Element Size-Any effect to solution other than accuracy? 2

[huiying]

Hi,
Would like to know as in my previous qn if element size has a part to play in the solution other than accuracy. This is because after reducing the element size (to get a better accuracy), i'm unable to get a solution (i.e. error msg in the .f06 file). However, it was ok when i was using a coarser mesh. Pls help. Thanks.

Regards.

 

[pjhype]

What is the error message?

pj

 

[huiying]

This is the error i've got...


^^^ USER FATAL MESSAGE 9050 (SEKRRS)
^^^ RUN TERMINATED DUE TO EXCESSIVE PIVOT RATIOS IN MATRIX KLL.
^^^ USER ACTION: CONSTRAIN MECHANISMS WITH SPCI OR SUPORTI ENTRIES OR SPECIFY PARAM,BAILOUT,-1 TO
CONTINUE THE RUN WITH MECHANISMS.

 

[rjw57]

Looks like you lost model constraints when you changed the mesh size. I believe this is telling you that you have a mechanism (free body motion). Constrain the model properly or set AUTOSPC to ON (I recommend the former over the latter).

 

[huiying]

Thanks, rjw57.
A few qns...
1. How is it possible to lose constraints when mesh size is changed? Is it something to do with the relation between the constraint and the elements. How do I check whether it is lost?

2. I kept the constraints to the same, only changing the mesh size, how do I constraint it properly? Delete the constraints and apply them again?

3. How to check for free body motion before actual running the whole analysis?

Regards.

 

[pjhype]

You probably deleted the elements but kept the nodes with the constraints. These nodes are not necessarily connected to the model, resulting in the error. If the nodes of the refined model match the nodes of the coarse model, you could perform a coincident node check and keep the constraints. Alternately, you could delete all of the constraints, delete the extra nodes, and re-apply the constraints. You generally cannot check for rigid-body motion during pre-processing.

pj

 

[rjw57]

huiying,

If you have modeled the constraints to be nodal based rather than geometry based, you will most likely lose the relationship between the node and constrain when you remesh. Not sure what pre you are using, but FEMAP and others have geometry based constraints which stay in place regardless of mesh mods. Yes, delete your old constraints and reapply. See if geometry based constraints can be applied and use them when possible. As for RBM, you may be able to use your preprocessor to check the maximum separation of constraints in your model. If any constraint spacing is zero in whichever coordinate system you are using, you may have inadequate constraints. You can also do a modal analysis to see if you have an ~ zero frequency mode. If so, there is a good chance you may have a constraint issue.

rjw57