Smoke going through a pipe

[Deflaminis]

Hello, I'm working on making smoke going through a pipe. Any tips for this?

The problem to solve for me seems to be detail/vs speed. I'd like the smoke to just travel the pipe but obviously things like buoyancy make it a bit tricky. In reality this pipe has it's own vacuum so there's no real impedance but I'm not sure how to fake that here. I took a shot at using a negative discharge rate on another source to try to suck up the smoke, but it seems that only work if the smoke touches the emitter. I can't seem to get it to suck up the smoke from far away even though it's a closed system with no air intakes. (maybe that doesn't matter, I'm not sure.)

When I make the emitter within the pipe bigger so I can get it over the initial bend of the tube, the smoke sort of fills the space so perfectly it just looks like a gray line in there. If I make the emitter smaller but have more discharge, it doesn't seem to want to make it down the pipe. Negative buoyancy seems to be hard to control with this weird setup I have, and I'm sure the problem is on my end so I'm just looking for any tips for a situation like this one. Would it make more sense to put a pflow down a path in there and use that instead? (Like how you might approach it with something like afterburn?)

I realize that in reality the smoke would be almost impossible to recognize as it would be whipping by quite quickly, and I'm looking for a less realistic but more recognizable setup here, which is a pain I know...

Also weird is that when I press f9 to do a quick render, sometimes it renders all the smoke and sometimes it doesn't. I can press render 2x in a row and get a different image. Examples labeled A and B here. Single omni light with ray traced shadows in the scene. It's super weird so I attached the scene here in case you want to investigate.

[Svetlin.Nikolov]

The most important setting in if you want the smoke to travel a long distance is the conservation quality - you need to increase in order for the smoke to reach the other end more easily. This way the sucking source will work better too, in case you still need it.

About the lighting issue, are you sure you have atmospheric shadows enabled on the light as well?

[Deflaminis]

Didn't have atmospheric shadows on... thank you. After a few minutes I noticed that turning it on didn't fix the problem though. Setting to to Adv. Ray Traced did, I had it on regular ray traced for some reason. Probably was both things causing my problem.

I'll work with the conservation quality on this some more. Based on the help files on the web site, am I right in assuming the highest you really need to go is 2x the longest dimension? Meaning here, 2x the height in units? (multiplied by cell size?) (as demonstrated in the movies with the swirling smoke on the dynamics section of the online help?)

EDIT: The conservation tip was great, even using 466 (based on a 233 height cell) it's very quick to simulate and reaches the bottom with 10x less discharge rate. So, that problem was well solved there. With a high conservation, it becomes clearer that the negative exhaust is definitely successfully creating a vacuum inside the tube.

How can I make the smoke separate a bit more inside the tube though? Should I use multiple sources with different temperatures so they don't mix quite as easily? (Less solid inside and more smokey looking?)

Thanks for your help!

[Svetlin.Nikolov]

Ah, in case you render with vray, you need vray shadows - other types will work only by accident.

Getting smoke detail inside the tube will be tricky though. You could try using forward transfer advection, and if that doesn't work well, increasing resolution and making the smoke emitter smaller than the pipe is the only alternative I can think of.

[Deflaminis]

Sounds like a plan, thanks for the heads up. Will take a shot. Yup vray shadows worked great. I saw some tutorial somewhere saying to use ray traced but they must have been using scanline and I didn't notice. Kind of a no brainer after I thought about it, thanks for setting me straight.

EDIT: I wonder, is it possible somehow to get the vacuum working before I start actually discharging? Not sure if I'm explaining this well...

When the discharge first happens it starts all the swirls and eddies start to form which causes the other smoke to start to follow. That beginning part is what causes a lot of the expansion to fill the tube, which I want to reduce. However once the (wind?) gets going it's closer to what I want. Is there a way to get that wind and vacuum going before I start emitting? Or could I use a fd helper that emits only velocity at first and then use a 2nd one with an offset discharge to start discharging after that velocity or whatever is already there? Basically getting a sort of starting funnel/tunnel for the smoke to follow once it starts discharging? Or am I way off here?

Man, this must be hard to read. I hope I'm making sense.

[Ivaylo Katev]

you can try a vacuum source (negative discharge) in the opposite end of the pipe

[Deflaminis]

This has been pretty fun so far. It's working very logically for the most part. Thanks for the help, both of you.

Is there a good way to preview forces inside a negative emission? Would it be velocity or? That's exactly what I ended up doing, but I'm not sure how to tell when it's in full force so I can start emitting easily.

If the entire object is closed, is that a problem for it without some positive discharge, like sucking on a soda pop bottle? If so, should I use 3 sources? For example: 1 negative at bottom, 1 positive emitting only velocity(?) to get the vacuum effect going steadily, and 1 that emits smoke after the vacuum system is going?

[Ivaylo Katev]

yes, you can use the velocity channel to visualize the forces.
it's not a problem to use closed container with not balanced sources inside, at least with the direct solver. svetlin mentioned that the cg solver can produce instability in such case, but i tested it and it seems to work stable.

[Deflaminis]

Thanks will do. I really appreciate all your help.