Isogeometric Postprocessing

Moderator: GiD Team

Post Reply
mvscheven
Posts: 16
Joined: Thu Dec 04, 2014 4:26 pm
Location: Stuttgart, Germany

Isogeometric Postprocessing

Post by mvscheven »

Hello,

we have take an brief look into the isogeometric post-procesing.

Is there any documentation about the .geo file? If we refine the Nurbs description in our FE code we have to write the files ourselves, we assume.

Is there also a possibility (or plans) to visualize stresses, i.e. gauss point results?

Regards
Malte
User avatar
escolano
Posts: 1922
Joined: Sun Sep 05, 1982 10:51 pm

Re: Isogeometric Postprocessing

Post by escolano »

There is a short documentation of the .geo file ASCII format in the customization manual / help
PREPROCESS DATA FILES>Geometry format: Modelname.geo
(last developer versions only)

You are right, the results file with results 'OnNurbsSurface' are related to the control points of the NURBS definition of preprocess (<modelname>.geo file)
If do you refine the NURBS in your calculation code and you express the result in a modified geometry then you must write for postprocess an extre <modelname>.post.geo file according with the result.
When changing from pre to post, in case of having 'OnNurbsSurface' results in the .post.res file then this extra file will be read instead of use the NURBS definitions of preprocess.

Note: if do you know 'a priori' where do you want to refine the NURBS you can try to do it in GiD preprocess.
Geometry->Edit->Edit NURBS Surface/Line...
From the window that will appear you can elevate degrees or insert knots
Then you don't need to write the .post.geo extra file, but use the same definition of the preprocess geometry to calculate

off course instead of do this refinement manually you can do it programatically with some Tcl procedure.

I think that it is now possible to write 'Matrix' results 'OnNurbsSurfaces' (with six components of the symmetric 3x3 matrix typical of strain/stress tensor: Sxx Syy Szz Sxy Syz Sxz for each NURBScontrol point),
but until now we have tested only 'Scalar' and 'Vector' cases of results 'OnNurbsSurfaces'.
If do you have any trouble send us the example, this is a relative new feature in GiD and it is already evolving (new developer version will have some news related with this topic, like allow binary results format)
mvscheven
Posts: 16
Joined: Thu Dec 04, 2014 4:26 pm
Location: Stuttgart, Germany

Re: Isogeometric Postprocessing

Post by mvscheven »

Thank you for your detailed explanations.

We are interested in the description of the .geo-file because usually we are not using GiD as a preprocessor. GiD uses even for very simple geometries trimmed surfaces. From the computational point of view trimmed NURBS surfaces have a few disadvantages, so we prefer to work with untrimmed surfaces as long as possible.

Regarding the stresses my concern was not about the number of components but with respect to the polynomial order of the stresses. As the stresses are derivatives of the displacements, they cannot be interpolated with the original NURBS basis functions.
User avatar
escolano
Posts: 1922
Joined: Sun Sep 05, 1982 10:51 pm

Re: Isogeometric Postprocessing

Post by escolano »

Creating surfaces in GiD it si possible to try to create the surfaces 'as untrimmed' as possible.
for example if do you create a surface by its boundary lines in a plane you can force to be created untrimmed
e.g.
Geometry->Create->NURBS surface->By contour
then click the contextual menu (right mouse-button) and select
NoTryPlanar
and then select the boundary lines (but the shape will be good if lines could be grouped in four sides of a topological rectangle)
without selecting NoTryPlanar the surface will be a rectangle (4 control points linear NURBS surface trimmed with the curves)

And to elevate the degree in u and v directions of a NURBS surface (e.g. the surface number 1) you can write this in the lower command line

Code: Select all

Mescape Geometry Edit EdNurbsSurf 1 DegreeUElevate DegreeVElevate escape 
(or do it from the Geometry->Edit->Edit NURBS window)
Post Reply