open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I'm trying to couple the hydrodynamics of our very large 3D model with Delwaq. In the *.hyd file provided by telemac the "number of exchanges" was not mentioned as a number; instead only ******* was written. I had a look in the code and could calculate the number of exchanges manually and this gave me a 7 digit number. When coupling smaller models with Delwaq I don't get this problem, so is it possible that the variable that holds the "number of exchanges" is limited to 6 digits?

If this is so, are there other variables that could have the same problem and make that other generated Delwaq files are not correct as well?

Best Regards,
Sven

Hello Sven

I'm not familiar with Delwaq but after having a rapid glance on the subroutine wrihyd.f (sources/telemac2d) you can see that this routine writes a lots of hydraulic data for delwaq and that your problem is at line 224 where the number of exchanges is hard-written as an integer of 6 digits.
In your case the number of segments of your mesh (added to number of open boundaries nodes) probably exceeds 6 digits which explains the problem.
To fix this issue just change I6 at line 224 by I7. I hope that this will not generate a problem from DELWAQ side :)
I don't think that there are other problems similar to this one in the remaining routine. But by precaution change the other I6 as well.

I hope that this helps

with my best regards

Riadh ATA

Hi Riadh,

Thank you for your suggestion!
I was already checking this subroutine but found that changing I6 to I7 didn't solve the problem. I'm running the model in parallel and found that all the way in the bottom of subroutine tel4del.f there is an if loop IF(NCSIZE>0) then a command file for DELWAQ is written (NCOB) and after this loop wrihyd.f is called and NCOB is transferred to this subroutine. I can't really figure it out, but is it possible that I have to change something in the if loop in tel4del.f as well?
My variable NSEG is a 7 digit number

Cheers,

Sven

Hello Sven

Sorry I didn't have time to check again your issue.
Did you fix the problem?
If not, another idea maybe to see if the problem comes from variable MBND which is a counter for open boundary nodes. If you can print it out and see if it is the cause of these troubles.
Just a naive question, you say you have a huge model that you run in parallel. CPU time with Delwaq (which runs only in serial mode, or in shared memory parallelism (openMP ?)) is not prohibitive for you ?
can you give me some statistics about the model size, the total simulation time and the generated CPU time.

with my kind regards
Riadh

Hi Riadh,

I started working again on this project. With a smaller model I have no problem. With our large model (472400 nodes x 5 planes; we simulate periods ranging from 5 days to 90 days and run the model (depending on processor availability) on 50 to 180 processors. The speed up time varies between 5 and 10). Your first suggestion to change an I6 into and I7 in the wrihyd.f subroutine was good, but did not solve the problem. On the user conference you suggested to have a look at GRETEL, so I did and there in gredelhyd_autop.f I needed to change also the I6 to I7. I'm compiling the code right now and will test if this works.

Concerning the cpu time for DELWAQ for this large model, you are also right. I'm afraid it will take a long time. But we have already a calibrated mud model in Delwaq, developed on a Delft3D model of the Schaldt estuary. We want to use the Telemac hydrodynamics for this model without any further calibrating and compare. I'm also working on a sediment model in SISYPHE. We will also compare results with the DELWAQ model.

Hello Sven

This looks interesting. Please keep us informed about your investigations.

with my kind regards


Riadh

Hi Riadh,

if you go to sources/utils/gretel to the subroutine gredelhyd_autop.f and go to line 343 I changed I6 to I7 and in the next line a space is missing behind ...-direction.
"grid-cells-second-direction",NSEG2+MBND
--> "grid-cells-second-direction ",NSEG2+MBND,
I changed also in sources/telemac2d wrihyd.f in line 224 I6 to I7. This to solve the problem in serial mode.

These changes give a correct .hyd file for Delwaq for larger models.
Can you insert these changes in the trunk?

Best regards,

Sven

Hello Sven

I integrated your remarks on the trunk and this will also be done for V7P2 release.

with my kind regards

Riadh