Inflation simulation with bending

MIT aeroMorph project
Dear experts,
I am trying the inflation simulation with Kangroo2 recently, but the inflated cloth simply inflates without bending like what showed in the picture.
This project can be understood like this: they used 2 same clothes with TPU coating on a side facing each other, clothes melted and glued together in the heated area. And heating-welded areas can not be inflated.
In my algorithm, I added force to the mesh vertex and set the heat-welding area vertex into rigid points, but the result can only inflate without bending.
What is the mechanism behind this bending phenomenon and how to simulate it?
It has hazed me for months and my hair is getting thinner.
Hopefully you can help me. Regards!

This thread might be helpful

Hi Daniel, thanks for your reply. I have seen this before. However, It seems they did not realize the bending phenomenon. Or am I missing something?

the bending I mean is being folded like the crane in the picture.
or like the angled object in this picture

This one is more like what you describe.

I really appreciate this. Close but it still has some distance from my goal. In reality, the folding action is autonomous instead of an outer force like a hand.
I changed the file a bit: in my case the welded area is stiff so I used rigid points. The problem is that the bending or folding is not autonomous. The file is attached.
8.19 (21.4 KB)
hope you can have a look

My wild guess is that the folding action is due to unbalanced force( they possibly be balanced when in the same planar, but when the forces get a bit noise force the balance break . Under the influence of the forces they come into a new balance and that postion is angled.
I have no idea whether it is true and how it can be explained and presented in kangroo.

Yes - that’s in the paper you linked at the start - it describes that symmetry breaking coming from the welding process.
In the definition above it reproduces the geometry from the paper after you give it a little nudge in one direction to break the symmetry.
Adding the asymmetry of the welding process would be a bit more complex but I think doable.
I’m not going to do the whole thing for you though.

Actually it seems that it doesn’t even really need any nudge - the symmetry breaks naturally without further input, just the direction in which it buckles is random. (19.7 KB)
If you wanted to include something to make it buckle one way instead of another it could be done with a line of hinges on one side.


1 Like

Wow! That is truly amazing! You are my lifesaver ~ ~ Now I really get it!

1 Like

I get a new bottleneck: the calculation gets really slow if the mesh has many faces( in some cases the welded patterns are complicated ). Is there any way to speed up this calculation? Thanks a lot.

At the moment the main thing you can do is to choose the largest edge length for the the mesh that can capture the features, so it’s not too dense. Note that in the last file I shared this edge length was set as an absolute value, so if you make a shape many units across you’d end up with a very dense mesh unless you adjust it.
I am also working on some updates with better parallelization for mesh bending and stretching, but no promises about when that will be ready.

1 Like