AnsweredAssumed Answered

Nonlinear Analysis Failed - Improper Contact Surfaces for a Contact Area

Question asked by Louise Yi on Feb 18, 2021
Latest reply on Feb 22, 2021 by Louise Yi

Like • Show 0 Likes0
Comment • 4

I am running a nonlinear analysis for latch/snap. The model and motion are pretty simple.

The analysis is not failed during the bending, but failed at where it's almost done. And the stress distribution is not even. As the screenshot below, it's concentrated at one side.

Sometimes it will pop out "Improper Contact Surfaces for a Contact Area", sometimes it just says it's failed.

I didn't find much info about this error message, so just wonder what's the reason for it?

Is it possible because of improper contact set setting, mesh size, or is it because of the steep edge of the latch?

If I add a 45 or larger degree tilt to the rear end it will succeed. But why vertical wall does not work?

I attached the model(SW2019) for reference. Thanks for any idea!

Ryan Navarro Feb 19, 2021 1:48 PM

Correct Answer

I did use Surface to Surface contact. You're right that the "conventional wisdom" suggests you would be better off with Node to Surface, but in practice when dealing with convergence issues such as this you have to go with whatever works. The surface to surface contact seemed to still resolve the contact the way I would have expected.

Adding even a very small edge break (chamfer or fillet) to those sharp edges can also help a solution like this, and often cases matches reality since most tooling etc has some kind of minor edge break on it vs a "perfect" knife edge

See the reply in context

Attachments

Nonlinear.zip175.2 KB

No one else had this question

Outcomes

Ryan Navarro Feb 19, 2021 9:10 AM

Hi Louise,

I was able to get it to solve completely with a few simple changes, unfortunately I am running a newer version of SOLIDWORKS so I don't think you'll be able to use the files

added a Split Line to isolate the contact areas
set contact to these new split faces
applied mesh control (default settings) on contact faces
adjusted the global mesh size to "coarse" to speed up the solution a bit
edited the study properties and set the max timestep to 0.05 instead of 0.1.

Reducing the max timestep value will make sure the solver doesn't take too large of a step which can goof up contact problems like this, in exchange for taking a little longer to solve.

That error message you received just means a general solver convergence problem. I think the issue here is mostly just the mesh, but also if you think about it the straight walled version will want to "snap" suddenly once it clears the other component.

Snapping or sudden movement doesn't always jive well in a Nonlinear Static analysis and can trigger the solver convergence problems. Adding friction to the contact can help in those situations, or setting up the same study as a Nonlinear Dynamic analysis which allows for those more sudden changes. That being said it didn't seem to be necessary in this case - just needed to increase the mesh density on the contact faces and reduce the solver timestep a little. The original mesh was much too coarse for accurate contact

Also, it's not always a "problem" if the Nonlinear solution doesn't solve all the way - as long as you get the data you need out of the partial solution. For this snap design, I think you'd likely already have all the data you need even in the "partial" solution you originally posted

You may also consider using a 2D simplification to speed these up and also make convergence easier

Actions

Louise Yi @ Ryan Navarro on Feb 19, 2021 1:05 PM

Hi Ryan,

Thank you so much! It's so kind of you to give a detailed explanation.

I adjusted the model as you suggested and reran the simulation. But it still failed as what it was. The mesh control size is 0.11mm as default.

I wonder if it's possible that you use contact set type as "Surface to Surface"? I've used this type before and it succeeded.

But on Soliworks Tutorial it says "Not recommended when the area of contact between a source face and target face becomes too small or reduces to a line or point during deformation."

I think when the latch leaves the container, the contact area will reduce to a line? That's why I come back to use Node to Surface contact type.

I haven't used 2D simplification or Nonlinear Dynamic Analysis before but I will learn how to use them as well.

Thanks again!

