Anisotropy on a curved component

Rahul_P1

Hi all,

I would like to build a model like the one in the (low-tech) image attached.

It's 3D, and is effectively a multilayered, asymmetrical domed structure. The materials in each layer are anisotropic, and have axes that are usually parallel/perpendicular to the surface of the object (see drawing). Is there a way to assign material properties that follow this 'curve' to the 3D elements, without having to generate multiple local coordinate systems? My hestitation with multiple local systems is not only the time it would take, but that I don't want there to be abrupt changes in the direction of the properties when moving from one system into another.

The mesh is being imported from an stl file, so there is not a surface/geometry associated with it. I'm prepared to try generating a surface from the mesh if that would help.

Any advice much appreciated!

Rahul_P1

Are you using Radioss (Block format) or Optistruct (Bulk Data format)?
If this is Optistruct, you can create orthotropic material properties and even laminate composites (via PCOMP, PCOMPG, and PCOMP) and the default direction for each element is the element's coordinate system (N1-->N2 is the X-Axis and the Element Normal is the Z-Axis for that element. If you use quad elements and arrange them (map them) in the directions that you want, then you can align the N1-N2 element definitions to match a specific selected element. I have a script for the alignment if you need it.

If you are using Block formated Radioss, then you need to create and specify a coordinate system for each desired angle around the contoured face.

The question is if you are doing a crash-type simulation or linear or quasistatic type simulation.

Paul

Rahul_P1

Hi Paul,

It'll be a linear elastic, static analysis. It's based on an stl (generated from a micro-CT scan), and I think the stl will be simple enough to remesh as quads, but I've never tried mapping. Your script sounds like it would be very useful if you're willing to share it, so thanks! However wouldn't this only deal with 2D elements? How would this then be implimented in the 3D mesh that I need to build from those quads?

I usually export stuff to solve in Abaqus so am not that familiar with the Radioss setup or analysis. I'm prepared to learn different formats/solvers if absolutely necessary, but time's a factor so if there's a straightforward way of doing this that'll go to Abaqus that would make me very happy! I have also never worked with composites, and from my limited playing around in HyperLaminate I'm not sure they're appropriate because the thickness of the layers is variable within a layer.

Jen

Rahul_P1

Jen,

Perhaps these below are the scripts you are looking for,

You need to download the script and then run this script from the downloaded location by going to file > run tcl tk from within HyperMesh.

see http://www.altairuniversity.com/support-forum/?mingleforumaction=viewtopic&t=87

you will find this and other scripts in the altair script exchange, go to learning > script exchange from the altair university website or here http://www.altairhyperworks.com/Script-Exchange.aspx?%3ftop_nav_str=1

you will need to log in with your university mail id and login credentials you get after registering to the client centre.

search for the script by just entering orientation in the keyword section.

once you have found the script click on the download button, once you have downloaded the script place it in a convenient location, then you need to do a file > run tcl tk from within HyperMesh. make sure you have a relevant model loaded in HyperMesh when you run the script.

Thanks and regards

Rahul Ponginan

Rahul_P1

Yes, those look like they would do the job. I'll try them out, thanks!

 

Altair Forum User

 _