Liquid in a cyclone

[JuhaHo]

I made another tryout. This time I created all new Phoenix system, simulated it and started tyFlow. Simulation worked fine with full frame steps but crashed immediately as I changed framesteps to 1/2. 1/4 would be required.

What next? I have clients jumping up and down and it's not sign of joy in this case.

[tashko.zashev]

Hi,

I'm working on the scene again and suspect that the geometry errors for some of the simulation issues you have. Like the Air pockets...
http://ftp.chaosgroup.com/support/sc...6_10-34-43.gif
The geometry need to be re-checked for issues shown by the STL check. I believe this will help to reduce the problems you have with this simulation.

Also in order to help further with the crash reports, we can set a remote session using the ticket you already have in our system.

[JuhaHo]

Thanks,

I made an STL check and indeed it found some errors. The collision object was made with VDB boolean modifier, so it was volumetric model which I thought to be watertight. Anyway I fixed the model and received 0 errors in STL check. I started tyFlow and it crashed as usual.

What does this VDB means. I was told not to use AUR, but VDB instead in Phoenix. What is the difference there?

[georgi.zhekov]

Unfortunately we still can't crash it on our side.

VDB is an alternative cache format compared to the native Phoenix AUR format.

Generally the vdb files are larger and render slower compared to the aur ones, but the vdb format is widely used in other software packages. So if you wish to use the volumetric data simulated by Phoenix in other software that doesn't support the AUR format, you can use the VDB one.

In the case where you use VDB you can then load the resulting VDB files into a tyFlow Birth VDB operator and use the VDB Force operator to drive your particles that way instead. That way you would avoid the use of the Phoenix loader to read the cache data into tyFlow and also potentially avoid whatever issue is causing the crash.

[JuhaHo]

Thanks,

Is there a AUR to VDB converter? Then I could use AUR during rendering together with Cached tyFlow?

[georgi.zhekov]

Yes - check this out - https://youtu.be/j2sUDknZoKY?si=mghcuhorD862AoWi

You can find more about the cache converter tool here - https://docs.chaos.com/display/PHX4MAX/Cache+Converter

[JuhaHo]

Thanks,

I tried, but all I got is "Access is denied" as a result.

[georgi.zhekov]

Where do you get this message?

Can you show us a screenshot?

[JuhaHo]

Thanks,

I'm not sure if I got this right. Anyway Your server is not accepting my screenshots at the moment.

I also simulated small sequence, but I'm not sore how to use it in tyFlow. I made a test which failed.

[JuhaHo]

What I've heard. Others have tested my scene. Max 2024 don't crash, but 2023 crashes. I'm working with Max 2022.

I would expect, that the cause for this crash is investigated and fixed as soon as possible. Something in the way Phoenix FD handles sub-frames is broken badly at the moment.

[JuhaHo]

I downgraded Phoenix FD Back to official 5.2.00, but it didn't help - still crashing.

Edit: I installed today's nightly. It crashes at Frame 3.

Edit 2. I installed Phoenix 5.10.01 and it works. So it seems Phoenix 5.2 cannot be used together with tyFlow, when Frame sub steps is used.

[georgi.zhekov]

Thanks a lot for the help - we finally managed to reproduce the issue and found the reason why it's crashing. We're on it and I hope to have a fix soon.

Using a Phoenix 5.10 build should work correctly meanwhile.

As soon as we have a fix in the nightlies we will let you know.

Thanks!

[JuhaHo]

Now when we have workaround for this crash issue, can we finally start discussing about the fluid behavior in a cyclone? Please take a look at the attached screenshot, where You can see, that Phoenix is not simulating the cyclone correctly. The fluid goes immediately uncleaned to the exit pipe without going down first. I should go down circulating on the walls while picking up speed. Then on the bottom of the cyclone the flow turns up and starts to rise in the middle.

How can I make the fluid go down first? Is there any suggestions for workaround to make Phoenix simulate more like real life. I think I have asked this earlier (maybe last summer) in this thread.

Edit: The reason why I want to have it behave more accurately is that, this behavior ruins particle simulations too. If the fluid don't travel down, neither do the particles.

[JuhaHo]

I didn't have better idea, so I decided to drag the centerpipe wall all the way down, so the liquid must travel down, before it can go upwards. Then before the pipe lower end I made tyFlow icon which sends the particles to the next event, where the fluid force is ignored, so the stones will continue towards the junk trap and don't follow the liquid upwards.

Of course this requires two Phoenix simulations: one with center pipe for the particle sim and the other for rendering.

[JuhaHo]

The setup above didn't work at all. Now the fluid is falling directly instead of vortex. Is there a way for creating a vortex force to fluid when it's going down?