open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear everyone,

I am trying to reproduce a test case initially set up for Delft3D. It is described in Temmerman et al. (2007), Geology.

The domain is a rectangular grid, representing a horizontal domain of 80 x 100 m. The bathymetry is 1.6 m above mean sea level. The tide is simulated by imposing a sinusoidal water-level fluctuation at the north and south open boundary of the grid (amplitude = 2.4 m; period = 745 min; phase difference between both boundaries = 24 s).

That is the setup described in the paper.

However, when I try to reproduce it with TELEMAC, the phase difference between the boundaries is too short so that water enters from both sides, while it should come from one and leave from the other. It's like the propagation of the wetting-drying front is too slow...

Any idea how to solve this problem?

Hi Olivier

It could be good to have all the information about the paper. The consortium is currently looking for a review of validation test case...

I don't have a precise explanation to your problem but 2 remarks:
Based on the theory of wave propagation, the phase difference should be 20s so the value is in a good range but maybe this could be a sensitive parameter...
I don't understand your remark on wetting-drying front. For me we have 1.6m of water height in the domain at the initial time of simulation so no dry zone...

Hope this helps

The paper can be found here:

geology.gsapubs.org/content/35/7/631

My first message was a bit confusing about the bathymetry. The bottom lies 1.6m above mean sea level. At the open boundaries, the water level is initially zero and increases to 2.4m in the first quarter of the sinus fluctuation... So the domain is dry most of the time...

Water depth becomes first positive at the south boundary. Then, after 24s, water depth becomes also positive at the north boundary.

In theory, the water surface should be sloping smoothly from south to north, with water flowing from south to north.

However, after 24s, the wetting-drying front is far from the north boundary. As a result, water is coming from both sides (see snapshot.png from my first message).

Thank you for your help.

Hello,

You should inspire from a case with tidal flats, like the malpasset test case, the variant with positive depths (steering file t2d_malpasset-small_pos.cas).

Among other things:

TIDAL FLATS : YES
OPTION FOR THE TREATMENT OF TIDAL FLATS : 1
TREATMENT OF NEGATIVE DEPTHS : 2
TYPE OF ADVECTION : 14;5 (this may improve the behaviour of the front since the default advection solver is 1, the method of characteristics, which do not perform well for advancing fronts, I put here the NERD scheme, 14).
SUPG OPTION : 0;0
MASS-LUMPING ON H : 1.
MASS-LUMPING ON VELOCITY : 1.
SOLVER : 1 (should be OK and faster with TREATMENT OF THE LINEAR SYSTEM : 2)

I have no documentation where I am so actual keywords may be slightly different.

With best regards,

Jean-Michel Hervouet

Thank you Jean-Michel for your suggestions. Unfortunately, they did not solve the problem.

Here are some pictures to illustrate my problem.

The grid is shown on "grid.png". The tide propagates from south to north and the lateral boundaries are walls.

The two other figures are profiles along the direction of propagation of the tide, at x = 50m.

The problem is illustrated on "profile_start_wetting.png, which shows the very first moments when the domain is flooded. According to the theory of wave propagation, the wave should cross the entire domain in 24s. But here, it seems that it is stacked to the first triangle for a while. Therefore, as I impose water levels at both sides, water enters from the north after 24s.

If we look to further time steps on "profile_wetting.png", the problem disappears when the entire domain os flooded, even though some some wiggles remain. But I guess that could be solved by imposing a discharge for the outflow instead of water levels at both sides.

My main concern is about the propagation speed of the flood front. Maybe there is a criteria in the options of TELEMAC to activate faster the wetting of the second cell?

I hope my explanation is clearer now... and that you can help me

Hi Olivier

It look like there was a minimum threshold for water depth which stop the propagation at the first moment of your simulation...
Maybe you could have a look on this direction!

Regards

Thank you for your suggestion Christophe.

Any clue where to look at for that? I think I've tried everything I could from the reference manual, but maybe I have to look somewhere in the code?

FYI, here my CAS file so far, which reproduces the behavior described above.

I would actually have a second question that could help me to solve the problem...

I could live with a flooding front that is too slow, it might be a limitation of the shallow water equations after all.

But I need to define a proper boundary condition at the Northern boundary to prevent water to enter the domain at the beginning of the flooding phase. A Neumann condition of the type dv/dy = 0, associated with u = 0 should do the trick. Is it possible with TELEMAC?

Thank you for your help.