Rank Deficient 1

[dfanfan]

I am running an extensive FEMAP model and I keep getting the following errors:

*** USER WARNING MESSAGE 6137 (UDSNRD)
INPUT MATRIX IS RANK DEFICIENT, RANK = 12720 IERROR
USER ACTION: CHECK MODEL
User information:
One of your matrices is singular. See the NX NASTRAN Numerical
Methods User's Guide. for a discussion of singularity.
*** USER FATAL MESSAGE 6202 (MCE1)
THE SELECTED MULTIPOINT CONSTRAINT SET AND/OR RIGID ELEMENTS PRODUCE A SINGULAR RMM MATRIX. THIS MAY BE CAUSED BY A
CIRCULAR DEPENDENCY IN WHICH A DEGREE OF FREEDOM IS INDIRECTLY DEPENDENT UPON ITSELF.
*** USER FATAL MESSAGE 3005 (MCE1)
ATTEMPT TO OPERATE ON THE SINGULAR MATRIX SCRATCH IN SUBROUTINE MCE1
User information:
Subroutine SDCOMP or subroutine UDCOMP has detected a singular
matrix and the calling routine does not support this case.
A User Information Message defining the singularity has already
been printed.
0FATAL ERROR
1

It is telling me I have a singularity somewhere. Can anybody help on how to find this singularity in a quick, easy and efficient manner? Any suggestions are welcome.

 

[dfanfan]

I figured it out everyone. My model was not fully connected therefore, producing singularities. Thanks for all your help.

 

[dfanfan]

I take that back. It didn't work. Same error but now the rank = 13848

 

[dfanfan]

PLEASE HELP ANYONE!

 

[rb1957]

judging from your posts, the singularity in your structure has moved ... well ok, it's complaining about a different one ... are you printing the f06 file ? that just might help identify the culprite.

good luck !

 

[dfanfan]

Yes, this error was pasted from the f06 file. Just to give you a little background information, I am taking a solid model and created midsurfaces on all the parts. I meshed all the midsurfaces and connected the panels with rigid elements. The model is about 30 thousand nodes and 30 thousand elements. FEMAP should be able to handle a file this big. I did this methodology for two other models and it worked fine. Previous results were good. But this one has me totally stumped. The rank of the stiffness matrix? How can I find the singularity if it is not identifying where it is?

 

[rb1957]

i'd call the nastran help desk ... they were always very helpful when i needed help. i think there is enough information in the f06 but you need more than you said and i can't remember what it looks like. i don't think it's as helpfull as "dude, you've meesed up at node 1234 in freedom 4"

good luck !

 

[dfanfan]

Thanks rb1957

 

[ncho]

Check the connections of your rigids, they may be doubly connected meaning a node may be a slave to two different master nodes or you have a connection where two rigids have been connected to two nodes and one nodes master is the others slave. Either way the problem lies with the rigid elements.

-Nathan

 

[BlasMolero]

Dear dfanfan,
Simply edit the nastran input deck and in the bulk data section write "param,bailout,-1" and solve the model. If you have a singular matrix because your model is not properly constrained (you have rigid body motions) then the nx nastran solver will run with sucess. Later go to postprocessor and see displacements results, surely you will see values of meters o km, not problem, the important is that you see where the error is, you locate this way the piece that is unconstrained, simply define constrains properly and launch the solver again, and now I hope with success.

Let me know if this helps to solve your problem.
Best regards,
Blas.