open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello Everybody,
I'm testing a simple rectangular channel with high gradient 7.5% with telemac2d coupled with Sisyphe. I noticed that after some time, the sum fractions in layer one increases from 1 to even 3. I have checked the nodes at which the sum of fractions are more than 1, and these nodes have no problem.
I also changed the ZERO value from 10-7 to 10-3 but didn't help alot. I would appreciate if anyone could help me how to deal with this problem. attached are a snapshot of listing printout, case files, geometry files, FORTRAN file and etc.
any help is highly appreciated.
best Regards
Najib

and here is the listing printout

Hello,

I can have a look but a number of files are missing :

* Fortran file (init_compo.f)
* liquid boundaries file
* previous computation file
* stage-discharge curves file

Moreover some names of files (e.g. the results file) contain the character % and cause the Perl scripts to crash.

There are others questions like yours on the forum. The subroutine layer.f is very tricky to write, maybe we can find a better algorithm. However the few recent times I looked at a problem it was actually created before arriving in this subroutine.

With best regards,

Jean-Michel Hervouet

Hello Dear Jean-Michel,
Thank you for your reply. Here are the missing files.
liquid boundary file: for_test.txt
Stage discharge curve: H_Qrelation_SA_74_78.txt
previous computation file :test_case_7.5n.f
FORTRAN file : init_compo.f
Telemac2d steering file: tel.txt
Sisyphe steering file: sis2.txt
I removed the percentage character from previous computational file. I also think there would be some problem in my FORTRAN file (init_compo.f)where layer.f is also included.
Best regards
Najib

Hello,

Thanks, the previous computation file is missing (probably too large) but I could reproduce a similar error without it. I get a "case 4" error in layer.f and am looking at it... Wait and see..

With best regards,

Jean-Michel Hervouet

Hello,

A first analysis shows that the subroutine layer.f is a priori not guilty and is given misleading data. Actually in subroutine bedload_evol.f I see that an available layer of sediment is about 4.98 m, this is in array ELAY%R(628) for example. Then in layer.f only an active layer ELAY0 equal to 0.2 m is considered, and this causes a negative fraction. layer.f finds that we are in case 4 (only one layer eroded) whereas the value of ZFCL_W shows that we have eroded more than one layer, this is due of course to the wrong layer 4.98 m that is initially considered for erosion.

I would hint that there is a misunderstanding somewhere in the data, or a bug, for me ELAY should be, as it appears in bedload_evol, less than ELAY0. However if the data are wrong we should be able to check it and stop the program.

I have now to discuss the matter with my colleagues, so you will have probably to wait for next Thursday to have more details.

With best regards,

Jean-Michel Hervouet

Dear Jean Michel,
Thank you for looking at my files and helping me. What I basically want is to have two layer model ( active layer with thickness of 0.2m & a 2nd layer of 4.9m thickness). The total erodible bed must not be more than 5m. Maybe I've given contradictory inputs in my fortran file which is misleading the program. I'm also checking it.
Thanking you again
With regards
Najib

Dear Jean-Michel,
When I run the model without previous computation, i get this error. ( plz see attached picture).
Best Regards

Hello,

For some reason I do not get exactly this error, but we shall answer tomorrow, Pablo Tassi found that you have an error in init_compo.f and we'll do a common message.

Regards,

JMH

CORRECTION: the thickness for 2nd layer 4.8m