Trouble converging in non-linear static analysis with contact

[Anonymous]

Hello, 

I am having trouble with a non-linear static analysis converging.

Setup-

Non-linear static analysis between two parts. One part is held stationary, the second is forced into the first part. 

Load-

Enforced motion in the x-axis of top surface of part 2. 

Constraints- 

1-Bottom surface from part 1 fully constrained

2-Top surface from part 2 fully constrained, except for translation in x-axis

3-Top surface from part 2 translation in x-axis constrained. 

Surface Contacts- 

Manual, surface to surface, symmetric contact. 

Mesh-

Mesh has been refined in all areas that contact will occur.  

I have tried- 

• Changing NLTOL to 3
• Modify ESPU, EPSP and EPSW individually to, 0.1, 0.1, and 0.001 respectively. 
• Set solve method to AUTO, SEMI, and ITER
• Changed stiffness factor from 1.5 to 10
• Changed the mesh around the contact surfaces
• Changed type of contact surface (symmetric/unsymmetric) 
• Changed MAXBIS to 50
• Changed MAXITER to 200

And in each case I am still getting an error E5076 for max number of bisections being reached... and in each case it is the load that will not converge. I have had the work and displacement converge, but I can not get the load to converge. 

After viewing partial increment solutions, it is appears that the two parts will penetrate each other slightly. It also appears as if the pieces are beginning to behave as you would expect, but be fore a full solution can be realized the max number of bisections limit is reached. I have also tried increasing the MABIS to 50, and the analysis still failed.

I am running out of ideas to try... does anyone have any ideas that might be keeping the load from converging?

Thanks,

Kyle 

[Andrew.Sartorelli]

Hi Kyle,

You mentioned that you were able to work and displacement to converge, if you turn off load convergence does the model solve to completion? Technically we only need two convergence criteria to have an accurate solution. As displacement and load would ultimately mean we have convergence of work. In this case because we have convergence of work and displacement, I think it would be OK to run the model with just these two criteria.

Regards,

Andrew

[Anonymous]

Hi Andrew, 

Thank you for the response. 

Not sure if I know how to turn "off" the convergence criteria for load, but my guess would be changing EPSP to a much higher value, say 1. Is this what you mean when? Or is there another way to turn off a specific convergence criteria? 

Thanks again, 

Kyle

[Andrew.Sartorelli]

Hi Kyle,

It sounds like you are using Autodesk Nastran and not Nastran In-CAD, so here are the details on performing this through modification of the bulk data file. You can control which convergence criteria is used through the CONV field on the NLPARM card, field 8 row 1. To set it to displacement and work you would enter a value of UW, U for displacement and W for work.

If you are using Nastran In-CAD, go into Nonlinear Setup and then select Advanced Settings. Under "Convergence criteria and error tolerances" only select the checkboxes for Displacement and Work.

Regards,
Andrew

[Anonymous]

Appreciate all your help Andrew-

I am using Nastran IN-CAD, I am just finding the most helpful information in the Nastran Reference Manual that was included with the install of IN-CAD. 

I tried running a few simulations as you suggested, to no avail. 

I am starting to wonder if the Nastran algorithm can handle this simulation. I would assume it could handle it, but other wise I am stumped as to why it can not solve. 

Kyle

[Andrew.Sartorelli]

Hi Kyle,

It might be easier if I have a chance to take a look at the model. I will reach out to you directly with instructions on how to share your model with me.

Regards,
Andrew

[Anonymous]

Dear Andrew,

Could you possibly share the answer to this problem if you found it, unfortunately i am facing very similar one. No matter what are my attempts and changes, i get error E5076 and warning E5074 during second time step, with convergence of work and displacement on 100.0 and load on around 0.3. Is it possibly hardware problem? No matter how i extend the number of maximum bisections, it is still not enough. 

Regards,

Mikolaj

[Andrew.Sartorelli]

Hi Mikolaj,

One of my colleagues was working with Kyle on his model, so I can't speak to the specifics for solving the problem for his model. There can be some situations where obtaining one of the convergence criteria is not viable. In your situation where you have work and displacement convergence, I would say it would be safe to proceed only with those two criteria.

In Nastran In-CAD go into Non-linear Setup, click Advanced Settings, and the click the boxes for Displacement and Work.

Regards,
Andrew