Wood RBD Sim Unstable with Constraints

[TheDude123]

Hi,

I've got a "wood" structure that needs destroying and I'm getting stuck on the simulation part. The sim looks good for about 3-4 seconds and then the fractured pieces start to move of their own accord (i.e they freak out).

I've fiddled with substeps (both DOPNET and bulletrbdsolver), collision geo, constraint iterations, etc. I watched countless tutorials and searched the web looking for answers, but nothing seems to work.

I just don't know where the issue is coming from.

Any help or guidance would be greatly appreciated.

Thanks!

 

Edited July 4, 2023 by TheDude123

[Benyee]

nice setup for wood fracture

I did some optimizations for your scene,see the marked nodes

for the rbdcluster node,you should check off the "fuse clustered pieces" since you have to make collision shape of each piece separately in dop,and so you should keep the original "name" attrib after clustering

for the constraints,I split the "glue" constraint into 2,one is cluster inner glue(glue constraint) and the other is glue between clusters(hard constraint,to make the entire wood structure a little bit softer)

and other modifications,for the bullet solver and setup for wood mass,etc,hope it helps

Wood_RnD_Example_File_v2.hip

 

 

[TheDude123]

On 9/23/2022 at 7:38 PM, Benyee said:

nice setup for wood fracture

I did some optimizations for your scene,see the marked nodes

for the rbdcluster node,you should check off the "fuse clustered pieces" since you have to make collision shape of each piece separately in dop,and so you should keep the original "name" attrib after clustering

for the constraints,I split the "glue" constraint into 2,one is cluster inner glue(glue constraint) and the other is glue between clusters(hard constraint,to make the entire wood structure a little bit softer)

and other modifications,for the bullet solver and setup for wood mass,etc,hope it helps

Wood_RnD_Example_File_v2.hip

 

 

Hi Benyee,

Thanks so much for taking a look. I went through your scene and cached it out, but it looks like the problem I'm having is still there (the pieces start to float around and have a life of their own after some time).

Again, thanks for trying.

[islamoos]

On 9/28/2022 at 8:40 PM, TheDude123 said:

Hi Benyee,

Thanks so much for taking a look. I went through your scene and cached it out, but it looks like the problem I'm having is still there (the pieces start to float around and have a life of their own after some time).

Again, thanks for trying.

Sounds like your problem is not having the correct name attribute on your pieces.
For instance if you have two fractured objects each containing 10 pieces, you have to make sure when you assemble to change the 'piece' prefix to the name attribute to something custom. Otherwise you have two pieces with the same name and the solver doesn't know how to evaluate them.
Cheers

Edited January 24, 2023 by islamoos

[TheDude123]

On 1/24/2023 at 8:48 AM, islamoos said:

Sounds like your problem is not having the correct name attribute on your pieces.
For instance if you have two fractured objects each containing 10 pieces, you have to make sure when you assemble to change the 'piece' prefix to the name attribute to something custom. Otherwise you have two pieces with the same name and the solver doesn't know how to evaluate them.
Cheers

Hi islamoos,

The names should be correct. Feel free to check the file. I'm pretty sure it's an issue with the bullet solver and excess energy being stored in the sim.

[islamoos]

On 2/24/2023 at 9:56 PM, TheDude123 said:

Hi islamoos,

The names should be correct. Feel free to check the file. I'm pretty sure it's an issue with the bullet solver and excess energy being stored in the sim.

Hey, I couldn't find your original file so I just modified the file posted by Benyee.
You are definitely right, the movement you describe comes from Hard constraints.
The solution is to introduce bending in a SOP solver so the constraints don't return to their original state every time they bounce.
There are also a couple of other notes I have: There was a setup in there trying to break Glue constraints in a SOP solver via an angle attrib. which won't work because they by definition break if they experience a change in angle of any kind.
There was a setup that connected the Hard constraints to the floor where by design some of the constraints didn't have correct name attributes, this can lead to trouble down the line.
I made the clusters non-breakable and changed the way adjacent pieces are connected. You can change it back if you wish.
For high-res wood fracturing this system might not produce desired results in that case (IMO) the best approach is a combination of glue and bending hard constraints.
I just simmed 240 frames and there is no floating behavior, let me know if it works for you :)
Cheers

Wood_RnD_Example_File_v2.hiplc

[TheDude123]

On 2/27/2023 at 4:00 AM, islamoos said:

Hey, I couldn't find your original file so I just modified the file posted by Benyee.
You are definitely right, the movement you describe comes from Hard constraints.
The solution is to introduce bending in a SOP solver so the constraints don't return to their original state every time they bounce.
There are also a couple of other notes I have: There was a setup in there trying to break Glue constraints in a SOP solver via an angle attrib. which won't work because they by definition break if they experience a change in angle of any kind.
There was a setup that connected the Hard constraints to the floor where by design some of the constraints didn't have correct name attributes, this can lead to trouble down the line.
I made the clusters non-breakable and changed the way adjacent pieces are connected. You can change it back if you wish.
For high-res wood fracturing this system might not produce desired results in that case (IMO) the best approach is a combination of glue and bending hard constraints.
I just simmed 240 frames and there is no floating behavior, let me know if it works for you
Cheers

Wood_RnD_Example_File_v2.hiplc 1.4 MB · 6 downloads

Hi islamoos,

Dude, thanks so much for taking a look! I forgot that I had to remove the original file, sorry about that.

Re: Glue constraints in the SOP solver breaking via @angle attrib...Ahhhh, I can't believe I missed that. Good catch!

Yep, that definitely looks like it's taking care of the floating issue.

Thanks again!