Hyperelastic simulation does not run due to element distortion

  • 55 Views
  • Last Post 21 May 2020
LewisJ posted this 21 April 2020

Hello!

I have been trying to run a simulation on a relatively simple geometry all day. The simulation uses the Neoprene Rubber material (hyperelastic) and is intended to model the 'flipping' of a dome inside out. The model is a surface model, with 0.5 mm thickness. The only boundary condition aside from pressure (~9 Pa applied) is a fixed support. 

I have attempted the following things, but cannot acheive convergance due to 'excessive distortion of elements'

- Decreasing mesh size - 0.8 mm yields optimal results, higher or lower causes the simulation to terminate earlier. 

- weak spings/Energy stabilisation - no change 

- substeps: improved result up to a point. I tried increasing initial and minimum substeps. No results past 500 initial substeps

- Low order triangular mesh - doesn't help, similar result

Please the .wbpz archive here: https://polybox.ethz.ch/index.php/s/MY4VkCvr5vJ819N

Thank you! 

Lewis 

 

I would really appreciate help, and for someone to point out where I am going wrong. Thank you! 

Lewis

Order By: Standard | Newest | Votes
LewisJ posted this 21 April 2020

Wenlong posted this 21 April 2020

Hi Lewis,

I would suggest turn on the nonlinear stabilization since this structure has a very weak out-of-plane stiffness and easy to buckle. Maybe start with a small energy stabilization factor like 1e-3 and slowly increase it if it doesn't help. 

Regards,

Wenlong

 

  • Liked by
  • LewisJ
LewisJ posted this 01 May 2020

Hi Wenlong, 

Thank you for your advice. The simulation now works perfectly! I had to go up to a energy stabilisation of 0.5. 

Thanks for helping!

Best,

Lewis

Wenlong posted this 01 May 2020

Hi Lewis,

0.5 is a pretty high value for most simulations (I haven't seen any simulations running energy stabilization higher than 0.1). Basically what it does it to add artificial damping to the model to prevent sudden instability like buckling, but that artificial energy added to the system has to be small. 

Please add a stabilization energy plot and compare it to the strain energy plot. The stabilization energy has to be way smaller (you may use 5% as a reference) than the strain energy to make the solution accurate. 

Regards,

Wenlong

 

  

LewisJ posted this 20 May 2020

Dear Wenlong, 

 

Thank you for your detailed replies. I took a look into the Asnys Stabilisation feature and I am beginning to understand it more. 

I moved on from the last simulation, but I am facing a similar challenge (similar buckling analysis). This time I tried both energy stabilisation and dampening, however I cannot get the solution to converge past a certain point. I also tried to redefine my mesh (with symetrical surface splitting) but none of my attempts have yielded any improvement. The furthest I got was by splitting the load into multiple steps, with a high number of subseps. 

https://polybox.ethz.ch/index.php/f/1896594952

Could you please point me in the right direction to getting this to solve?

 

Thanks,

Lewis

Wenlong posted this 21 May 2020

Hi Lewis,

Since this is a different simulation, and in order to get the attention of a bigger audience, please post it in a separate thread. Please note that Ansys staff cannot download attachments from the student community, so please insert images as you did before. 

Thanks!

Regards,

Wenlong

 

Close