sharp penetration of a soft solid (contact & stress singularity) 1

[Aislingoh]

Hi all,

I am trying to model a sharp knife penetrating skin. I have some ideas about how to do this but I would invite any comments from all you experts on my method at this early stage.

I am using explicit. The knife is modelled as rigid body and is driven via boundary conditions. The skin is modelled as a hyperelastic material.

1. What is the best type of interaction to use? currently I am using general contact explicit and it seems to work ok. but perhaps surface-to-surface contact would be better? The abaqus example of a projectile penetrating a plate uses surface-to-surface contact but I found general contact solves quicker and appears to do the job better too.

2. I do not intend to model the actual damage. What I plan to do is to use a failure criterion (such as Von mises) along with element deletion. I am using the submodelling technique to get a higher mesh density in the region of interest, without sacrificing efficiency. Currently the geometry of the blade tip is modelled as a point (no tip radius). Will this lead to stress singularity and if so, will that matter since I am using element deletion anyway? The alternative is to introduce a small tip radius (around 1micron), but in this case I will need a very fine mesh about the blade tip to capture this geometry. And since it's a rigid body anyway, it seems like a waste of time and effort!

Please feel free to comment on any aspects of this, I would love to hear the opinion of the community.

Thanks,

Aisling

 

[rstupplebeen]

Instead of element deletion you could look at it as crack propagation? Is there a reason for starting in Explicit? Are there dynamics involved? A more detailed explanation of your problem will help with focused suggestions. I hope this helps.

Rob Stupplebeen

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

 

[Aislingoh]

I'm not sure what addition details would be helpful. The simulation is intended to eventually model a stabbing scenario (yes, stabbing!). But for now i want to drive a knife through a thin piece of skin that is encastered around the edges. I will try attach an image.

I understood that I would be simpler to implement element deletion instead of crack prop. What would be the benefit of using the crack propagation approach?

I am starting off with quasi-static speeds but I want to increase the rate of displacement of the blade up to 10m/s. Also I was under the impression that explicit can model contact better.

 

[Aislingoh]

here's a second image if it is still a bit difficult to visualise the problem with the first.

 

[MechIrl]

I think element deletion is a good start for what you're trying, at least it's the most straight-forward approach. A big down-side is that your solution will be highly mesh-dependent. If you are using element deletion I would suggest you do a careful study on how changing the mesh density can change your result, from my experience with it (ductile metal fracture) it can make a significant difference. At the moment your mesh looks quite coarse for this sort of problem (especially since you are using linear elements).

Using a crack propagation approach would help with the mesh dependence problem e.g. (XFEM) but would require use of the implicit solver (which can also be used to solve 'dynamic' problems to some degree)and may not be so straightforward to implement in your case.

For contact I think the explicit solver is better for your application (at least for the Abaqus version you are using which doesn't handle edge to surface contact well in Standard, its improved in later versions). I'd say stick with the general contact approach unless you have specific reasons to work with contact pairs, which I don't think you should for you application.

 

[Aislingoh]

Thanks for your help MechIrl, that's the advice I wanted to hear! The mesh is really coarse for now, but I have a much finer submodel around the contact area (not shown).

My next challenge is to get the element deletion working effectively. Can I still that using the general contact approach?

 

[MechIrl]

Yes general contact works fine with element deletion. Just remember to also include internal element surfaces in the contact domain so that when outer elements are deleted the blade can come into contact with inner element surfaces too. There's details for how to do that in the manuals (section called 'modeling surface erosion').