Actions

Ryan Navarro @ Louise Yi on Feb 19, 2021 1:48 PM

Actions

Louise Yi @ Ryan Navarro on Feb 22, 2021 4:18 PM

Thank you for the explanation! I've added R0.1 fillet to the retainer edge but it still fails. I guess node to surface is more difficult to solve under such a sudden drop. I may just use surface to surface as you suggested.

Actions

4 Replies

Ryan Navarro Feb 19, 2021 9:10 AM
Mark Correct
Correct Answer
Hi Louise,

I was able to get it to solve completely with a few simple changes, unfortunately I am running a newer version of SOLIDWORKS so I don't think you'll be able to use the files
- added a Split Line to isolate the contact areas
- set contact to these new split faces
- applied mesh control (default settings) on contact faces
- adjusted the global mesh size to "coarse" to speed up the solution a bit
- edited the study properties and set the max timestep to 0.05 instead of 0.1.
Reducing the max timestep value will make sure the solver doesn't take too large of a step which can goof up contact problems like this, in exchange for taking a little longer to solve.

That error message you received just means a general solver convergence problem. I think the issue here is mostly just the mesh, but also if you think about it the straight walled version will want to "snap" suddenly once it clears the other component.

Snapping or sudden movement doesn't always jive well in a Nonlinear Static analysis and can trigger the solver convergence problems. Adding friction to the contact can help in those situations, or setting up the same study as a Nonlinear Dynamic analysis which allows for those more sudden changes. That being said it didn't seem to be necessary in this case - just needed to increase the mesh density on the contact faces and reduce the solver timestep a little. The original mesh was much too coarse for accurate contact

Also, it's not always a "problem" if the Nonlinear solution doesn't solve all the way - as long as you get the data you need out of the partial solution. For this snap design, I think you'd likely already have all the data you need even in the "partial" solution you originally posted

You may also consider using a 2D simplification to speed these up and also make convergence easier
Like • Show 1 Like1
Actions
- Louise Yi @ Ryan Navarro on Feb 19, 2021 1:05 PM
  Mark Correct
  Correct Answer
  Hi Ryan,
  Thank you so much! It's so kind of you to give a detailed explanation.
  I adjusted the model as you suggested and reran the simulation. But it still failed as what it was. The mesh control size is 0.11mm as default.
  
  I wonder if it's possible that you use contact set type as "Surface to Surface"? I've used this type before and it succeeded.
  But on Soliworks Tutorial it says "Not recommended when the area of contact between a source face and target face becomes too small or reduces to a line or point during deformation."
  I think when the latch leaves the container, the contact area will reduce to a line? That's why I come back to use Node to Surface contact type.
  
  I haven't used 2D simplification or Nonlinear Dynamic Analysis before but I will learn how to use them as well.
  
  Thanks again!
  Like • Show 0 Likes0
  Actions
  - Ryan Navarro @ Louise Yi on Feb 19, 2021 1:48 PM
    Unmark Correct
    Correct Answer
    I did use Surface to Surface contact. You're right that the "conventional wisdom" suggests you would be better off with Node to Surface, but in practice when dealing with convergence issues such as this you have to go with whatever works. The surface to surface contact seemed to still resolve the contact the way I would have expected.
    
    Adding even a very small edge break (chamfer or fillet) to those sharp edges can also help a solution like this, and often cases matches reality since most tooling etc has some kind of minor edge break on it vs a "perfect" knife edge
    Like • Show 1 Like1
    Actions
    - Louise Yi @ Ryan Navarro on Feb 22, 2021 4:18 PM
      Mark Correct
      Correct Answer
      Thank you for the explanation! I've added R0.1 fillet to the retainer edge but it still fails. I guess node to surface is more difficult to solve under such a sudden drop. I may just use surface to surface as you suggested.
      Like • Show 0 Likes0
      Actions

Nonlinear Analysis Failed - Improper Contact Surfaces for a Contact Area

Attachments

Outcomes

Related Content

Recommended Content