open TELEMAC-MASCARET The mathematically superior suite of solvers

I am creating an tel2d model for test 2 in the benchmark models. The report summarizing the tests and all software packages can be publicly accessed from evidence.environment-agency.gov.uk/FCERM...els_Report.sflb.ashx. However, the data set used for the models is not in public domain so I cannot upload the model here.

Test 2 is to assess the package's ability to handle disconnected waterbodies, wetting and drying of floodplains, and to predict the inundation extent due to momentum flooding on a complex topography, with an emphasis on the final distribution of flood water rather than peak levels. I have attached a screenshot (on page 32 of the report) to quickly introduce this test. The left figure shows the model setup, and the right one shows the water depths of most packages.

As shown in the left figure, there is one single inflow boundary at the top left corner, and elsewhere is solid boundary. Initially the bed is dry. I am feeding a tiny initial depth at the inflow boundary so the model can calculate inflow velocity (as suggested by many posts).

I cannot get water to depression 5 (water depth zero), as pointed by the arrow in the right figure. I have tried the following:
-different velocity profiles for the inflow
-different turbulence models
-different scheme of advection
-different time steps
-finer grid resolution at the inflow boundary

I also attached my case file. Could you suggest what else I could try? Thanks

Hailiang

Hello Hailang

Can you change friction coefficient?
to let the flow arrive to a further point, you should maybe decrease the friction.

You can also try with finite volumes (But for such cases, finite elements and finite volumes will perform almost in a similar way.)

with my kind regards

Riadh

Hi,

I know from other 2D software participating in the UK benchmark tests which used the same friction coefficient (0.03) as you, so I wouldn't change it. And since most of them don't account for turbulence, for comparison reasons I would turn it off (or set to laminar viscosity) in Telemac-2D FE.
Maybe your problem is related to the mesh generation.. how you discretized the depressions and the zones in between?

Could you share the mesh?

By the way, some time ago I was also interested in testing Telemac-2D by means of the UK benchmark test cases but had no time to set up them.
It would be really nice to do that. What says the consortium to an official Telemac participation in these benchmark tests?

With best regards,
Clemens

Hi

The consortium planed to work on the UK benchmarks...
The main idea is to integrate the benchmarks in the list of Telemac tests cases and have some discussions with the Environment Agency in order to integrate Telemac in the future release of the benchmark report.
If some users would work on the preparation of some of the benchmarks case, The consortium will be happy to integrate theirs works in the test case database...

Regards

Thank you for the reply! The test specifies Manning's n coefficient 0.03 and cell resolution 20m so we are not flexible on these setting.

Clements, could you elaborate on turning the turbulence off or setting it to laminar viscosity? I have used Tuflow (one participant) and know it uses constant or Smagorinsky coefficient for turbulence modelling.

I did try to refine grid at the boundary but did not improve the results, although I do not believe other software also refined the boundary cells. I did not treat depression zones separately. Cells in the domain have similar density except for refining the inflow boundary. I have also setup a Tuflow model (square grid) and can see water at depression zone 5, so do not believe we have to play with telemac mesh.

It looks to me water does not have enough momentum to arrive depression zone 5. I saw the same "issue" for test 3, where not much water gets to the 2nd depression.

I will double check whether I am allowed to post the mesh here. Thanks again for your inputs.

Hailiang

Hi
I don't have in mind all the parameters of this test case but we should keep in mind some difference between codes could sometimes explain some discrepancy in the results.
In Telemac, the information and results are supported by the nodes of the mesh...
In the other software, if the results are given by cell, this could lead to small effects...
It could be great to have a view of the topography of the model built...

regards

Hi,

ok, I didn't know that the cell resolution is given.

You could try
TURBULENCE MODEL = 1
VELOCITY DIFFUSIVITY = 1.E-6
and
comment out
SOLVER ACCURACY =1.E-6
MASS-LUMPING ON VELOCITY =1
IMPLICITATION FOR DEPTH =1

in order to apply default values.
I suppose you tested already the chosen time step of 5 seconds with regard to oscillations / stability of your model.

With best regards,
Clemens

First of all Clemens, I have tried your setup and still get no water at depression zone 5.

Riadh, thanks for providing the info. I am very interested to know the results of test 2 and test 3 from the German teams. We are also working on these tests. We did not test the tests requiring 1D/2D linked modelling including test 7 and test 8B. But we also see very good fit with the results from other modelling packages for the other tests.

Any chance i can get the contact info of the team? My email This email address is being protected from spambots. You need JavaScript enabled to view it. Thanks,

Hailiang

Dear Hailang,
I'm a member of the "german team". I send you an email on 6th of february.

Best regards
Thomas

Dear All

best wishes for the new year (Clemens and Cristophe ).

These cases are being run by germain teams within a germain benchmark project where they use several codes among them Telemac. Rebekka is following this project and thus I had some feedbacks. Telemac results looks not bad at all for cases I have seen.
==> Clemens you will have some inputs during the next sci-committee hopefully.

Actually I do not remember if this case has already been tackled or not (I have to verify).
As suggested by both of you, these cases are really interesting and we can introduce some of them into the validation (qualification) database of Telemac.
To be discussed...

with my kind regards

Riadh