Liquid passing through object

[Thanos Kagkalos]

Hi, i m uploading this scene if anyone can take a look. I have this animation of a mixer shaking and the liquid is passing through.
I ve gone even to 32-64 substeps and still a problem. I tried thickness on the model, and again the same.
Is there something im missing? Also i cant tell the difference between the distance in collision tag and in volume tag.
thanks a lot

scene file: https://www.dropbox.com/s/f8jz9gh9rsi2c ... t.c4d?dl=0

[Thomas Schlick]

I'm on it, but that's a very hard nut to crack.

[Thanos Kagkalos]

Dont waste ur weekend on this Thomas haha. I m just looking for the overall theory behind scenes like this, cause sometimes im not quite sure if i do something wrong or if the plugin cant deal with it. I really appreciate the help here!

[Thomas Schlick]

I was able to make the blender watertight, but only with Liquid-SPH and a really high number of substeps (600/600) and a pretty big surface offset - and the simulation took ages. The problem with this model is the sharp edge in the model's lower part. This is the area where the particles go through when they're suddenly accelerated. I'll pass this scene to our developers to see if there are ways to improve the collision geometry with one of the next updates, but I think it's something difficult to fix. The only thing I can imagine for now is to (drastically) increase scale, make everything much bigger, but then you'll need much more particles as well.

Also i cant tell the difference between the distance in collision tag and in volume tag

The parameter in the volume tag really affects the collision geometry object and makes it bigger or smaller. In the collider tag it's just a virtual distance and you can make the particles "fly" above the collision geometry. To make the collision geometry visible go to the collider tag's "Display" tab and check "Show Collision Geometry". If it doesn't show up go to the last frame of C4D's timeline and then back to 0.

[Thanos Kagkalos]

Thanks again Thomas for the help. I did the sim eventually by making it x10 times bigger yes and i had a surface offset of a good amount as well. Weird thing (in a positive way) is that in the final render the offset didnt seem much and that is due to the meshing i think, made everything a bit thicker (i ll upload when i finish the project).
One last question on this, is 600 substeps viable? What are the reasonable numbers for generic simulations?
Thank u very much

[Thomas Schlick]

Yes, the surface offset can be compensated through thicker meshes and you actually won't see any gaps.

One last question on this, is 600 substeps viable? What are the reasonable numbers for generic simulations?

That's always the problem with simulations: there's not really a common range for parameters. Each simulation is different and requires different settings. There are so many things you have to take into account, like objects, scales, forces, the number of particles, particle type, etc. Anyway, for Liquid-PBD, the number of substeps should range between 2/2 and a maximum of ca. 30/30. Higher settings normally don't make sense, because they'll affect your fluid too much - it becomes more and more viscous/rigid.

With Liquid-SPH it's different. There, substeps of even several thousands can be adjusted, but then you'll lose everything Dyverso was made for: speed. Although it might be necessary with very high viscosity settings, but for this purpose we have the viscous and viscoelastic materials. Several hundreds of min/max substeps are Ok (the default is 1/300), but when a single frame takes 10 or 15 minutes to simulate due to high substeps. I normally try to find alternatives. Something around 5 min/frame are OK for me with SPH, but that's just my personal limit and also depends on the number of particles. If you're fine with 10+ min/frame then just do it

[Thanos Kagkalos]

Thanks again Thomas for all the help and for the clarifications. Cheers