Thermal model error

I'm working on a thermal model using CAELinux LiveCD 2008, and I'm currently stumped. I've got a mesh, and I'm using the wizard to setup my .comm file. When I solve, it gives me an error message, and doesn't produce the results that it would produce if it had been sucessful.

From the results file, I see an error message, I put it into google.com/translate, and it notes a near 0 value at a node. I've tried changing several parameters, but with out luck.

Here's a snippet of the error message I get.

---

!
! <F> <FACTOR_21> !
! !
! !
! Matrice non factorisable : !
! pivot vraiment nul à la ligne : 14590 !
! pour le noeud N13358 et la composante TEMP !
! !
! !
! !
! Cette erreur est fatale. Le code s'arrete. !


After I run it through google.com/translate I get this.

Matrix not factorise:
really central to the zero line: 14590
for node and component N13358 TEMP

However, I don't know what the means. I've done the tutorials and they worked just fine. Well after that space in the name issue was figured out, it was fine. So I don't think the problem is with my system, I'm think it's part of my model file.

Does anyone have any suggestions on where to start looking for a answer?

I plan on posting the file when I get home. It appears it failed upload so I guess I'll try again in a seperate post. For now I guess I'll just have to post a question.<br /><br />Post edited by: Jared Harvey, at: 2009/02/12 23:15

Here we go, this link should allow the file to be downloaded if someone wants to take a look at it.

www.nedharvey.com/jaredharvey/southworth/study.zip

I should probably delete the below attempts to upload the file.

Lets see if I can get the file uploaded. Should be attached.

Post edited by: Jared Harvey, at: 2009/02/13 02:15

Didn't work, it's over the file size. I'll try an alternate approach in a couple minutes.

Post edited by: Jared Harvey, at: 2009/02/13 02:17<br /><br />Post edited by: Jared Harvey, at: 2009/02/13 03:19

to complete the answer of Kees:

correspondences mechanical &lt;-&gt; thermal

stiffnes matrix &lt;-&gt; heat conduction matrix
displacement BC &lt;-&gt; temperature BC

This means, that You have not enough temperature constraints in Your thermal model. You must define a fixed Temperature at (at least) one node.

good luck, Johannes

Great and thanks.

I seem to be able to get it to work on just a part, and I'm still testing if it could be an assembly issue. Perhaps there is a boundary issue I'm not seeing. Here's a quick overview of the steps I'm taking.

3D Modeled assembly then exported as step 203 file.
--- geometry --&gt; import --&gt; select step file --&gt; Create at least 2 groups
--- mesh --&gt; create new mesh --&gt; tetra w/ 1D avg length = 25 --&gt; compute
--- aster --&gt; aster thermal wizard --&gt; 3D, geometry, lambda 1, set two faces with different temperatures, no flux, create comm file
--- if successful, check plot to see it looks about right.

I've tested the above with a step file that is only a part, and it all seems good. I get some pretty pictures at the end. I've had my problems on the assemblies, and I'm working on making a simple 2 part assembly to try and keep it simple.

Perhaps I need to some how fuse the parts before I run the mesh? All the tutorials I've seen, with imported parts, have been single parts, not assembly step files. I've ensured that all the parts touch each other and do not interfere with each other. Perhaps there parts aren't fused and that's what registers as an undefined boundary?