Nonlinear Structural Analysis with Finite Element Method

[Hassan2]

TL;DR Summary

Hello everyone.

I need advice from anyone who is experienced with numerical methods of nonlinear structural analysis.

I have several years of experience in developing codes for linear and nonlinear magnetic field analysis. I have some experience with linear structural analysis too. Now I am being asked to develop a commercial code for nonlinear structural analysis but the topic looks complex and frightening!

Unlike in electromagnetics, the nonlinearity of mechanical structures is not only due to the material property. It could be due to large deformation and contact as well. Even though I will be implementing the existing and popular methods, I am still scared and I feel it is not a job that can be done in less than 6 months. Do you have any comments or suggestions?
I have seen some papers and online sources on such topics. Do you know of any comprehensive book that covers the details?

Thank you very much in advance.

Hassan

 

[jrmichler]

I used ABAQUS to analyze a range of problems involving large deformations and contact, and it took me a couple months just to learn to properly apply it to my particular problem. The topic looks complex because it is. You need to deal with:

Elastic-plastic material models, including strain hardening and strain rate hardening.
The element(s) must respond properly to large deformations, including deformation where the element is subject to bending.
Contact between surfaces that are initially in contact.
Contact between surfaces that come into contact due to deformation.
Will you model contact only at nodes, or at surfaces?
Friction.
Large deformations may require remeshing. You need built in checks for this.
Provision for element damping to prevent diverging oscillations.
Tensile failure.
And more, that is just what I can think of in a few minutes.

It's been a while, but the ABAQUS manuals that I used were the best software manuals I have ever seen. It's worth reading them just to learn what inputs are needed to solve your class of problems. My recollection is that they had a whole book on their elements, with some detail on the element formulations.

I think that your six month estimate is very low. Six years might be better, and that only if you are careful to limit the range of problems that software should handle. Good luck, because you will need it.

 

[Hassan2]

Thank you very much indeed. Your advice clarified many aspects of the job. I need to resist my manager's ambitions then.