open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I want to impose U and V on liquid boundaries such that the flux on this liquid boundary is equal to a given value and such that the vector (U,V) is normal to the liquid boundary (the liquid boundaries are straight lines in my case) and oriented downstream.

The thing is that the flux given by the "flux_boundries" telemac variable is not equal to the flux that I intend to impose. So my question is simple : where does the difference come from ?

Regards :)

Hi
In the flux, there is a section taken in account so how you manage the H value?
Did you use it to compute U and V?
How is the difference between the value computed by Telemac and the one you want to impose?
What about using Q directly and let Telemac compute U and V?

Regards

Hi Christophe,

I don't understand your first question. On the liquid boundaries where I impose U and V I don't impose H and I let TELEMAC handling H on those liquid boundaries. I compute U and V as sqrt(H) and then I multiply U and V by a "well chosen" value such that the flux on this liquid boundary is the one I want to impose. I guess the subroutine Q doesthe same.

The difference between the flux I want to impose and the value computed by TELEMAC can be important such as 54 m3.s-1 (TELEMAC) instead of 72 m3.s-1 (the value I want to impose).

Another qesution is : I impose U and V in the subroutine bord. The values imposed in this subroutine can be modified by TELEMAC through the computation ?

Hi Sebastien

I just mention that a discharge is the product between velocity and an area. If you fix U and V, the value of H is implicitly use to compute the area and then the discharge...
Maybe you could try to impose directly the discharge on the boundary with a velocity profile define as 4 (sqrt(H)) in order to check how it works...

The difference you obtain is large, but did you try to rely it to the width of the river?

About the last question, the answer should be no. Telemac let the value you impose. I think you could check in the result if you probe the U,V values at the boundary node...

Regards

Hi Christophe,

At the beginning of a time step I impose U and V and check that the triplet (H,U,V) "gives" the flux I want to impose by using the formula of line 450 of subroutine flusec_telemac2d. Through the resolution of the Saint-Venant equations, at the end of the time step, H is not the same that at the beginning (!) so I guess this is why I do not get the flux I want to impose at the end of the time step.

This leads me to this question : how TELEMAC handles U and V so that the flux on a liquid boundary at the end of a time step is the one prescribed by the user ?

Hello,

Answer: the finite element formulation leads to essential boundary conditions which are e.g., in the case of a continuity equation, the quantity of water entered at the boundaries. It is this quantity that is used in the proof of mass conservation and it is this quantity which is entered in the domain. In this sense it is somewhat disconnected from the actual U and V imposed in the momentum equation. Of course these velocities are computed to give the right prescribed discharge, given the local sections involved, but we could be slightly wrong in doing that, without spoiling the mass conservation, only the user could complain that the velocities do not correspond to the actual discharge.

With best regards,

Jean-Michel Hervouet

Hello all,

To make sure I do not make any mistake I let Telemac computing U and V on the liquid boundaries where I want to impose the discharge. I only prescribe the discharge I want to impose in the call to the debimp subroutine at the end of the subroutine bord. But still the discharge at the end of a time step is different that the one I want to impose. For example: I want to impose -3.3498683920336494 and 213.24276228594124 on two different liquid boundaries and the flux on those liquid boundaries at the end of the time step is respectively: -3.2680369753246441 and 213.59527526222192. And I do not understand why. Can you please help me ?

Regards