Particles go through geometry

[DragonsSpirit]

I have made a scene once in RealFlow 4 ( http://www.youtube.com/watch?v=Z4DU6CODktY ) .

Because RealFlow 4 consequently failed to mesh the whole sequence ( dont know why, when I klicked "build" in mesh node, some of the frames which made RF crash meshed flawlessly), I wanted to simulate and mesh the exact same scene again in RealFlow 5. I had the substeps on 20 in RealFlow 4. Now when I simulate this in RealFlow 5, then a lot of particles keep falling through the geometry (cube, simulation is inside), even if I increase the collision distance and substeps (already way beyond what I had in RF4). I finally had the substeps to 333 standard value and a big collision distance, but still particles kept falling through after some frames. Only if I kept particle count under 2 millions, no particles (or at least less) fell through. Why could that be? To make them more stable I also reduced the pressures by factor 10 (later had 0.02 internal and 0.1 external). Because the particle cloud ( simulation starts with a circle emitter with a certain volume ) always exploded a little when I had millions of particles and did not reduce the pressures.

Any ideas? Again, in RealFlow 4 it simulates all the way through with 20 substeps, 0.02 collision distance and no particles pass the grometry ever. Fluid is stable.

[gus]

Hi,

It will be very helpful to troubleshoot this if you could send us over the scene.
We don´t want to jump to conclusions without having a look at what you are exactly doing.

g-)

[Far Star]

I read in the un offical RF forum that this is an issue that has been fixed and will be in next update sometime in July.

[atena]

The problem that has been solved is only if the object that is interacting with the fluid is in dynamic mode. If it is a normal static object this still might be an issue and it would be of great help if you could send us the scene.

Thank you very much.

[DragonsSpirit]

Hi folks....thanks for the answers.

I have now uploaded the file to wikifortio.com

The File ID which u need for download is :

765147


File will be available untill 14 days from today.

[atena]

Thank you so much for giving us access to the scene. We have simulated your scene here getting rid of the 20 max steps limit in the adaptive integration scheme (setting the 333 default value), now simulating frame 106 and no particle went through the cube so far.

One of the main changes in RF5 has been the integration scheme for the particle-based solver, many things have changed. For this reason one simulation step in RF4 is not the same than one simulation step in RF5. Simulations that went ok in RF4 with a limited number of steps might fail in RF5 with the same limitation. What is guarantee is that without limitation the same simulation will take less time in RF5 than in RF4.

Again, thank you very much for the report.

[DragonsSpirit]

Thank you guys. Ill check that...Got another question: When i downloaded the fire preset from your earlier website (RealFlow 4), I mean the scene which uses attractors and dumb particles, i noticed that the entire simulation would explode when I changed the Thread count from 2 (your settings in the downloadable scene) to 8 or any other. What causes that issue? I have also had that problem when doing my own tests, that simulations with dumb particles would not work on another machine with different thread numbers. I did not test if that is still the case but since some of my customers still may use RealFlow 4, it would be nice to know what to go for / avoid when using dumb particles. I dont have a scene to show the issue currently...ill make one and upload it if it does not work out.

[atena]

Sorry I don't know if I understood well the problem. You mean using the fire preset in RF5? We don't have reports on that issue and I've just tried to create a simulation with dumb particles in RF4 with 2 threads, then load it into RF5, change the number of threads to 4 and the simulation went ok. If you are able to reproduce the problem we would love to have the scene.

Thank you very much.

[DragonsSpirit]

No this was about RealFlow 4. RealFlow 4 did not simulate it proper when I changed the thread count. Ill try to make a scene as soon as possible.

[atena]

Thank you for the clarification. Hopefully this problem has gone with RF5, I wasn't aware about that problem in RF4.

Thank you.