Help with Press-Out Analysis using DTRAN timestep solution after a NLSTAT Shrink fit solution

The_Ghost

Greetings,

I am having a problem with a press-out analysis. I am using version 2022. The setup is as follows:

Objective

Preform a press out analysis of an aluminum bushing that has been assembled into an aluminum machined casting. I wish to obtain the force in which the bushing begins to slide and break free.

Two Load Case Setup

Load Step 1- Shrink fit the bushing into the casting with a given interference fit. The bushing is already in place, just the contacts need to be equivalenced and create the condition of a bushing that was press fit into the casting.

Load Step 2 - Perform an enforced displacement on the bushing to start generating a reaction force to determine when the load ramp starts to flatten out. Basically, finding the knee bend in the force deflection curve. This is the force that I am most interested in.

NLSTAT analysis for both Cases

Load Step 1 - This solves with no issues. I can get the initial shrink fit to work using a friction constant for the contact.

Load Step 2 - I apply the enforced displacement and get reasonable results. The solutions work with both large displacements turned on and off. However, this is still a quasi-static solution and not truly time dependent. I want to push out the bushing over a 2s time period.

NLSTAT for the Press fit and DTRAN for the Enforced Displacement

Load Step 1- The first solution works with NLSTAT and solves with no issues. I can get the initial shrink fit to work using a friction constant for the contact.

Load Step 2- I applied the enforced displacement, using a DTRAN load step, and the model will not converge. I have tried multiple combinations of options, time steps, variable changes, contact changes, etc. The most I can get is the first timestep to solve. When I post the results, the elements are distorted. It has to be the solver, trying to deal with the geometric non-linearity of the contact and sticking during the transient solution movement over a real timestep. I do this type of advanced nonlinear analysis in Nastran or Abaqus with no issues and get good results with basically the same model. I did try to run in version 2025 with no success.

This is a fairly straight forward analysis, and I just would like to be able to perform this type of solution directly within Opti struct without using those other solvers. Any thoughts, suggestions, or different approach would be appreciated.

Adriano_Koga

what are the options that you're using for this model?

Maybe changing some options using NLADAPT could help.

In newer versions maybe trying to use NLCTRL instead of NLPARM.

Also what are the options you're using in your contact? Any Contact Stabilization?

Any PARAM?