open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi

I am running T2D and SISYPHE internally coupled and I also wish for SISYPHE to read a TOMAWAC results file of temporally and spatially varying waves. I am reading the TOMAWAC waves using a modified version of condim_sisyphe.f that Pat Prodanovic posted (opentelemac.org/index.php/kunena/17-sisy...-file-1-from-sisyphe). However when I try and run the simulation it stops after

INITIALISATION EN CAS DE COUPLAGE : PART= 0

with a segmentation fault. I've attached my steering files. Does anyone know what might be causing this error or have an example of how they have correctly performed this coupling please?

Much Thanks

Fay

Hello,

One thing to check is your dictionnary of Telemac-2D: the string SUBMIT of BINARY DATA FILE 1 must end with SELAFIN and not PARAL, so that it is decomposed in parallel.
If it is not this we would need to see the Fortran file.

With best regards,

Jean-Michel Hervouet

Hi Jean-Michel,

Yes BINARY DATA FILE 1 is decomposed in parallel. Please find attached my version of condim_sisyphe.f. I have changed the way the SELAFIN file is read compared to PAT's version, so I can temporally interpolate between wave results.

I don't think it is relevant but please note my complete FORTRAN file also contains modified versions of meteo.f, prosou.f and bord.f.

Best Regards

Fay

Hello Fay,

I do not find any obvious mistake but I am afraid of the test IF(AT.EQ.0), I would use something like what follows instead:

LOGICAL DEJA
DATA DEJA/.FALSE./
IF(.NOT.DEJA) THEN
allocations of your various vectors here
various initialisations here
DEJA=.TRUE.
ENDIF

This will cope with the possibility that the subroutine is not called for AT=0.D0 or that the test AT.EQ.0 does not work when comparing a double precision number and an integer.
You should also check that FNPOIN is equal to NPOIN, just in case.
You can also remove the allocatable array VAR : if you read nothing instead of X the program will just skip the record of all the X, and then all the record of all the Y.
Then your prints of TITLE and other variables should show when the segmentation fault occurs.
Just put an order SAVE instead of listing the variables you want to save, that will save them all (just in case we forget one).

With best regards,

Jean-Michel Hervouet

Thanks Jean-Michel

I implemented your changes and found the segmentation error was caused by a silly mistake: I incorrectly defined time in the SELAFIN file as double precision instead of real. Please find the updated subroutine attached.

Unfortunately it still doesn't work: I have attached an image of the TELEMAC printout. condim_sisyphe.f correctly assigns the wave height, period and direction to the variables HRW, TWR and THETAWR as shown by the printout but then I get a MISSING WAVE DATA error message.

THETAWR, HWR and TWR are variable names in the subroutine call, however I did see that Pat used
USE DECLARATIONS_SISYPHE, ONLY: HW,TW,THETAW
and assigns values to these variables not THETAWR, HWR, TWR is this what I need to do? I don't understand, what is the difference between HWR and HW etc?

Thanks

Fay

Hello,

You are nearly done! At the end of your condim_sisyphe subroutine, there is a comment saying that you must add :

HW%TYPR='Q'
TW%TYPR='Q'
THETAW%TYPR='Q'

but there is a RETURN before so you should put it higher in the subroutine. Component TYPR is the tag that says that the data have been given and this is what is checked by sisyphe.f.

With best regards,

Jean-Michel Hervouet

Thanks Jean-Michel,

If anyone else would like to do this coupling I have attached the correct version of condim_sisyphe.f, you then need to add EFFECT OF WAVES = YES to the sisyphe steering file and assign your wave results (hm0,direction,Tp) to BINARY DATA FILE 1 in the telemac steering file. Condim_sisyphe.f will temporally interpolate your wave results to the model timestep.

Jean-Michel is my direction convention correct? I have assigned TOMAWAC MEAN DIRECTION results (which give the direction waves are propagating towards, measured in degrees clockwise from North) to THETAWR.

Cheers

Fay

Hello Fay,

Yes, Thierry Fouquet confirms the direction convention, clockwise from the North.

Best wishes,

Jean-Michel Hervouet

Hi

Using the setup described below, I am experiencing problems when using a PREVIOUS COMPUTATION FILE to initialise my simulation. I obtain an error message for MISSING WAVE DATA, can anyone please tell me do I need to include the wave results in the hydrodynamic file or the sisyphe file?

Cheers

Fay

Hello Fay,

Actually condim_sisyphe is not called at the first time step when there is a continuation continued, so the wave data are not given. The only quick workaround I find is to take sisyphe.f in your Fortran file and remove the test IT(.NOT.DEBU)... ENDIF that is around the first call to condim_sisyphe. Then you must make sure in condim_sisyphe that only the wave data are given and nothing else initialised.
This problem comes from the fact that what does condim_sisyphe is not very clear, it is both used for initialising and for giveng data at every time step.

With best regards,

Jean-Michel Hervouet