Exploding RBD's in FLIP fluid

[Jebbel]

Hi Guys,

 

I need some expert advice and insights on this project I'm working on. I'm doing some RnD first before moving to bigger scale so I can provide you guys with the HIP file. The objects however (barrels and table pieces) I cannot share with you. But maybe there's something in the dop nodes or solvers that you can spot right away.

 

I'm working on a flood scene where there's a lot of debris and rubble that needs to be transported by the water. Please have a look at the video. The barrels seems to acquire a spin which at one point is out of control! I can imagine this could be the rotational stiffness...

but the exploding (and stuck) piece of table is still a mystery at the moment.

As you can see I'm also having trouble controlling the buoyancy of the objects in the fluid. Right now I use the density to try and float pieces but it's not entirely working the way I want it. To sum it up:
- The wooden planks need to float

- barrels need less spin and need to be embedded more in the fluid.
- the tables obviously need to stop launching like superman .

 

any advice or tips  are greatly appreciated .

High-RBD_count-in-fluid-006.hip

[tricecold]

I didnt check your file but just an idea, if the RBD pieces are bullet and packed , you can put a pop speed limit to automatically cull crazy max speeds, this usually solves a lot of explosive behaivour

[Jebbel]

That's interesting . This one's going in my notes!

I might've solved the problem btw.

I'm currently not using the parallel constraint solver but rather the single threaded option. In there, there's an option called "split impulse". Now... If I tweak the penetration threshold and its corresponding settings... Explosions seem to have almost disappeared. The problem in the end lies in the solver trying to separate objects which have collision penetration I think.

So...in theory; If I use the parallel constraint solver the issue should also go away. I'll report back here.

 

Thanks Tim! 

[Diego A Grimaldi]

Good suggestions by Tim.

 

For the floating problem you can tweak a parameter called Feedback Scale in the Flip Solver > Volume Motion > Solver.

 

By default this parm is set to 0 (idk why) but if you play with it you'll find the RBDs starting to interact and float over the water.

 

Good luck

[Jebbel]

Thank you all for the replies. much appreciated . I was aware of the feedback scale and using it in conjuction with density settings on my objects to make things float. Unfortunately it's still a bit like working in the dark but eventually I'll get there. 

I think I'm gonna try to implement that speed limit Tim was talking about as well. Sounds like good practice for this sort of sim .

[dyei nightmare]

On 17/7/2015 at 3:27 AM, Jebbel said:

Thank you all for the replies. much appreciated . I was aware of the feedback scale and using it in conjuction with density settings on my objects to make things float. Unfortunately it's still a bit like working in the dark but eventually I'll get there. 

I think I'm gonna try to implement that speed limit Tim was talking about as well. Sounds like good practice for this sort of sim .

hi there,  did you found the solution to this problem?

[Jebbel]

No... I ended up animating stuff by hand which was a pain the !#$@. The water was very turbulent and lighting very dark so there was some space to cheat. The Feedback scaling helped but I never really managed to fix this. 

[Jubuxyz]

On 7/17/2015 at 3:27 PM, Jebbel said:

Thank you all for the replies. much appreciated . I was aware of the feedback scale and using it in conjuction with density settings on my objects to make things float. Unfortunately it's still a bit like working in the dark but eventually I'll get there. 

I think I'm gonna try to implement that speed limit Tim was talking about as well. Sounds like good practice for this sort of sim .

great