MMG2D giving different results for the same metric, depending on the regions defined in the domain

User Help
#1

Dear all,

I am using MMG2D inside KratosMultiphysics to remesh 2D meshes, and I have seen that the final remeshed result is affected by the regions that I define in the domain, even if the same metric is used.

I can only upload two images so I will directly show the final result. For both images the metric is the same in the whole domain. The difference is that in the second one, a group of triangles is defined forming a line in the upper part.

image
image.png796×679 25.4 KB

image
image.png701×676 25.8 KB

We can see that in the second case, where the group of triangles is defined, the final result is affected by it even though both have the same metric.

Is this the intended behaviour? I would expect it to give the same result regardless of the regions defined.

Find attached both .sol and .msh for both cases (files starting with line_… correspond to the second case).

Thank you for your help!

Marc

line_output.zip (52.8 KB) normal_output.zip (49.8 KB)

MMG2d Null metric det error
#2

Hello Marc,

Mmg is more constrained when you have 2 domains in your mesh (<=> the strip of triangles inside your second mesh), because:

  • the remesher must preserve the two domains and their colours;
  • the meshes of both domains must be valid for finite element computations (no element with 2 boundary edges and no internal edge connecting boundary points): it leads to insert at a points layer inside the domain defined by the strip of triangles;
  • mmg must respect the geometrical features and approximation of the interface between the domaines: it computes a metric field linked to the curvature of the interface and intersects this sizemap with the physical metric provided by the user;

I hope that it will helps you to understand the difference between the 2 outputs.

Regards,

#3

Hi Algiane,

Thank you for your response.

I understand your explanation and I think we can close this post!

Regards,

#4

Thanks for your feedback!

Best

#5