Min Substeps / Max Substeps?

[JC_Chicago]

I have a stairwell that I am flooding, with cascading water down the steps. The top landing area is a simple plane with a hole for the sprinkler system standpipe to transverse through.
The emitter initially floods the flat stairwell landing fine, but after about 500 frames of simulation, the entire plane of fluid falls through the landing geometry. I believe that this is a substep issue, however, I have yet to find anything that clearly defines the role of substeps - both Min and Max. I have tried quite a few combinations, currently testing 60 Min, and 333 Max - but what does this mean???
The image is from behind the landing. The "tube" of water in the foreground is what "escapes" through the opening for the standpipe. The geometry has 14 million polygons (only the top 15% is visible here) ...
Any thoughts on why the fluid would fall through is much appreciated.
Regards.

[gus]

Hi JC_Chicago,

How are you?

We have had some similar issues with particles going through geo in certain circumstances, but managed to find the reason, and it has now been fixed for the upcoming patch.
Maybe it would be a good idea to have a look at your scene to see if it is the same and is no longer there. Would you mind sending it over to us? XML format and the geometry involved on a .zip file would be great if possible.

Cheers,

g-)

[apguest]

Hey,

I had the same happening on our Realwave sea, with hi particle count, we had to change the Collision distance to a higher value, for our scene it was 1.0.

That seemed to help.

[JC_Chicago]

I will send the file over this morning.

As with collision distances, a setting of 1 is just to great. Correct me if I'm wrong, but I would have to then make the scene that much larger to offset this in order for the water to look realistic. Right?

Thoughts?

[JC_Chicago]

Here is the XML document.

[gus]

Hi Jeff,

This is a workaround that should fix the unwanted leaking effect.
Create a cube that surrounds your fluid simulation, and then on start run the following code in a batch script. You may even want to save this as an initial state so you don´t end up running over and over the script until the patch is released.

Code: Select all

dummyCube = scene.getObject("<generate a cube that surrounds your fluids and put its name here>")
vertices = dummyCube.getVertices()
emitters = scene.getEmitters()
#get positions from dummyCube for particle insertion
for vertice in vertices:
	for emitter in emitters:
	#add dummy particles to emitter
		particula = emitter.addParticle(vertice.getPosition(),Vector.new(0.0,0.0,0.0))
		#freeze dummy particles
		particula.freeze()

Give it a go and let us know if it fixes it.

g-)

[JC_Chicago]

If I read your post correctly, I should put a cube around the entire stairwell model, not just the landing.
I have tried this, and it doesn't fix the problem.
I am perplexed why a simple flat plane would cause so much grief.
Do you have another solution? I can't use a simple plane from RealFlow, for it needs the hole for the sprinkler pipe...

Jeff

[gus]

Hi Jeff,

The cube is just to fix the position for the particles that will get created and added to the emitter(s). Did the script place those particles in the corners of the cube at all?
What are your substeps like for this simulation?
Also, what is the scene geometry scale and the particle collision distance for the object 'leaking' ?

Cheers,

g-)

[JC_Chicago]

Nothing was placed at the corners. To be quite honest, I may not be using the script correctly. Having said that, when I replace the geometry name in the script (per your direction), it runs without an errors. What should I have seen exactly? Also, should I make the cube the same size as the individual landing (top of the stairs), or the entire geometric scene?
The min/max are at 20 and 100, respectively.
The geometry has since been shrunk to try an avoid the issue/leaking - it is now roughly half the size it should be. The particle collision is set at 0.01.
Simulation is set at "High".

Your help is greatly appreciated, for I just can't fathom why a simple plane with a hole (for the standpipe) has been so difficult. I would think this is the simplest scenario.

Tessellation of the geometry (ridiculous levels) doesn't appear to fix it either.

I am wondering if I put the landing at a slight tilt (eliminating the perfectly flat plane concept) would help???

Regards,

Jeff

[gus]

Nothing was placed at the corners. To be quite honest, I may not be using the script correctly.

If there is nothing in the corners of the box, then it hasn´t worked properly.

Having said that, when I replace the geometry name in the script (per your direction), it runs without an errors. What should I have seen exactly? Also, should I make the cube the same size as the individual landing (top of the stairs), or the entire geometric scene?

Size should no matter as long as it contains all the fluids scope.

The min/max are at 20 and 100, respectively.

By min/max, are you refering to the simulation substeps? In that case, try to crank the minimum value up to 50 for example, and the max to 333, to see if you still see the leaking.

The geometry has since been shrunk to try an avoid the issue/leaking - it is now roughly half the size it should be. The particle collision is set at 0.01.
Simulation is set at "High".

Tricky to see a workaround without looking at the same scene. Maybe you could try and zip it and send it over again. This time try to reset, zip the flw with the objects involved, and send it to our support email.
If you are not using any body dinamics, then don´t worry about changing those settings.

Your help is greatly appreciated, for I just can't fathom why a simple plane with a hole (for the standpipe) has been so difficult. I would think this is the simplest scenario.

As you say, it is a simple scenario, but if there is a bug, then everything stops from being simple.

Tessellation of the geometry (ridiculous levels) doesn't appear to fix it either.

Again, if there is a bug, things can go nuts and tessellation my not be even having an effect.

I am wondering if I put the landing at a slight tilt (eliminating the perfectly flat plane concept) would help???

My guess is that you will still see the leaking. If we can get the scene, I am sure we will get a proper look at it and find a bug, if there is any, or a workaround if things are not setup properly or working as expected.

Regards,

g-)