Telemac2D.

All the tests I did are resumed in the next board :



Here is the model :



Have you some idea to solve this issue?

Regards

Hello Nicolas

I ran your 7 tests to be sure to understand the error nad i have differents explanation according to the test.
First of all, I am not sure that it is clear that when you couple Telemac with Tomawac. It is Telemac that asks Tomawac to run during ome time.
That time is 'COUPLING PERIOD' * 'TIME STEP', then Tomawac will do n time step to run that time and then will do the output according to its period for graphic output.

So in Test 3 as coupling period is 24 and time step 5 ; Tomawac will run pack of 120 sec which will be 0, 120 240 and 360. As period for printout is 12 (which means 60s) you get the output for 0,60,120,180,240,300,360.

In test 4 it is the same (calculation from 0 to 360 by pack of 120) but as printout period is 24 you get the output for 0, 120,240,360.

In test 5 pack of calculation is 60s (from 0 to 300) but you ask in Tomawac a period of output of 24 so it is 120 s so you get 0, 120,240.

In test 6 and 7 the explanation is different, it comes from the fact that output in Telemac is more subtle. In Telemac if the last time step is not a output according to period of output, it records it. Tomawac doesn't do that. that is why you got the records with Telemac2D for 295s and 305s and not for Tomawac.

regards

Thierry Fouquet

Hello Thierry,

If it is assumed than Tomawac is just run at the time 'COUPLING PERIOD' * 'TIME STEP', I have question about some tests :

- In test 3 what is the validity of the output in Tomawac at 60s, 180s, 300s? Because Tomawac was not run at these times. Telema2D run until 300s, so what is the validity of the output at 360s? Tomawac should not have been run for this time...

- In test 6, which validity for output at 300s ?

- In test 7, which validity for output at 360s ?

Regards

Hello Nicolas,

I guess that i didn't explain properly. I didn't mean that Tomawac is just running at the time 'COUPLING PERIOD' * 'TIME STEP', but communicates with telemac at those time, so the intermediate time steps do exist.
Telemac say to tomawac make me a calculation for a simulation of 'COUPLING PERIOD' * 'TIME STEP' seconds then tomawac uses its time step and divides that time into some time interval

regards

Thierry

Thank you Thierry, it is now clear for me.

If the more subtle graphic output, as in Telemac, could be implemented in a future release it will be practice.

For now, I will pay attention to the way to fulfil the parameter files in order to not have surprise in the graphic results!

Regards