open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello Telemac Community !

We are facing a very unexpected issue ; w want to make a dumb case (flat bathymetry=0, no meteo, no erosion ..) with a uniform horizontal flow but something's wrong.

The mesh has a rather strange shape : a kind of rectangle with a circular head.
Let's call the bottom rectinilear segment R, the rest of the boundary being called C (for curved, even though two parts of it are straigt).
The overall size is roughly 10km*10km, the resolution is 1km on the boundary and 10m in the center.

I tried many different boundary conditions and none of them gave satisfaction (strange results or crash):
*566 everywhere : forcing H=10m, U=0, V=2m/s
*466 everywhere : H being free, U=0, V=2m/s
*466 on R and 544 on C
*566 on R and 544 on C
*466 on R and 444 on C
*We also tried to force the rectangular part with 466 and let the curved part free (444), in order to create a canal in the bottom. It crashes after only three iterations.
--> See the attached picture for some of the permanent regimes after 1h20'.
As you can see none of them give a nice homogeneous vector field.

The time step is 5s but we also tried with 2.5s.
The law of bottom friction is set to 3 with a friction coefficient of 40.
Other parameters can be read in the .cas file.

You can find attached all the files you need to recreate this case.

Thanks in advance for your help !

Hi
There is no picture

Did you try first with a 2D computation?
For me, rather than tryning various combination of boundary type, I would try to try fo increase slowly the velocity as the initial condition is only a constant water level, trying to impose instantaneously a velocity on all the boundaries, this will probably generate instabilities...


regards

Thank you M Coulet.

I implemented a slowly increasing velocity and it is much better, thank you.
I modified BORD3D : adding min(AT/3000,1.D0) as multiplying factor of VBORL%R(IBORD)=VIT3(IVIT,AT,N,INFOGR).
That way I can easily modify the prescripted value in the steering file.
(needed to do that because I want only V and not a velocity perpendicular to the boundary).

But then I faced another very strange effect, one of the nodes see it's depth going from 10m (initial value) to 0m. The node is "emptying" until it becomes dry. This of course impacts my velocity field on the right side (see picture).
This node has exactly the same boundary code as its neighbours (466).
Anyway, this is just for curiosity as I moved to a model with a simpler geometry, a cylinder. Now the velocity field is nicely increasing almost homogeneously, even though there is a strange acceleration in the middle (pink is the target speed).

Now I struggle to modify the vertical mesh using MESH TRANSFORMATION=2.
I modified the subroutine MESH_TRANSF, commenting this :

DO IPLAN = 1,NPLAN
TRANSF_PLANE%I(IPLAN)=2
ENDDO

and adding that :

DO IPLAN = 1,NPLAN
TRANSF_PLANE%I(IPLAN)=2
ENDDO
ZSTAR%R(1)=0.D0
ZSTAR%R(2)=0.35D0
ZSTAR%R(3)=0.40D0
ZSTAR%R(4)=0.45D0
ZSTAR%R(5)=0.50D0
ZSTAR%R(6)=0.9D0

And I receive the error that you can read in the third picture.
Where am I wrong ?

Erratum, I commented this :

DO IPLAN = 1,NPLAN
TRANSF_PLANE%I(IPLAN)=1
ENDDO

Hello,

It seems that you start to write the line at the 5th column and not the 6th, so that your variable is unknown (STAR instead of ZSTAR).

Hope this helps,

Chi-Tuan