[Gmsh] anisotropic 2D mesh at face boundary

Steffè Walter Walter.Steffe at thalesaleniaspace.com
Fri Jan 14 15:43:31 CET 2011


Hello

 

  I have just seen about the possibility of specifying a boundary layer as a special type of Field.

So I have started to experiment with it in order to compress the boundary triangles towards contour normal. 

The BoundaryLayer is based on an Attractor which refers to all the edges in my miodel.

I have specified:

   blayer->options["hwall_t"]->numericalValue(meshsize);

   blayer->options["hwall_n"]->numericalValue(meshsize/1.5);

   blayer->options["hfar"]->numericalValue(meshsize);

   CTX::instance()->mesh.anisoMax=2;

   CTX::instance()->mesh.smoothratio=1.05;

 

Unfortunately I can not get the kind of mesh I am looking for.

The mesh is influenced by these parameters but it never becomes really anisotropic.

 

I have also looked inside the BoundaryLayerField code to understand how the metric is generated.

To me it is quite obscure the mechanism for retrieving the edge tangent. 

  std::pair<AttractorInfo,SPoint3> pp=cc->getAttractorInfo();

  GEdge *e=GModel:current()->getEdgeByTag(pp.first.ent):

is it the edge which is closer to the given point (x,y,z) or is just the first edge in the Attractor EdgeList ?

 

The edge tangent is used to define the direction of hwall_t and should be related with the nearest edge.

 

May you please give me a hint on the way to follow in order to define a metric that is able to strech the boundary

triangles in the contour direction or, which is the same, to compress them in the normal direction.

 

 

Thanks in advance

Walter Steffè 

 

 

 

  

 

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/gmsh/attachments/20110114/e1f9a2ec/attachment.html>