COMSOL: Newton's Iteration reached - Stress testing sensor

Click For Summary

Discussion Overview

The discussion revolves around troubleshooting issues encountered in a COMSOL simulation involving a sensor with a fixed surface and prescribed displacement. Participants explore convergence problems, particularly related to the use of hyperelastic materials and contact conditions in the model.

Discussion Character

  • Technical explanation
  • Debate/contested

Main Points Raised

  • One participant reports persistent convergence issues despite increasing iterations, refining the mesh, and applying parametric studies, along with a warning about MUMPS out of core.
  • Several participants inquire about the connection method of the parts and the use of contact in the analysis, suggesting that contact may complicate convergence.
  • It is noted that the parts are joined using a form union, with plans to implement contact later.
  • Concerns are raised about the presence of other forms of nonlinearity in the model, specifically questioning whether geometric nonlinearity is enabled and what material model is being used.
  • A participant mentions that the model converges when set as a linear material but fails with a hyperelastic material, indicating that nonlinearity is enabled.
  • Another participant suggests that the hyperelastic material might be unstable within a certain strain range and recommends trying a lower load or a different hyperelastic model, while also asking about the source of data for material constants.

Areas of Agreement / Disagreement

Participants express differing views on the best approach to resolve the convergence issues, particularly regarding the use of hyperelastic materials and contact conditions. The discussion remains unresolved with multiple competing perspectives on how to proceed.

Contextual Notes

Limitations include potential instability of hyperelastic materials within specific strain ranges and the impact of contact conditions on convergence. The discussion does not clarify the definitions or assumptions regarding the material models used.

uppityd
Messages
5
Reaction score
0
It is a sensor with a fixed surface and a prescribed displacement on the opposite

To remove this message error,

I have tried to
  1. Increase iterations
  2. Refined mesh
  3. Applied parametric studies
  4. And some other ideas from forums
However it still cannot be solved

It also has a warning of MUMPS out of core
 

Attachments

  • sensor.PNG
    sensor.PNG
    16.9 KB · Views: 153
Engineering news on Phys.org
How are these parts connected ? Do you use contact in this analysis ?
 
FEAnalyst said:
How are these parts connected ? Do you use contact in this analysis ?
They are joined using form union, i would like to put contact between the sensors together in later.
 
uppityd said:
They are joined using form union, i would like to put contact between the sensors together in later.
Right, it’s better to avoid contact until the analysis converges with Form Union. Especially that contact in Comsol is very susceptible to convergence difficulties.

Contact is not used here but there are other forma of nonlinearity in this model, right ? Do you have the geometric nonlinearity enabled ? What about the material model - which one do you use ?
 
FEAnalyst said:
Right, it’s better to avoid contact until the analysis converges with Form Union. Especially that contact in Comsol is very susceptible to convergence difficulties.

Contact is not used here but there are other forma of nonlinearity in this model, right ? Do you have the geometric nonlinearity enabled ? What about the material model - which one do you use ?
Sorry, I should have specified a lot better.

The main reason is that the block underneath the spherical disks is a hyperelastic material. If the whole model were to be set as a linear material then it converges, however when set as a hyperelastic material, the block underneath, then it does not.Non linearity is enabled.
 
The hyperelastic material might be unstable in a given strain range. Try with a lower load and maybe a different hyperelastic model. What is the source of data for material constants in this case ?
 

Similar threads

  • · Replies 9 ·
Replies
9
Views
7K
  • · Replies 11 ·
Replies
11
Views
4K
  • · Replies 463 ·
16
Replies
463
Views
74K
  • · Replies 6 ·
Replies
6
Views
6K