Large deformation thermal expansion modeling issues

[AWooll]

Hello all,

I am new to this forum as well as Abaqus FEA. I am attempting to model a hydrogel like substance. It is a 3D model with the overall geometry being a simple 50mm long by 2mm tall by 0.2mm thick geometry. This overall geometry is composed of two halves (in which is the height at 1mm high is split parallel to the length direction). The two halves are different elastic materials TIED (constraint) together. The basic idea of my model is to apply a temperature change to the model and see the shape caused due to the two materials having different co-efficients of expansion. The final shape I am trying to create is largely deformed from the original and my solver aborts when the deformation gets too high (i.e. the sheet loops to almost a circle). I have tried a bunch of different solutions and small changes to the solver so to discuss all of them would be a waste of time to mention them all off the bat and instead divulge them as the thread continues.

The basic idea is applying only a temperature change to the whole body and encastring one end of the sheet and letting the relative strain between the two parts cause the deformation.

I would like help in choosing between:
-Dynamic/Explicit AND Static, General (which is more appropriate)
--I have non-linear geometry on for both
-Any changes to what elements or meshing type to use
-Any other tips to make it work

I am attaching the input file that has two models in it. One for the explicit and the other for the standard analysis.
Any help is greatly appreciated.

 

[rstupplebeen]

Are you modeling axisymmetric? Also, you can merge together your 2 solids retaining the boundary which will make the Tie unneeded. I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[AWooll]

WThanks for the reply,

What would the benefit in applying axisymmetry? Also, I have been looking into trying to make the two solids one but I do not know how as I am new to Abaqus. Is it done with a partition? Is it possible to merge and define each half as different materials.

 

[AWooll]

I was able to merge the parts. I am wondering what this will do by eliminating the tie? Is it for saving computational time or does it serve other purposes aswell.

 

[rstupplebeen]

Computational time and now that they are merged it is not needed. Axisymmetry reduces the degrees of freedom dramatically since many less elements and nodes are needed.

In your case their may be a weird buckling mode that would not be included in an axisymmetric case which will aid in computations. This is the only reason that I can think of not to use axisymmetric which would be if you are trying to model this phenomenon. You could probably still use quarter symmetry though.

I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[AWooll]

Thank you, the merger has helped. The wrinkling/buckling does occur in 3D when a dynamic/explicit model is used but not in a standard model. In 2D, neither models show the weird buckling. At the moment I am attempting to decide which model is correct though. I am trying to model close to what I will see through experimentation so I am afraid to ignore the buckling if this is what is actually caused. Also, my explicit and standard solver generate VERY different solution. I do not know which solution to trust. Do you have any input on which technique is best suited for my analysis type? Thank you once again for your help

 

[rstupplebeen]

I always start with using whatever symmetry available so in your case axisymmetric. Then if you find there are non-axisymmetric results use quarter symmetry.


I start with Standard for all non-impact problems. If inertia is needed to accurately predict results then Explicit is needed. Explicit also handles extremely large strains and sliding contact better.

If you re-post your model I will try to look at it next week.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[AWooll]

Thank you very much for your help so far. Here is a post of the model I am concerned with. It has only the Standard model input I made however I have an Dynamic/Explicit as well where only difference should lie in the element type and the swelling step I have defined (one is dynamic/explicit the other is standard). The results from both are very different. Again, it is difficult for me to decide which is more correct. Both displays certain geometries that could be expected; just my luck ha.

As a side note I also noticed that using reduced integration affected results drastically as well. Would you know why this is?