open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I'm working on a weir, in a simple case to begin and making some tests. I have some isues that I can't solve :

My model is composed by two squares (coming from the validation case weir) linked by a weir. The bathymetry is constant in each square but different between the squares. I'm working in parallel mode.

In the first test, the water is coming from the left in the square with an imposed level frontiere. I impose a simple sinusoide. The evolution of the water level is well represented on the left square. On the right square, the level rise is well done, but there is no low-level even if the water line is higher than the weir level. This model is in attachment called "reservoir_double_gauche"



In the second test, it's the same method but in the opposite way : the water is coming from the right square. The evolution is well represented on the right square but nothing happens on the left square, the water doesn't cross the weir. I have checked the volume computed in clsing.f and it seems ok. If I reverse the node numbers in the weirs data file, the water level evolution on the right square is abnormal compared to the liquid boundary and on the left square the water level fall down while the the weir level is higher. This model is in attachment called "reservoir_double_droite"



So here is my questions :
A weir is it directionnal?
When the water cross the weir in one way, the coming back is it possible?
Why a weir work in one way (from left to right in my case) and not in another (from right to left in my case)?

Thanks by advance

The attachments :

Hello,

You are probably right, though I see no fundamental reason why the flow could not go in both directions. So far it cannot work because e.g. in subroutine loiden.f, the elevation upstream must be higher than the weir elevation, which must be higher than the elevation downstream, otherwise the sqrt crashes. We could change the implementation and get negative discharges instead of putting 0. (in loiden and loinoy) and hopefully it works.

Regards,

Jean-Michel Hervouet

Hi,

I have checked the computation of the flow in loiden.f and loinoy.f. For me the sign of the flow and so the direction is not compute here.
In clsing.f there is a test to check which side is higher than the other (A or B upstream). The sign of the flow is adjusted if it is necessary (Qab=-Qab). When I check the sign and the evolution of Qab, it seems ok in both simulations.
In the last part of clsing.f, clhuvt.f is called. By removal, I think the mistake comes from here but I don't find it!
Any idea?

Hi
Have you some water on both sides of the weir?
I didn't checked but as discharge boundary doesn't work without water, it's possible there is also a limitation for the weir (which works as an internal boundary).

Hope this helps

Yes there is some water on both sides of the weir.

Even if there is a limitation, the limitation should be the same for water coming from left to right or right to left...

Are you talking about a limitation like in a culvert (QMAX in siphon.f)? I don't see such a limitation in the code for a weir.

Regards