open TELEMAC-MASCARET The mathematically superior suite of solvers

I would like to ask somebody of you for advice that could solve my problem. I have a mesh that represents floodplain area with river channel inside. Boundary conditions are liquid with prescribed flowrate, or elevation respectively. I know that during the real situation that is simulated water remains in the river bed between embankments and doesnt occur in adjacent floodplain. What should I do to force the model to hold the water in the channel and just when free surface reachs the top of dykes, water can overflow to floodplain. Please, see attached pictures with mesh and results. Thank you very much. I am TELEMAC2D beginner.

Hi,

Your problem is probably caused by a poor discretisation of your dikes. You will have to zoom on your dikes or share your mesh to get hints from the community.

Hello,

Yes, you need to have a good refinement of dykes, i.e. several points at the top of the dyke. Another possibility is the new feature :

THRESHOLD DEPTH FOR RECEDING PROCEDURE

By default it is 0. but you can try a few centimeters, let's say 0.02 m to start with, and it may suppress the problem, it did for some users.

With best regards,

Jean-Michel Hervouet

Hello,
thank you for your previous advices. I have tried to improve our mesh (focused on dykes) but results are the same. Water still appears in floodplain. Maybe I enter bad values (boundary conditions, initial conditions), maybe my mesh is incorrect. I am starting to be confused and frustrated newbie, because I have no idea, what is wrong. Could you be so nice and have a look at attached compressed archive, that contains mesh (selafin file), boudary conditions and liquid boundary file? Discharge on input is constant in time – steady flow. Thank you one more time for your hints.

Fany

Hi
indeed it's not really good to impose the upstream discharge on such section (not perpendicular to the flow) your problem comes from your initial condition.
You have water in the flood plain and no possibilities for emtying it!
Change your previous computational file to ensure the water height is 0 in the flood plain and it should works

regards

Hello,
I adjusted initial conditions as follow: flood plain – water heigh = 0, channel – water height = 0.1 and after that results are much more real. However, I have one more question. After each iteration (time step) areas that should be dry were flooded by water. At the beginning, is it better to inundate the whole domain and let the floodplain dry continuously during the computation?
Thank you.

Regards
Fany

Hello,

This is the difficult problem of initial conditions on large domains. Note that pre-processors like Blue Kenue and Fudaa can help you setting an initial free surface. Though it may be a long process, you can consider also starting from a dry domain (except some water at the entrance) and fill it progressively.

Regards,

JMH

Hello,
thanks for previous helpful answers. I would like to ask you a few more questions. What is the principle of variables time step and number of time steps? Is my assumption correct that longer duration of calculation with smaller time step brings better and more accurate results? Sometimes it seems that results from time step e.g. 300 (half time of simulation) is more realistic than results from the last time step 600 (end of simulation). What values should be roughly assigned to variables time step and number of time steps in such a large domain (about 7 km long and 3 km wide) with 50 meters mesh resolution and steady flow of unknown duration (input discharge)?
Thank you.

Regards.
Fany

Hello,

Smaller time steps do not always give better results, only smaller time steps AND smaller mesh size should tend to a convergence. Generally speaking a reasonable Courant number gives good results, let's say around 1. For example advection schemes with very small time steps may be very diffusive. A good indication is the number of iterations of solvers. 10 to 20 iterations is OK.

JMH