ANALYTICAL RIGID SURFACES AND DISTANT CONTACT DEFINITION (STATIC ANALYSIS)

[ivanpanzera]

Dears,

I'm trying to simulate a cold roll-forming process for a metal sheet. I started trying with static general analysis. In order to simplify the model and to avoid possible preliminary problems, I started considering only two rolls groups defined as rigid analytical surfaces.

https://res.cloudinary.com/engineering-com/image/upload/v1574873516/tips/Immagine_27_zjqgyu.bmp

After initial step, I have defined "interference fit" step in order to applying preliminary pressure between sheet and the rolls of the first stations. In a further step, I' ve appliyed rotation at each rolls and the first rolls group push forward the sheet efficiently thanks to the initial pressure.

https://res.cloudinary.com/engineering-com/image/upload/v1574873678/tips/Immagine_28_pmskrx.bmp

Problems occur following: interaction between sheet and second group of rolls doesn't work and pairs interpenetrate one another.

https://res.cloudinary.com/engineering-com/image/upload/v1574873874/tips/Immagine_29_zjf8ec.bmp

Interactions between sheet surfaces and analytical rigid surfaces are defined as "surface to surface" with tangetial penalty friction and normal hard contact. This is the same defition I used between first rolls group and sheet and it works correctly for these rolls. I think the problem could be the initial clearance between pairs.
This problem could disappeare if I use deformable definition for the rolls, but I want to mantain analytical rigid surfaces in order to reduce computational costs for my simulation.

How can I fix it?

Thanks in advantage,
Best regards,

Ivan

 

[FEA way]

Try using general contact instead of contact pairs. Also if it stil doesn’t work in Standard, consider switching to Explicit.

 

[Mustaine3]

Even if it does work in Standard, those kind of analysis, with permanent changes in contact, usually run much faster in Explicit. The user has just do some test runs to figure out a good value for Mass Scaling.

By the way, Surface-to-Surface and General Contact use both a penalty formulation. When contact forces become really high, the penalty stiffness might not be sufficient to prevent penetration. In such cases the user can scale up the penalty stiffness or switch to node-to-surface. But again, this would not be necessary in Explicit.

 

[ivanpanzera]

Dears,

Many thanks for your suggestions.

I think that static analysis is the most suitable for this kind of problem. Explicit analysis are often used but in many jobs I found it is used in quasi-static conditions because the effects of inertias do not affect the process in real cases.

I have just solved my problems switching to discrete rigid definition for the rolls and activate NLGeom within step settings.

The problem now is the enormous amount of time and memory needed for the complete roll-forming process (28 seconds, 13 group of rolls) .

During the process the sheet has contact only with two groups of rollers at a time, while the remaining 11 groups of rollers rotate uselessly. A possible solution is to separate the process into multiple analyzes or to activate during analysis from time to time only the rollers needed for roll-forming, avoiding having a complete model from start to finish.

Can you recommend one of these strategies or something else?

Thanks in advantage!

Best Regards,

Ivan Panzera

 

[FEA way]

You can perform quasi-static analysis in Abaqus/Explicit as well and introduce mass scaling or artificially speed up the event (not too much or inertia effects will affect the solution at some point). Thanks to these tricks you will be able to significantly reduce the simulation time. This is the typical approach to rolling simulations in Abaqus.