Steam Flow velocity issue

I try to make a steam thrusters for airballoon.

So they push the steam very fast downwards…
This is what I managed to make:

I would love to make it even more powerfull, faster. This could be done with velocity.

But the velocity cannot be over 6000, otherwise the steam just disappears, as you can see form the video. What would be the solution for keeping the steam visible even with higher velocity value?

Hello @pekka.varis! That looks really cool! I’ve asked the dev team for some advice on how to make it more powerful!

1 Like

First thing I would try is to increase temperature.

1 Like

I face a similar situation but instead my Flow is shaded and set to be a horizontal flame generated by a large industrial burner where fuel is spayed out under high pressure into the center of a 60 mph air stream generated by a large fan, creating a 10-foot long by 7-foot diameter flame conical in shape.

The fuel travels about 18" out from the nozzle in a wide-angle cone before combustion begins. As you would expect, the resulting flame is not gentle, with slowly lapping swirls and tongues of flame that Flow does so well, but extremely fast and chaotic.

I am unable to get Flow to simulate this flame, whose entire body is at the same high velocity from source to tip. Instead, the flame starts out at nearly zero velocity then progressively increases to the desired speed at the flame/smoke transition. I know this is due to the logarithmic nature of gravity and buoyancy, so I was hoping setting the flowEmitterSphere’s Velocity to a very high, 4,000 value, the same as Pekka Varis shows in his video above, would make the entire volume move at that speed. Velocity does push the start of ignition farther away from the source, which I like, but the initial flame speed is still nowhere near real world, nor is the entire volume of the flame at the same velocity. Buoyancy has nearly zero affect in the real world, because there’s not enough time for it to act on the flame itself, whereas Expansion is the primary force.

Also, under such extreme parameter settings, the flame becomes unstable, and most times will not reignite after stopping and restarting the simulation.

I have a feeling that the CFD algorithms Flow uses are not designed to create this high-pressure fuel, high-velocity airflow type of flame. Correct me if I’m wrong.

At any rate. I can produce a nice looking and nearly properly-shaped flame, but it plays in slow motion at 1/20th or more real-world “speed” (see video in my next post below). My RTX3090 frame rate is at a respectable 30fps, even though the flame is inside a rough, but reflective metal drum, so LOTS of rays being cast, because I disabled all raytrace options for Flow – seen in reflections, etc.

Anything close to this video would be great if Flow can replicate it! But, again, I don’t see how it can do this kind of speed. Same for Pekka’s steam needing to appear “more powerful.” His simulation needs to obviously extend much longer downward that this video does horizontally and have no flame-/cloud-like, detaching filament wisps.

1 Like

Such a sweet large industrial burner you got there, Dean!

@areidmeyer - Superior.
That was the fix, thank you!

Dean, I mark this solved now. Nice thing is that the more velocity I add to this steam, the more it looks like your real burners flame!!

Somewhat close, but your sim “breaks,” with voxel boundaries clearly visible at the end of the flame, err steam.

I see what you did there… increased velocity up to 12,000 and temperature to 90. I will just have to reset my sim to defaults and start over, upping velocity and temp to those levels, etc. I was able to get up to 4,000, but didn’t adjust temp up that high. Any higher and the flame became so thin, tenuous, and unstable that it would collapse into a cloud of points and fail to ignite when the sim was stopped/restarted.

Thanks for letting me butt in.

1 Like

High speed flow are definitely problematic to simulation at 60 Hz. It just isn’t a sufficient sample rate. The video you shared sort of shows that effect, where the flame seems to almost just be noise frame to frame. Would need a high speed camera to see the movement properly.

One other note is Flow today uses FP16 velocity, so around 65K it risks going inf. Will consider adding a FP32 mode in future updates for these cases.

Is your goal real time? If you don’t need real time, you can definitely increase the Flow simulation rate so it takes smaller timesteps, and increase the maximum steps Flow takes each frame. This will help with fast motion.

The block artifacts in the one video. One thing to check is the Flow Monitor window, that will tell you if you are running out of blocks. That is the most common cause of that effect. High speed effects the default block allocation behavior might also not be sufficient. You can use just a box/sphere emitter with all couple rate values to 0.0 but allocationScale > 0.0 to force allocation behind the nozzle. This will improve behavior.

1 Like

I increased Allocation Scale = 2, Steps/sec = 120, Max Steps per Sim = 5, Velocity Sub Steps = 2, Fuel = 0.6, Temperature = 15, Velocity = -9,000/800/0, Vorticity Force Scale = 9, and several other “minor param” tweaks and was able to produce a flame look and profile closer to “real” world.

Some of our burners producer longer, fuller flames, so this is a happy middle-of-the-road effect. At least the speed and staccato/strobed flame licks are so much closer.


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.