open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,there:)

During the tracer simulation, I tried to take decay into consideration. In the difsou.f , the YASMI was set TRUE. Due to the change, error happened. It is said "GRACJG (BIEF) : EXCEEDING MAXIMUM ITERATIONS". Then I increaed the interation number. It still didn't solve the problem.

By the way, if the YASMI remains FALSE,the computation goes well.

Does anyone have any idea about the reason for the error?

Thanks in advance.

Hello,

It could be that your implicit source term SMI is not correct. You can look at the header of subroutine cvdftr.f in library bief. SMI is in the right hand side of equation df/dt, in a term SMI*F(n+1)/H, it must be so for mass-conservation result. So it could be that you put a wrong sign, or did not take into account the division by H. For a decay this term should be negative. You can inspire from the example at the beginning of difsou.

With best regards,

Jean-Michel Hervouet

Hello,jmhervouet,thank you for your advice.

But I still can't figure it out.

I saw the example in the DIFSOU.f and increase some similar sentences,which are as follows:
CALL OS('X=0 ',X=TIMP%ADR(1)%P)
CALL OS('X=0 ',X=TIMP%ADR(2)%P)
DO I=1,HPROP%DIM1
TIMP%ADR(1)%P%R(I)=-0.00001*HPROP%R(I)
TIMP%ADR(2)%P%R(I)=-0.00001*HPROP%R(I)
ENDDO
YASMI(1)=.TRUE.
YASMI(2)=.TRUE.

So i think I noticed the points you mentioned.

The divergence occured after some time steps.

Bisides, acccording to my understanding, explicit source terms means source and sink and implicit source terms. I'm not sure about that.

Best wishes

Hello,

Another possibility is that you have tidal flats. I corrected a bug recently into cvdftr.f (library bief) :

change the sequence:

DO I=1,SMI%DIM1
IF(T10%R(I).LT.1.D-2) THEN
T4%R(I)=0.D0
ELSE
T4%R(I)=SMI%R(I)/H%R(I)
ENDIF
ENDDO

into:

DO I=1,SMI%DIM1
! CORRECTION JMH 26/04/2012
! IF(T10%R(I).LT.1.D-2) THEN
IF(H%R(I).LT.1.D-2) THEN
T4%R(I)=0.D0
ELSE
T4%R(I)=SMI%R(I)/H%R(I)
ENDIF
ENDDO

because the previous implementation has a risk of division by 0 if the depth changes from more than 1.D-2 into 0 in a single time step.

If you downloaded with svn you just need to update bief sources and recompile.

Last thing, in your implementation in difsou the two CALL OS are useless and you should put D0 at the end of 0.00001, but it is only optimisation and a detail.

With best regards,

Jean-Michel Hervouet

Thank again:)

I didn't downloaded source codes with svn.And I couldn't find the subroutine in existing source codes in my computer.


So how can I include the new CVDFTR.F in fortran file?

Best wishes

Hello,

You will find cvdftr following the path :

...\V6P1\bief\bief_v6p1\sources\cvdftr.f

... is the directory where you have put the installation of Telemac. You can put it in your FORTRAN FILE after modification.

Best regards,

Jean-Michel Hervouet

Dear jmhervouet,

I found water depths in some points and some time are zero.

With your modified fortran file, the problem has been solved.

Thanks a lot:)

Best wishes

Dear jmhervouet,

I have another question:

In the printout file, i saw QUANTITY CREATED BY SOURCE TERM.

Could you please tell me how the variable is counted out?

Best wishes

Hello,

The quantity added by source term is so far only the quantity added by point sources or by an explicit source term. Implicit source terms are not taken into account in the balance (so you should see a loss of tracer occuring when you add decay). I hope I have time to implement a balance including implicit source terms in the next version.

With best regards,

Jean-Michel Hervouet

Dear jmhervouet,

I found a new problem after the tracer simulation is finished,that is, negative values appear in some areas around the tracer source location.

During the process, the initial value of tracer is set zero.

Do you know what's the reason for it and how to solve it?

Thanks in advance.

Best wishes