open TELEMAC-MASCARET The mathematically superior suite of solvers

I'm modelling a river with sediment transport, coupling Telemac2D with Sisyphe. The bottom of the river is erosionable but the side wall are not. The problem is that Telemac2D assume that the dimension of the bottom is the same as the size mesh so the side walls of the river gets eroded.

For that reason, I identify de NPOIN of the mesh present in the side wall and modify the noerod.f file. The modify code is:
...
open(unit=1,file='/home/emilio/Desktop/PuntosRiberas.txt')
Do J=1,18816
read(1,*) Pto
I=MESH%NBOR%I(Pto)
ZR(I)=ZF(I)
end do
...

where PuntosRiberas.txt is the file where the NPOIN from de side walls are saved and J=1,18816 is the length of PuntosRiberas.txt.

It run well, but still eroding the side walls. What could be the problem?

There is another problem: when I run in parallel, It looks like in parallel NPOIN is not more valid. It's there something like the BoundaryColour for NPTFR nodes?

Best regards,

Gonzalo

Hello,

What you do in your loop is not very clear :

Do J=1,18816
read(1,*) Pto
PtosNoErod(J)=Pto
I=MESH%NBOR%I(Pto)
ZR(I)=ZF(I)
end do

If Pto is a global number the loop should just be:

Do J=1,18816
read(1,*) Pto
ZR(Pto)=ZF(Pto)
end do


If Pto is a boundary number the loop should be:

Do J=1,18816
read(1,*) Pto
I=MESH%NBOR%I(Pto)
ZR(I)=ZF(I)
end do

and I do not see the use of PtosNoErod(J)=Pto. Moreover using unit 1 should be forbidden because it is already open for a geometry file, try 99 instead.

You could just look at the boundary conditions on U to know if you are on a solid boundary:

USE DECLARATIONS_TELEMAC2D, ONLY : LIUBOR

...

Do J=1,NPTFR
IF(LIUBOR%I(J).EQ.2) THEN
I=MESH%NBOR%I(J)
ZR(I)=ZF(I)
ENDIF
end do

and this would work also in parallel.

With best regards,

Jean-Michel Hervouet

Thank you for your answer.

I include your suggestions, and it run well if it is not in paralell.
But if it is in parallel, still eroding the river sides.

What could be the problem? The nodes where I set no erosion, are exported from BlueKenue so I think they might be global nodes. Or i'm wrong?

Thanks,

Gonzalo

Hello,

In parallel the point numbering is locally changed, so it depends on what you have really implemented and we would need to see your Fortran file.

regards,

JMH