3 Replies Latest reply on Mar 20, 2011 11:59 AM by Jerry Steiger


    James Denney

      Have a large assembly that's been giving me fits. Was meshing and solving fine until I made a change to the assembly a few weeks ago. Was having a heck of a time getting the assembly to mesh with the new parts. Even after removing the parts, I couldn't get the assembly to mesh in its original form. Spent several hours troubleshooting and just ended up having to rebuild the FEA version of my model. Got it meshing and executing the FEA solver. This was using SW2010 SP 4.0 on my old XP 64 bit machine.


      We did a new computer build to a WIN7 machine. I transferred the assembly to the new computer, and everything was meshing and solving quite nicely. I was happy with the performance increase with faster mesh times and solving times. I made a change to the model and introduced some new parts. The assembly meshed fine, however, during the FEA run, but I kept getting the "Status Code: 9" error messaged. Tried running the FEA with the intertia relief and soft springs flags turned on to no avail. Did some digging in the forums, and noticed my assembly was "under defined". I found a floating part in the assembly and constrained it so the assembly was now "fully defined". Re-ran the study, and still was getting the Status Code 9 error message. I removed the parts I added and still get the message. I went back to the original study (I had a few tabs of different studies looking at various loading conditions) and still get the message.


      I ran the SW cleanup tool to clean out any old temp files, including CHKDSK. I also deleted all the old studies, and created a new study. None of these actions appeared to have worked.


      Some things of note:

      1) That floating part may have been present all along...and yet it seemed to run successfully at times. No noted anomalies when looking at the displacement charts.

      2) Does it seem like the more you tweak with a large assembly, the more problems that seem to arise on the FEA side?


      I'd hate to pack-n-go the model every time I have a minor revision that I follow up with an FEA run, but it seems that once a study gets "corrupted" or whatever the case may be, it hoses all your previous studies.


      Any thoughts or similar experiences here?

        • Re: Status Code:9 MATRIX_SINGULAR_OR_INDEFINITE?
          Anthony Botting

          Yes - this happened to me a number of times when I mess with adding or removing new parts or fasteners, esp. It seems we must be very careful when adding components and remember to run SW interference check with "treat coincidence as interference" to ensure the body is indeed touching another (so the default bonded contact will work), and check all contact definitions to make sure they get applied to the correct faces. You are right. The "singular" message normally means the stiffness matrix is not invertible, i.e., there are rigid body modes or bodies that can move since they're not connected to anything.


          From experiences using the switches to artificially stiffen the matrix, most notably, the soft spring and inertia relief options seem to "take" best when using the Direct Sparse Solver. I have seen the same error message on FFEPlus (iterative) solver even with the switches on, but not if using the Direct Sparse Solver (as long as the database was not corrupted).


          Also, I have noticed a few times when I could not get it to go thru and turned in the issue to tech support, I get a response with the phrase "..your database was corrupted.." (and I think to myself - gee, I don't recall that EVER happening in a UNIX OS). Anyway, I and have resorted to creating a new study, dragging and dropping all the BC and materials and fasteners, etc.,  to the new one, and deleting the old one, save, close, exit SW, restart SW and everything seems to work.


          Another case (where the problem was not a corrupted database) and I was able to discover the moving body - run a frequency analysis. It will find the rigid body modes so that you can plot the displacements, undeformed, in color, for the mode shape(s) that have zero frequency. The red-colored parts are the ones that are not restrained or connected (by bonding or a fastener, say) to another part.

          Yes, tweaking a model with a large number of components seems a bit tricky, and between tweaks, I tend to run lots of intereference and clearance checks, and check all the fasteners and contacts each time, just to be sure they are defined on the surfaces where they are supposed to be.

          Hope that helps, if even a little.


            • Re: Status Code:9 MATRIX_SINGULAR_OR_INDEFINITE?
              James Denney

              Thanks! I appreciate your advice. I might be a victim of the Windows 7 administrator account issue with the iterative solver. The sparse solver worked fine...

              • Re: Status Code:9 MATRIX_SINGULAR_OR_INDEFINITE?
                Jerry Steiger

                Tony and James,


                Just so you won't feel too bad about the problems you are seeing with Simulation, I will tell you some of the problems I have using ANSYS. Almost every time I change my SW geometry and rerun a study, I have to check all of the materials and contacts to make sure that they are following the original definition. More often than not, materials seem to be randomly assigned. Sometimes contacts also seem to switch bodies. My simulations using involve just a handful of parts or bodies, but it can still take a while to sort through all the possibilities. I can hardly imagine the problems that would happen if I had hundreds of parts.


                On the other hand, I don't remember ever seeing the kind of problems you guys are facing. Choose your poison!


                Jerry Steiger