open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I want to introduce a different layout for marine works by changing the boundary definition of the mesh. I have tried to create a 'hole' in the existing mesh and replace it with a new mesh with the new boundary and bathymetry (following the BK tutorial, paragraph 6.2). However, when I merge the two meshes, the boundary conditions are not sorted. I get a 'boundary' definition around the whole of my submesh.

I get this behavior only when the 'hole' contains part of the boundary and not when within the mesh.

Is there a way to overcome this problem?

Regards,
Costas

Hi Costas

Could you illustrate your message and also describe te different step you follow to acieve this task.
I think I already done some similar things but I'm not sure

Kind regards

Hello Christophe,

This image best describes the error:



The different step I took, was to export the submesh edges to a shapefile, import to AutoCAD, replace the old boundary layout with the new (making sure that I keep the common nodes intact) and import the new outline back to BK.

I generated a new mesh without resampling the outline and merged into the hole of the existing mesh. Everything looks normal, until I try to define the boundaries, hence the image above.

Is it possible to edit the 'node type' of the outline nodes somehow?

Regards,
Costas

Did you merge both T3S mesh of did you use a geo file (selafin)?

I think you could also use the large mesh as a submesh for the small one but I'm not sure exactly how this should works.

As far as I know, it's not possible to change the node type of the boundary to convert it to an inside node. To obtain such situation, this mean in fact that the triangles are not connected and the you probably have 2 different nodes at the same location (on the common nodes)

Regards

Hi Guys,

Your analysis is correct Christophe. Also, the large mesh can be used as a submesh to the small one or vice versa. It makes no difference.

Blue Kenue does not use the node types field when generating a BoundaryConditions object so changing these fields would have no effect. The code works by identifying edges that exist in only 1 element (ie. exterior and island edges.)

Costas: The process you followed seems reasonable to me.

If you have the large mesh and the detailed harbour mesh as 2 separate objects you can use the "Edit->T3 Mesh->Merge Mesh..." function.
This is faster than regenerating the new mesh with a sub-mesh.

Could you please post the mesh here so I can try to figure out what went wrong.

Cheers... Martin

Hello,

The meshes are not large, so generation time is not an issue. The aim is to have 2 meshes that are coincidental, excluding the area around the structure.

Here are the constituent meshes just before the merge operation. They were not extracted from a selafin object.



I have tried this a couple of times before and always got this problem when the submesh contained part of the boundary. So it is either BK, or I am doing something systematically wrong.

I hope the files will provide more insight. Thank your for your help.

Best Regards,
Costas

Hi Costas,
There are no attachments. please try reposting.
Martin

Oops, I forgot you can't post any file extension . Here is a zip archive:




Costas

Hi costas

I think you're facing a rounding problem!
Even you're thinking the boundary nodes are the same, if you select the same node on both mesh, you will see that the coordinates are differents.
In such case, the merge cannot work properly which explain the creation of a boundary condition inside your model as in fact, th mesh are not connected!

I didn't try but you probably have to make a shift in your model to avoid the roundig problem. As i see, it seems this is the boundary of the submesh which is rounded at 10-1.

Hope this helps

to illustrate, I join two screenshots

Hello Christophe,

You are right. Doing all the operations in local coordinates certainly corrects this issue. Looking more into it, it seems that the rounding issue seems to affect only the boundary nodes during mesh generation (without boundary resampling). Creating just the submesh in local coordinates and shifting back to grid coordinates and merging back also works.

So it looks like a BK precision limitation that one should be aware of.

Many thanks for your help.

Best Regards,
Costas