Personal tools
You are here: Home Forum Use Meshing strategies for a fine-to-coarse mesh

Meshing strategies for a fine-to-coarse mesh

Up to Use

Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 23. 2011

Hi, 

I'm trying to mesh an extruded airfoil profile. The meshed part is  the fluid around the airfoil (see "whole_domain.jpg"). I would like to have a fine mesh around the airfoil, and a coarser mesh in the rest of the domain. 

 

I tried two different strategies, but both failed. 

 

1) The strategy is to use 1D-2D-3D Netgen. It is not bad, but for any reason the mesh density is not constant in the extruded direction. See "irregular.jpg". If I export the geom in .step and mesh it directly in the Netgen GUI program, I don't have this problem. 

 

2) The strategy is to create groups of lines to control the element size on the lines. 

- I first create a relatively coarse mesh using "Tetrahedron (netgen)". I then have a regular mesh. 

- I create a submesh of the lines of the box to coarsen the mesh away from the airfoil. I re-compute the mesh. This step is ok.

- I create a submesh on the lines of the airfoil to refine the mesh in the airfoil. But the mesh doesn't fit (see "doesnt_fit.jpg")

 

Any suggestion? 

Thanks

Attachments

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 23. 2011

...image...

Attachments

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 23. 2011

...the last image....

Attachments

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Saint Michael at October 24. 2011

Hi

As about "doesnt_fit.jpg", it seems that you just have not updated visualization of a part of your mesh.

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 24. 2011

Hi, 

Maybe that was the reason....I repeated the process of "strategy 2", but decreasing  the size of the element on the airfoil little by little. Now, as you can see on the image, the elements fit. 

However, it seems that this algorithm has no idea how to manage the transition of the size. 

One solution would be to use the 1D-2D-3D netgen, but then I obtain the problem described in "strategy 1". 

Thanks, 

PL    

Attachments

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Saint Michael at October 24. 2011

You can combine the both strategies: create submeshes and use 1D-2D-3D netgen.

However, it seems that this algorithm has no idea how to manage the transition of the size.

Which version do you use? In 6.3.1 this problem seems to be solved.

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 26. 2011

I am using 5.1.3 (salome-meca 2010), maybe it's the problem...

No idea about the strange "refine pattern" that appears in the extrusion direction?

PL

Re: Meshing strategies for a fine-to-coarse mesh

Posted by SMESH expert at October 26. 2011

No idea about the strange "refine pattern" that appears in the extrusion direction?

Maybe curvature of the surface is incorrectly evaluated along the extrusion direction.

You can use 1D algo to discretize those edges (your strategy (2)) to workaround this problem.

Regards

 

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Pierre Laf at October 27. 2011

Hi, 

Problem SOLVED. Thanks for your help. 

SOLUTION: I installed the latest version of Salome, and the strange pattern in the extrusion direction disappear.   

[For your info "SMESH/MED expert ": You are right, using 1D algo to discretize those edges resulted in a "normal" mesh in the extrude direction, but then I had the other problem, i.e the transition was very bad, like in step5.jpg]

PL

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Lluis M. Biscarri at October 28. 2011

hi,

I'm new on salome and I don't know what strategy to follow in order to do it but, to capture properly the physics of your problem, it would be convenient to insert a boundary layer modelization with triangular prisms over the triangularization of the airfoil surface, and then filling the rest of the domain with tetras, trying to obtain a smooth transition to the coarse mesh of the rest of the domain

I'd like to learn the procedure to create such a mesh

best regards

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Saint Michael at October 28. 2011

Hi

it would be convenient to insert a boundary layer modelization with triangular prisms over the triangularization of the airfoil surface... I'd like to learn the procedure to create such a mesh.

First you get somehow your geometry. Then you create 2D mesh using e.g. NETGEN 1D-2D algorithm, using sub-meshes to get more refined mesh at certain places. Then you assign "Tetrahedron(NETGEN)" 3D algo with 2 hypothesis: "Viscous layers" and "NETGEN 3D Parameters". The latter allows you to attune speed of element size growth as you need via "Growth Rate" parameter.

St. Michael

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Lluis M. Biscarri at November 08. 2011

sorry for being late, Saint Michael, I've not received notification of your post

it helps a lot, thank you very much!

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Lluis M. Biscarri at November 08. 2011

in the procedure you post, you mean...

to create 2D mesh of the whole fluid domain geometry?

then should I need to have 2D geometry faces, for instance obtaining them from explode function, in order to create the 2D mesh and the refinment sub-meshes?

I don't figure out how to combine a 3D mesh (the one that has  the2 hypothesis: "Viscous layers" and "NETGEN 3D Parameters") with the 2D submeshes needed to mes surfaces from which boundary layer prisms have to be extruded

thanks again

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Saint Michael at November 14. 2011

in the procedure you post, you mean...

to create 2D mesh of the whole fluid domain geometry?

Yes

then should I need to have 2D geometry faces, for instance obtaining them from explode function, in order to create the 2D mesh and the refinment sub-meshes?

You create sub-meshes on geom faces where you need more refined 2D mesh, and sub-meshes on geom edges if you need more refined mesh on them. You can use Explode to get geom faces to create the sub-meshes.

I don't figure out how to combine a 3D mesh (the one that has  the2 hypothesis: "Viscous layers" and "NETGEN 3D Parameters") with the 2D submeshes needed to mes surfaces from which boundary layer prisms have to be extruded

When you create a mesh you define
3D:  "Tetrahedron (NETGEN)" +
"Viscous layers" + "NETGEN 3D Parameters"
2D:  "Netgen 1D-2D" + "NETGEN 2D Parameters"

Then you define sub-meshes on faces
e.g.
2D: "Netgen 1D-2D" + other "NETGEN 2D Parameters"
or
2D: "Netgen 2D" + other "NETGEN 2D Parameters" + "Wire discretization" + "Nb. Segments"
or
...

St. Michael

Re: Meshing strategies for a fine-to-coarse mesh

Posted by Lluis M. Biscarri at November 16. 2011

thank you very much, St. Michael

your advice is a great help

Lluis

Re: Meshing strategies for a fine-to-coarse mesh

Posted by bullmut at November 05. 2019

Hi St Michael

I know this is an old thread but its very much related to what i am trying to do now.

Do you have a python script for the workflow detailed below?

I am have the tui commands for creating the 3D mesh but i cant seem to build viscous layers after the fact.

 

 

Thanks

 

 

Previously Saint Michael wrote:

in the procedure you post, you mean...

to create 2D mesh of the whole fluid domain geometry?

Yes

then should I need to have 2D geometry faces, for instance obtaining them from explode function, in order to create the 2D mesh and the refinment sub-meshes?

You create sub-meshes on geom faces where you need more refined 2D mesh, and sub-meshes on geom edges if you need more refined mesh on them. You can use Explode to get geom faces to create the sub-meshes.

I don't figure out how to combine a 3D mesh (the one that has  the2 hypothesis: "Viscous layers" and "NETGEN 3D Parameters") with the 2D submeshes needed to mes surfaces from which boundary layer prisms have to be extruded

When you create a mesh you define
3D:  "Tetrahedron (NETGEN)" +
"Viscous layers" + "NETGEN 3D Parameters"
2D:  "Netgen 1D-2D" + "NETGEN 2D Parameters"

Then you define sub-meshes on faces
e.g.
2D: "Netgen 1D-2D" + other "NETGEN 2D Parameters"
or
2D: "Netgen 2D" + other "NETGEN 2D Parameters" + "Wire discretization" + "Nb. Segments"
or
...

St. Michael

 

Powered by Ploneboard
Document Actions