open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

we are simulating a dam break scenario for which we know the inflow hydrograph but we don't have any informations about the downstream condition and where we should locate the downstream condition since we don't know a priori where is the water going out of our domain.
The zone of interest is quite far away from the boundary so I thought to use a closed domain (only solid walls) and provide the unsteady inflow via the sources file.
My idea is: in the zone of the sources nodes the bottom has a steep gradient in only one direction and therefore gravity drives the water in the right downstream direction and disregarding the momentum at the inflow is justifiable.
First results look nice and impressive but I don't know much about the plausibility of the results.

Is there any drawback in doing so or am I wrong and is there another common solution for this type of problem?

Many thanks for any hints,
Clemens

Hello Clemens

What do you mean "provide the unsteady inflow via the sources file"?
Do you mean using source nodes?
If it is the case, I think that your solution is a bit complicated and I'm not sure that it is the most optimal one.
You have the inflow hydrograph, thus the easiest way is the use of a liquid boundary condition with a prescribed discharge 4 5 5. The hydrograph could be introduced easily end efficiently by the use of the keyword LIQUID BOUNDARIES FILE.
Your problem is how to tackle the output condition? In my opinion, the use of solid boundary will generate reflexions going back from the boundary to the zone of interst (most of time the dowstream part of the flow is subcritical, which explains these reflexions).
To overcome this problem, the key idea is the use of a liquid boundary far enough from the zone of interest. You can use the Thompson condition with a free output. An other solution is to modify the bathymetry locally in the vicinity of the output in order to generate a supercritical regime which will be insensitive to the condition you impose.
You can even impose an elevation. What is recommendable, with all these ways, is a sensitivity analysis regarding the output boundary condition. By varying this latter more or less, you have to see that the zone of interest is not affected, otherwise you have to extend the domain to make the boundary further.

I hope that this helps

Kind regards

Riadh ATA

Hello Riadh,

thank you very much for the answer.
Yes, I meant using the source nodes in the inflow area and to provide the hydrograph in the source file.
Actually this was my way in other projects with steep bathymetries to generate the initial conditions (water depths only in the creek) for successive flood calculations in which I applied then liquid boundaries.
And now I thought why not use this procedure for the whole flood simulation, so I don't have to care about possible supercritical inflow and stability problems.
So I'm wondering what is the difference between providing the flow hydrograph via sources nodes and providing the flow hydrograph as liquid boundary?

Regarding the outlet zone you are perfectly right. Our zone of interest is quite far away from the boundary so the influence of the reflections from the solid boundary is noticeable some time after the arrival of the flood wave peak in the zone of interest.

Thanks,
Clemens

Hello Clemens,

In theory there is no difference between introducing the hydrograph by source nodes or at the boundary. This will only be handled numerically in a different way, but for the global model it will generate the same consequence far from the boundary. Personnally, I prefere the first solution (imposition as a boundary condtion) combined with a velocity profile of type 4 or 5 (proportional to root square of water depth). This solution allows to tackle properly the prescribed flowrate when bathymetry varies a lot at the cross section where this condition is considered.
I agree that your solution could be more efficient to generate intitial condititon for steep bathymemtry, however, this can be also done with the other solution

Kind regards

Riadh