hexagon logo

Linear Contact

I'm trying to figure out a linear contact problem in SOL101. My model is a structure where some constraints are allowed to move up to a limit, after which they react like they have a linear spring applied to them. My solution cannot be in SOL400 due to customer requirements, and I have a permanent-glue contact which prevents me from using SOL106 and simply throwing gap elements at the problem.
 
I'm not a regular user of contact analysis, so a lot of my issues are learning-curve-related. However, I pulled down the tutorial on www.mscsoftware.com/.../steel-plate-contact to develop some familiarity before I apply the contact to my model.
 
The tutorial problem is not solving. Nastran gives me a fatal error:
 
 *** USER FATAL MESSAGE 4551 (nl3con)
   *** STOPPED PROBLEM DUE TO FAILED CONVERGENCE.
 
 *** JOB DOES NOT CONVERGE AT THE CURRENT TIME STEP OR INCREMENT.
 *** SOLUTION DIVERGES FOR SUBCASE   1 STEP   0.
 
Since the solution was using default parameters with 10 load steps, I increased to 40 load steps and kept getting the same error. Since I'm using a tutorial, I assume that the problem is well-posed and should solve if set up properly, so I can only assume I didn't set something up correctly when I sent it to Nastran.
 
Are there any ideas as to what I set up wrong, so that I can move forward on my analysis?
  • I'm not sure where you might have made a mistake, and I am not sure if those directions are 100% inclusive of everything you needed to do. I went through and made a model that works. I'll attach it. But, let me ask- when the displacements reach the certain point in your real situation, do things become 'rigidly' connected? Or do they hit a 'stopper'? If so, it might be simpler to use the SOL101 only Linear Gap capability. Read up on it here:
     
     

    Attached Files (1)
  • Thanks for the quick response!
     
    I found a few areas where your model file was different from mine - mostly in the BCTABL1 entries. I changed mine to match yours, and now it's telling me it can't resolve a penetration:
     
    *** USER FATAL MESSAGE 8135 (N3DSEP)
       MAXITR*10 iterations have been implemented to remove the penetration.
       Based on the current convergence rate, it is estimated that it may take more than 10000 additional iterations to remove the penetration.
     
    The only remaining difference which I can see is that you used quadratic tet elements whereas I used linear hex elements. The tutorial specified hex elements - which is why I used them - is that part of why I'm having issues?
     
    I was not read-up on the linear gap method, but it appears that may be the winner - I just created a quick model with two meshes separated by linear gaps, which did exactly what I need. I'll do away with the contact method, then, and just use linear gaps since those give me exactly what I want without the hassle of defining and managing contact pairs.
     
    Thanks!
  • should be no issues related to tet10 vs hex8... if you want to upload your version I can see what's going on... glad linear gaps seem to work!