How to make a fire tornado

All you need to make a fire tornado is:

  • A bin made from metal mesh. Something like this.
  • A lazy susan
  • The little tin case from a tea light. This sort of thing.
  • Lighter fluid
  • A BBQ lighter

Put a little bit of the lighter fluid into the empty tea light. Not too much. Put in less that a millimetre. That way it'll safely burn itself out after a short time.

Put it all together like in the video and light it. Give the whole thing a little spin.

I experimented with different diameters of mesh tube and different containers for the lighter fluid. I discovered:

  • A wider mesh is better. I'll explain why later
  • A smaller container is better. This produces a much more dramatic change in high when the vortex kicks in.

Why does this happen?

A flame draws air in from around it. That's because it heats the air. Hot air is less dense and so it rises up. Air is drawn in from the sides to fill the space left behind.

In our case the air is drawn in through the mash of the bin. As it passes though it gets pulled around in a circle. It's now got some kinetic energy. When the air is pulled further into the middle, that energy is conserved. It's still going round in a circle but the circle is much smaller so it's orbital speed will increase to keep it's kinetic energy the same. This is the conservation of angular momentum.

So the reason a wider mesh is better is that for the same orbital speed a wider mesh will be travelling faster and imparting more kinetic energy.

I'm not sure why the flames get bigger. Any ideas?

You might also like these giant steel balls you can burn things with.

Tags: , , , , , ,

  • Michael Eve

    . as the fuel heats up it evaporates more rapidly so rate of consumption increases.
    . as the apparatus heats up, the convection current up the axis increases.
    . something else.

    • Steve Mould

      So the question is, why does this happen more when the air is spinning?

    • SethK

      Along with both of these, increased turbulence in the air/fuel mixture could be causing more complete combustion (which would actually amplify the other two effects if it were true).

      • SethK

        Forgot to mention, increased turbulence would explain why those effects were more pronounces when the system is spinning.

        • Anonymous


  • Blaze99960

    Typically the evaporating lighter fluid expands in all directions and thus is not dense enough in any one place for it light to light. The increased draw of air in forces the evaporating gas upwards and since that is a smaller area it is more condensed and is able to ignite.

    • Steve Mould

      Thanks Blaze, I like you’re explanation!

  • jtarndt90

    Maybe fire has a small but significant amount of mass to be able to stack and roll on its self to create a twisting effect while the air being warmed around it helps it shoot further up out of the basket…

  • noodleincident

    The spinning air creates a vortex like when a sink is drained, this lets more oxygen get to the fluid and for the CO2/other waste to exit more quickly, just like a vortex lets water flow more quickly. Thus, more flame. If you want to see the same principle in a safer/cheaper/more useful way, let a full bottle of water drain by holding it upside down, then do it again while spinning it once or twice.

    • Steve Mould

      That sounds pretty good to me!

  • Pingback: Burning paper with giant steel balls | SHIFT_beep

  • Bsfa

    If you spin something, e.g. a ball on a rope and let go of the rope, the ball moves tangential to the orbit it followed when still held. That is, it moves away from the center (you). However, hot air is lighter than cold air, so it is the cold air inside the mesh cylinder that moves away from the center, and the hot air is kept  in the center. It doesn’t mix with the cold air very well. Because it is lighter than cold air, it is pushed up by cold air from the bottom. Because it doesn’t mix with cold air, it keeps rising. As a result, it will take up evaporating gas upwards, as Blaze said.
    My prediction is that the experiment works as well/better if the mesh cylinder is a solid cylinder, with just the lower edge being (wide) mesh. Hope you can do the experiment again.