open TELEMAC-MASCARET The mathematically superior suite of solvers

Your problem is certainly that prescribing the depth in an open boundary is not a well posed problem, and any velocity field may occur if the velocity is free. Many people do this however but generally their boundaries are far from the coast, in deep waters. The correct way to deal with such boundaries is the Thompson method, where you give the known depth and if it is known also the velocity, and the method uses the theory of Riemann invariants to find the correct prescribed values. See e.g. the case called 034_thomson in telemac-2D test cases, and the keyword OPTION FOR LIQUID BOUNDARIES, that applies to every liquid boundary, and thus can be multiple, like 1;1;1;2 if you have 4 liquid boundaries and want Thompson only on the fourth.

With best regards,

Jean-Michel Hervouet

Hi Jean-Michel,

Thanks for the reply..
1. the 'thompson case' I found has only one boundary so in the cas.txt file appears : OPTION FOR LIQUID BOUNDARIES = 2
INITIAL GUESS FOR H = 1
IMPLICITATION FOR DEPTH = 0.6
IMPLICITATION FOR VELOCITY = 0.6

2. I have 3 liquid boundaries and one solid boundary and define it as you can see in fig 1(with BlueKenue): 2open boundaries with prescribed H (B1 and B2) and 1 open boundary with prescribed UV (B3).
Then I change B3 in Fuda-Prepro to 'Liquid Boundary' with Thompson method
OPTION FOR LIQUID BOUNDARIES = 1;1;2

But then I have to define: PRESCRIBED ELEVATIONS
- I use 2.51m and 2.61m, for B1 and B2(corresponding to the first value of each of the time curves--> these curves represent the tide). For B3 I have to use 'zero'?

I run it, but it takes much more time and throughout the simulation is always giving the message:EXCEEDING MAXIMUM ITERATIONS ..

Do you have any suggestions?

Best wishes
Ana

Hello,

Yes you can give 0. for the third boundary.

For the Thompson boundary conditions, you must say in the boundary conditions file which conditions are known, at the beginning of every line :

4 4 4 if nothing is known (test case thomson)

5 4 4 if only depth or elevation is known

4 6 6 if only velocity is known

5 6 6 if depth and velocity are known


then Thompson will decide which one to take and which one to reconstruct according to the theory of characteristics, when there are too many conditions imposed, like 5 6 6, it will relax them, when too few are given, like 4 4 4 it will prescribe some.

With best regards,

Jean-Michel Hervouet

Hi Jean-Michel,

Thanks for the reply...but still the results are coming out strange. The boundary conditions file was already defined (544, etc), as you can see in attach.

I'm testing 2 cases with varying wind:

Run1. one ocean boundary with imposed tide (option for liquid boundaries: 2)--> it seems that the velocity field is coming from the bottom right corner, in fig1 and fig2

Run2. three liquid boundaries with imposed tide on the east and west, the southern side is treated with the Tompson method (option for liquid boundaries: 1;2;1)

Could you please check the files and give me some suggestion?






Best regards

Ana

Hello,

I am sorry I will not have enough time to have a close look at your cases.

What you can try :

1) relaxation in subroutine thomps.f (library telemac2d). Put this subroutine in your Fortran file and try to change the value of TETA which is hardcoded as 1.D0. We do not have too much experience yet on this parameter but it may help in your case (it slows down the variations).

2) enlarge your domain so that the boundaries with Thompson conditions have a larger depth (and thus smaller velocities).

I hope it helps,

Jean-Michel Hervouet

Hi Jean-Michel,

The boundary code 5 4 4 corresponds to condition when depth or elevation is known. Let's say I want to impose this condition on a downstream side and I have a time variation of stage at the downstream side, while modeling a river.

Will the code understand by itself that the prescribed time series is for free-surface or depth ?

Thanks in advance.

Regards,
Sumit

Hello,

You can see what is done in subroutine bord.f: the code expects the free surface elevation (the function SL is called, SL is acronym for Surface Libre, free surface in French) and then this free surface is translated into depth, in the line :

HBOR(K) = MAX(0.D0,Z-ZF(NBOR(K)))

Regards,

JMH