How to have an unbreakable fractured chain

[thinkinmonkey]

Hi all,

I'm having some trouble using fractured objects in my scene.

This is the scene:

The little ball is path constrained to the circle and animated by keframe, I'm using it to simulate the wrecking ball machine (?).

Then we have the cable, I'm using a wire object that is pin constrained in the animated ball.

Then we have the chain, three imported objects that are pin constrained to the free tip of the cable. This is the chain:

I'm using a fractured object to import the chain in DOP Network.

So the problem is about the chain: it always breaks!

I raised min/max substeps and collisions passes in rigidbodysolver after reading the help, the volume seems having enough polygons for simulation, but it always break.

Any suggestions?

Thanks in advance.

fracturedChain.zip

[sam.h]

Try messing about with the settings on the RBD solver. Try cranking up contact passes and collision passes. Also shock propogation should maybe be set to the number of objects you have (number of links).

[thinkinmonkey]

Hi Sam,

as I said in the first post, I read the help, so I raised shock propagation and other parameters, but with no success.

Anyway, I'm using these parameters for the simulation:

Actually shock propagation is no. links + 1 = 16 instead of 15, just to be sure, but even with 15 it breaks.

I suppose fracture is not the good choice to simulate a link, I'll try importing all single rings and let's see what happens, meanwhile I'll try a little experiment with a fractured sphere to observe if a fractured object is designed to break even if GLUE STRENGTH is set to -1.

Maybe it's because I'm using intersecting holed object, i.e. linked rings, and they are in contrast with fracture philosophy of H that doesn't like that kind of objects, but it's just a theory.

[michael]

also check you mass settings...your chain links might be 1000kg each or something...

[Stalkerx777]

1) Check collision SDF.

2) Check objects mass (density)

3) Rise contact/subcontact passes

I don't know why, but sometimes rising substeps in RBD solver doesn't do the trick.In this case, try to rise substeps on DopNetwork node itself, this should help.

Edited December 8, 2011 by Stalkerx777

[thinkinmonkey]

Ok, based on all your suggestion, I made a little simple scene where I put high settings, more than I thought, and the little link (made by only 5 rings) never breaks.

Then I made the same with the main scene, but it's very slow to simulate, they are 15 rings and now also the wrecking ball is taken in account for the simulation, so I'm taking a new direction for scene/simulation.

This is the simple 5-rings link scene: simpleChainTest.hipnc

Again thanks for your help.

Edited December 8, 2011 by thinkinmonkey

[Fabiano Berlim]

Do you really need that to be RBD?

Because you can make it faster and reliable with wire object. It's not going to be as perfect as RBD. But if the chain is not your main character, you probably won't notice that.

wire_solver_chain.mov

chain.hip

[thinkinmonkey]

Fabiano,

thanks a lot, it's a great solution.

I thought something similar, but I changed my "scene", so I'm working with another kind of solution, let's say, very cheap.

But I examined your scene and I learnt something new about wire.

Thanks again.

[GallenWolf]

Hi! Are you still working on this? I did some tests on this, and it seems the solution for this seems to be increasing substeps. I've attached a sample scene here, that procedurally allows the addition of additional rings. On 5 rings, I was setting the substeps to 2/4, with 15 rings about 2/8 - maybe give that a go?

Also, it seems like it is highly dependant on the motion from the base chain - small slight motions don't seem to need substeps, but with the animation I've put on, needs far more.

EDIT: I did more more tests, and it seems like substepping is not the only thing required. Using a longer chain and vertical motion so that the chain stretches, even with 4/10, penetration still occurs. The resolution was with the Collision Passes parameter. I upped that, as well as set my substeps to 1/10, and that seemed to do the trick. There may be more to it but that's what I've got so far.

EDIT2:

HTHs!

GW

rbdChain.hip

Edited January 9, 2012 by GallenWolf

[thinkinmonkey]

Hi GallenWolf,

I'm still on it, it's a part of a little animation, actually I'm solving other things, but I count to be on chain animation soon.

Anyway, I saw your file and movie and it reacts very well, especially when headOfChain rotates very fast!

I'll take your config in consideration, but my need was a bit different. As you will see in the attached file, I pin-constrained the first ring of chain and I animated it very fast, the result is that ring, the first one, simply goes out from chain because the violent animation of the pinconstraint.

With my previous experiments, I tried to increase parameters for simulation, but with no success.

At the moment, the solution suggested by Fabiano seems to solve the problem in a straight and nice way, but I'd like to know if a physically calculated chain is possible with no compromise.

By the way, thanks for your time.

rbdChainWithPinConstraint.hipnc

[GallenWolf]

Hiya!

Just checked out your hip file, and indeed, the 2nd link keeps on falling off. I was testing purely on two links, and it was still falling through. One thing I noticed is that the simulation seems "off".... does not look correct. My gut tells me the pin constraint may not be receiving proper feedback, but that's a theory. Based on that however, I switched over to the rbd spring constraint, and it actually works there. You might want to switch the pin to a spring constraint and give it a go and see if it helps. Fun, this

I haven't seen Fabiano's setup, but I would think a single wire used for a chain (I've done this for chainmail) would be much more stable and way faster. Personally very curious to do a fully simulated chain though

GW

[Macha]

Constrain them? Applyrelationship dop is perfect for that sort of thing.

Edited January 12, 2012 by Macha