Center/OffCenter simulator- different results

[flaviuscristea]

Hi there!

I was doing my tests on half/full simulator and I have noticed some differences. Please see attached images. So when the simulator is off center the results are different. I have ran all the simulations twice to confirm I get the same outcome.

If it is centered the wave looks nice (I ran this with smaller and bigger simulator, I get similar -good results)
Smaller centered simulator- results are like the bigger one.

Slightly off centered simulator- Results don't seem as good to me, the wave is broken.

what is weird is that a really of center simulator shows similar results to the centered one.

Is this a correct behaviour? And if yes, why is that? I'm curious how this works and why it happens because it has confused me a lot...


P.S. My anchor pocket is a separate object which intersects the hull. Is this OK for Phoenix ? I read that there might be issues with single sided polygons and SELF intersecting geometry. But what about two objects intersecting? Is this fine?

[Svetlin.Nikolov]

Ah yes, if you have an object that should not interact with the simulation, it would be best to make it non-Solid from its right click properties or even just exclude it from the Interaction rollout. It really looks as if something's blocking the foam birth or flow in the two off center shots...

[flaviuscristea]

That's the thing, I didn't move or do anything on the objects, I have only moved the simulator and I get different results. It seems very peculiar to me. I have ran each simulation twice and each time, center simulator is OK, off center gets weird results.

About intersection, I meant that the anchor pocket polygons intersects the hull polygons. I was wondering If I need to mold that, I mean make one object with continuous clean edge loops between hull and anchor pocket.

[Svetlin.Nikolov]

Ah, got it. Sounds like there could be issues with the geometry this way - you can check it by enabling some debug options, but you will need to simulate at least a frame in order to find out. Go to the Output rollout and set the Special channel to 'voxelization errors'. Then run the sim and in the Preview rollout, turn off everything - show mesh, all voxel channels, and the particle previews, and enable only the preview of the Special voxel channel. If any voxels show up, then the geometry needs welding or shelling. Don't forget to disable the special channel then, or it would consume ram and will make the cache files larger...

[flaviuscristea]

I have been doing over 100 tests today with this, I'm going slightly mad because getting inconsistent results. I need to get a clean wave in the front. ). Thanks for this channel thingie, sounds very interesting I will do it right now!

[flaviuscristea]

I have made a single mesh, clean topology no separate objects. And I get very different results if the simulator is centered or off centered. (like it is shown in the ship tutorial 2018 from phoenix fd doc). Any idea what might cause this?

[Svetlin.Nikolov]

Can't say more without the scene or at least thorough screenshots of the scene setup and the Phoenix settings.

[flaviuscristea]

I will organize and prepare the scene, and send it. Thank you very much!

[Svetlin.Nikolov]

Thank You!

[flaviuscristea]

Hey there,

I noticed that if I change the position or the size of the simulator I get different results... For example, I am trying to get a nice wave so I don't care what is happening on the rest of the ship so I made a small sized liquid simulator..However, if I move it off center or make it bigger the results change quite dramatically without changing any other settings. I only changed the dimensions of the simulator, I haven't touched cell size or any other setting. See in the attached images how different it looks(simulated up to frame 120 ). I am extremely curious if this is a bug or there is something I don't understand.

I am trying to save simulation times by adjusting things on the model only on areas I am interested in. Like, the rest of the simulation looks fine it is just the front part I am not happy with. But getting different results on different liquid simulator sizes/position makes this job impossible.

Thanks a lot!!

P.S. - Scene sent with a screenshot some explanations. But just simulated it quickly, even 90 frames are enough are think. Then change the position or size of the simulator a bit and see the differences.

[Svetlin.Nikolov]

Yes, I suspect that what it would most likely come down to is that since Foam and Splashes are born in a way that depends a lot on the liquid velocities, you need to leave some extra grid space around the places where the interesting velocities happen.

For example, since the ship pushes liquid in front of it, you need to leave more room head of it and to the sides, so shrinking the container too much and cutting off these velocities might be the reason for your pain. Note that at the borders of the simulator, the velocities seize to exists, so any effects they could cause back into the container would be skewed unless you give the large velocities enough room to unfold so that at the borders of the sim they have faded naturally. You can use the Velocity preview under the Preview rollout to keep an eye on that (need to export Grid Velocity when simulating). If you have large chunks of velocity right near the simulator wall, you will get skewed results.

[Svetlin.Nikolov]

If this is the cause of the results you are getting, then it's analogous to how fire/smoke simulations too. When simulating a large rolling smoke plume or fireball, like a nuke for example, you need to leave some space to the sides, so velocities can swirl there. Even though there is no visible data there, not giving room for the vortices to play out would make the plume stop rolling and dissipate quickly, so I suggest you check the velocity preview (also try turning off auto range so you have more control over what data you see) and see if there aren't any major velocity streams getting cut off at the end of the simulator - both to the sides and to the front.

[flaviuscristea]

!! This is quite an enlightenment and it makes sense. I will do further tests with it, leaving more space and see if there are differences. I probably pushed the trick from the tutorial (with half the simulator) a bit too much

Somehow I don't see the velocity grid. I can see the liquid one though..

Thanks again

[Svetlin.Nikolov]

Ah yes, turn off Auto-Range from the Preview rollout and knowing that the velocity goes up to 158 from the Cache File Content under the Simulation rollout, you can adjust the minimum and maximum visible velocities in the preview to the range that is interesting - say from 10 to 160 (a bit above the upper limit so velocities won't get skipped when scrolling the timeline and if there are higher ones in the other frames...)