wireframe,surface or solid and finite element analysis.

following is the text from sdrc ideas tutorial
"I-DEAS uses a solid representation of part geometry.Because it's a
complete representation, solid geometry supports any type of design and analysis application that requires geometry information. For example, it gives finite element simulation "
Now my question is , suppose I supply a similar representation of a particular solid in pure wireframe format( say using autocad),can I do FEM analysis on it.
My friens who is working on Hypermesh told me that hypermesh has very limited modeling capabilities but whenever need arises he can create a wireframe model of a solid and does fem analysis on it. He also told me that many times the CAD models are not complete and has some incomplete features.So they repair it in hypermesh using line,spline commands and then it is used for FEM.
He said to me that geometry is required only for reference and meshing is a totally different issue.
what is correct? I am totally confused. (as usual :).
regards, Yogesh Joshi
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

-----------------------------------------
Finite element analysis always requires the creation of analysis elements (planar mesh, solid mesh, bar, rod, etc.). Nurbs, bezier, etc. surfaces used for most present day "solid" reps are of no use to the solver, but most pre-processors can reference them for mesh generation so you are money ahead sending a solid (or surface) vs. wireframe rep to it.
" ... he can create a wireframe model of a solid ..." The wireframe is just a reference for element / mesh creation.
"... many times the CAD models ar not complete..." If the model is supposed to be a solid rep any "incompleteness" is due to something getting lost in translation. (Actually, most solid models contain too much detail for FEA purposes.)
Basically, the I-DEAS statement is marketing psycho-babble. All the attributes given to the "solid" can also be given to any set of surfaces that define a volume; e.g. a closed surface or quilt. Once a volume is defined, all that's necessary to have a "solid" is a database entitiy for mass.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks Jeff! but still I have some doubts.. Suppose I draw a cube using only 3D lines and I draw the same cube using solid modeling operations(extrusion). How will fem solver treat them..? in similar or in different manner.?
regards, Yogesh Joshi
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

The solver won't deal with either one. It can't see the wires or surfaces. In either case the user will have to mesh the objects using whatever functions are available in the pre-processor. The meshing operation is quite similar to other surface modeling techniques (exactly like creating an old Acad polyface mesh, if you are familiar with that) for shell / plate models. Where surface data is present most pre-processors can auto mesh the surfaces for a shell / plate model or fill the volume with a 3D mesh for a solid solution, with at least some degree of success. I don't have the foggiest how the solver internal math works, but it requires a matrix of "nodes" to do the computations. These nodes are the corner vertex (at least) definitions of "elements". Neither wireframe models nor surface / solid models created by CAD systems have any representation of this node matrix. (The exception being the previously mentioned polyface mesh models. These will translate directly to FEA mesh; all that the user has to do is merge coinicident nodes, define material and properties, etc.)
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
yogesh wrote:

Sure, your friend is technically right, you can do FEA with just wireframe. But once you need to do a fairly complex part, using a solid model and a decent modern mesher (not sure SDRC qualifies 8-) will be MUCH faster. Of course that's not to say you shouldn't learn what FEA is all about and how far the tools will take you. Vince Adams' book on FEA is a very good thing to read.
I think there was a day when modelers automatically redrew everything, mainly because meshers used to be pretty poor, as well as some CAD systems. Now days, if you don't have a setup that can mesh a fairly complex molded part, solve it, and get the right answer (converged) rather quickly, then you need to be out looking at a new system.
Might be fun to compare FEA systems too. There ARE some differences...
-meld
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.