Contact analysis

mabefa

Hi, 

I have just set a non-linear quasi-static analysis with a contact suracface  but I'm non sure if the results I obtained are corrected. In the loadcollector with card NLPARM I created, I filled the NINC field with the value 20, but I did it haphazard. Is there a value I have to set which is the best or does it depends on the load I have in the analysis? the forces I have is about 40000 N. Should I increase the value in NINC or should I decerease it?

Thank you

Simon Križnik

Hi,

 

the number of load increments required for convergence depends on possible nonlinearities (contact, material, geometric). The default value of 10 NINC on NLPARM card is a good starting point, but will no be suitable for all problems. If analysis converged then there is no use for further increasing load increments. If unsure you can activate the nonlinear expert system to aid in the convergence of small displacement nonlinear problems with analysis>control cards>param>expernl>yes or CNTSTB. Description from Optistruct Help:

 

YES

Activates a nonlinear 'expert system' that aids in the convergence of small displacement nonlinear problems (NLSTAT). This version is designed to facilitate obtaining converged, high accuracy solutions without much concern for computational time. The system monitors the convergence of nonlinear processes and, if needed, implements measures designed to improve convergence for poorly converging cases. These measures include: performing additional iterations, under-relaxation, automatic adjustment of the load increment, as well as backing off to the last convergent solution and retrying. This may lead to a large number of nonlinear iterations for poorly converging problems.

 

CNTSTB

Additionally introduces temporary stabilization on contact interfaces (CONTACT or GAP(G) elements) that may improve nonlinear convergence, especially in cases where individual parts lack full support and are supported only by contact. The stabilization is applied only during incremental loading and is not present in the final solution for the respective nonlinear subcase. (Refer to the CNTSTB Bulk Data Entry for default and priority information).

 

mabefa

Thanks for the answer, so I have another question: how can I know if the analysis converged? I mean, if the analysis ends and the analysis' results are available to see in hyperview did the analysis automatically converge? Or the analysis just ended and the results may be available but completelly wrong even if  the Optistruct file .out says the job is completed?

Simon Križnik

If the analysis ends and the analysis' results are available it means analysis has converged. 

The basic principle in assessing nonlinear convergence is to compare an error measure of the solution with a pre-determined tolerance level. When the error falls below the specified tolerance, the problem is considered converged. In a case of multiple, simultaneous convergence criteria, all criteria need to be satisfied for the solution to be converged. Convergence monitors relative error in displacements, loads, work or any combination. These norms only measure the error of the nonlinear iterative process. Their values do not represent the accuracy of the finite element solution, only the fact that the nonlinear process has converged properly.

 

However, the results shouldn't be trusted blindly so verification and validation steps have to be taken before results can be given any credibility.

https://blog.altair.com/verification-and-validation-in-fea/

https://www.engineering.com/DesignSoftware/DesignSoftwareArticles/ArticleID/11356/Verifying-Your-Finite-Element-Analysis-Results.aspx

https://www.nafems.org/publications/browse_buy/browse_by_topic/qa/verification_and_validation/

mabefa

Thank you a lot for the explication.