FEM with CATIA V5 - Error: Singularity detected

[Justyna3087]

This is whole error message:

Factorization matrix computation.
Singularity detected (relative pivot too small)
Possible reasons: missing restraint or connection specifications.
Display deformation or displacement vectors to diagnostic the problem.

I have restraints where I need to, mesh is OK, 2D and 1D properities are OK (I supose), Smooth Parts are OK, so what can I check more? I have displayed 'Deformation' and 'Displacement' from toolbar 'Image' but it shows nothing or I don't know how to use it.

Please, can anyone help me? What can I do more to solve this problem.

 

[rb1957]

"I have restraints where I need to" ... are you reacting all 6 degrees of freedom ? your model may need constraints that aren't real but necessary for your model to solve it's math.

 

[Justyna3087]

Can you give me example of constraint that isn't real? In my model I haven't released any DOF. Maybe this is wrong.

 

[rb1957]

not sure we're on the same page ...

say you model a simple beam. doing hand calcs, you know it's a 2D problem. but the FEA doesn't, so it worries about the out-of-plane load and beam torsion. so for FEA you have to constrain the out-of-plane forces and the beam torsion even though nothing's happening (zero load, zero reaction).

but your two statements "I have restraints where I need to" and "In my model I haven't released any DOF" confuse me. also not sure you answered my question, are you reacting all 6 degrees of freedom ?

 

[franc11HD]

Justyna3087,

When viewing the displacement result, click on the scale bytton and put something like 0.00000001. Then you should be able to see your part with vector display to show a displacement vector on each node. Then increase the scale a little bit and you should be able to see how the part is moving and which DOF needs to be constraints. As rb1957 what saying your model needs to at least be constrained in all 6 DOF to avoid rigid body motion of the part.

Try to select the "clamp" restraint and apply it on your model to see if it get rid of the singularity.

Another thing that I found problematic with 1D model,s is sometimes you put a restraint on a node but in fact, you select a point that is not linked to the node. Therefore, no restraints on your FEM.

 

[johnhors]

Do a natural frequency modal analysis. If the model has any unrestrained rigid body freedoms, then these will appear as zero frequency rigid body modes. From displaying these it should then be obvious where the model has insufficient supports.

http://www.Roshaz.com

quality, cost effective FEA solutions