open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I want to create a mesh using the triangulation method, primarily as I want the nodes to match with the XYZ points. It does this for most of the channel however when created the mesh outline does not match the highest parts of the mesh. I've attached two images. The first shows The 3D mesh with the jagged edge, with the black XYZ points overlain: there is not match in the Z. The second is of the mesh in 2D where it shows the mesh matching with the XYZ points.

Does anyone know why the mesh hasn't matced on the Z?

Many thanks

Hi
Which interpolation method do you use?
You could try "nearest node" this could works

Regards

Hi,

Thanks for your reply!

I ended up using linear to get the closest fit. It was also party due to my data processing with various awkward cut-offs after interpolation beforehand.

I was wondering whether triangulation was compatible with Telemac in lieu of the standard T3 Mesh generator?
I get the "negative determinant error: element 2408" error when trying to run it. I've made various adjustments - changing the constraint outline, altering the boundary points making sure all elements are CCW - but I still get the error. The element that got flagged doesn't look odd to me. From my view I can't see much wrong with the mesh bar some irregular triangulation at the boundary (see below). Is this the likely source, or does Telemac not like triangulation?

I've attached the mesh just in case it requires further thought.

Hi
Take care of the precision for the mesh coordinates.
This kind of error occur when a triangle is flat...
But in fact this is not so surprising with such mesh...
As you use a channel mesh, you should define a finer criterion for the mesh processing to avoid such results...
Or you should enlarge your domain...

Regards

Hi,

Comparing the original point coordinates to the triangulated mesh suggests that the mesh is precise in how it triangulates between them. However, on the edge it looks to be some mis-alignment - is this what you are referring to?

Regarding flat triangle, I imagine this is when two nodes are in the same place?

As for any finer criterion, I am using the standard delaunay triangulation which doesn't have many options. This is is because I want the channel nodes to align wiht my XYZ points.

Kind regards

Hi
What I mean with the coordinate problem is just about the precision of coordinates in the selafin file.
By default, it's single precision to sometimes, you could generate problem with rounding with leads to wrong triangle...
Flat triangle could also be a triangle made with 3 points which are on the same line, not necessary with 2 nodes at the same place...
About the criterion, it depends on the methodology you use to build the mesh. according to what I see, I suppose you use the channel mesher because it's not a normal Delaunay triangulation, but maybe I'm wrong...
Nevertheless, maybe you could build a mesh with a finer criterion between the limit of the channel and the real boundary of your model. Ideally, the mesh size should be in the same order of the edge-length of the mesh.

Regards

Hello!

I downloaded the latest BlueKenue Beta and the new double precision option for Selafin seems to have overcome the problem.

However, the run now crashes with a MURD3D:ITERATION NO. REACHED 100 ,STOP error. This didn't happen with my previous T3 Mesh, so I assume it's a problem with the mesh. What about the mesh can cause such an error?

The above error can come with using the wrong advection scheme with tidal flats, but i'm using the new PSI scheme released with v7p2 which can work with tidal flats and worked fine for the previous mesh. It works when I reduce the time step to a very low number (0.001!) suggesting that the courant number was too large. I'm assuming this it may be due to high advection velocities, but I'm not sure how this could come about from the change in the type of mesh?

Hello,

Check (with the CHECKING THE MESH keyword option or using another external software) if you have really small triangles (slivers, triangles with really bad aspect ratio...) in your mesh domain reducing the time step could be masking the issue.

Regards,

José Díaz.

Hello,

Thanks for pointing out that keyword. Turns out there were tiny triangles in the domain. Everything is now sorted.

Many thanks