open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I encounter a issue with TELEMAC 2D models on boundary condition built by Lekan (see: github.com/vcloarec/ReosProject).

I was wondering if this is a bad building of the boundary condition files, maybe I am missing something, or if it is a TELEMAC issue.

I have built a simple test cases to reproduce this issue:

• a constant inflow at upstream
• a constant level at downstream
• a quite simple geometry with a large downstream limit with corners far away from the main flow.
• important information: equation is Finite Volume, there is not issue with Finite Element

Here the mesh


You can found the model files for the case 2 in attachment.

Lekan builds open boundary condition for water level condition as follow:



If the the top right corner is right-angled, no problems, see dynamics results:



But if the geometry is changed with an obtuse angle on the top right:



Here a zoom to the corner:



Other changes, but with acute angle:



The bias is inverted.

Hello,
Interesting remark! According to T2D user manual, section 4.2.1:Thus, for a symmetric outflow boundary, the boundary must span between (boundary) nodes 13 to 76, NOT 77, like so:



Also, overconstrained triangle at the boundary it is not advisable:



Swapping edges is generally recommended and should be done in a sensitive case like yours:



Best Regards,
Costas

PS. Excellent graphics!

Thanks Costas for your answer,

I changed the mesh accordingly to your advice:



And there is still this big bias in the flow:




Even if I put the extremity of the liquid boundary further from the corner:




Here the files of the last case:

Which version are you using? In v8p3r2 I cannot reproduce the BIAS in your original case-2 set (HIGHEST VELOCITY plot):

hmmm, maybe a too old one, mine is v8p3r1... I have to update...

v8p3r1 is not that old. See the changelog for changes in default values or any other development that can affect your case. Anyway, v8p4r0 is out, try switching to the latest one.

In the test I shared, the equation is Finite Element. And it seems to works with FE. Can you try to switch to Finite Volume?

(see github.com/vcloarec/ReosProject/issues/9...uecomment-1462154636)

Unfortunately, the BIAS error remains with v8p3r2 when switching to FV.
I only had to add

Setting FINITE VOLUME SCHEME = 5
appears to eliminate the BIAS for me.
Here is my steering file:

Perfect!!!
It works also on my side with v8p3r1.

That will make me add an option to change the VF scheme from the UI, to not use the default one...

Thanks a lot!