Astrophysics

The candle that burns as a sphere in free fall

A candle flame is tall and yellow because the hot gas it makes is lighter than the air around it and rises, pulling fresh air in at the bottom. Buoyancy is the weight of the displaced air, and in free fall nothing has weight — so a candle lit in a falling box, or in orbit, has no draught at all. Its flame becomes a dim blue ball fed only by molecules wandering in from the still air, about a hundredth as fast as the draught that fed it on the ground. The equivalence principle, which says a falling laboratory cannot tell itself from one without gravity, turns out to be a statement about how fires breathe.

Assumes: The floor that cannot be told from gravity · The balloon that leans the wrong way

A candle flame has a shape everyone recognises: a narrow teardrop, pointed at the top, with a blue base and a yellow body, standing straight up whatever the candle is stood on. The shape is not the fire’s own. It is the fire’s arrangement with gravity. The flame heats the air around it, the hot air is lighter than the cold air beside it and is pushed upward, and as it rises it draws fresh air in at the base and carries the flame’s products away above. The teardrop is the outline of a plume.

So a candle is a gravity experiment, and the equivalence principle makes a prediction about it. The floor that cannot be told from gravity stated the principle: inside a freely falling box, nothing local can tell whether there is gravity outside it, because everything in the box falls together. A candle in a falling box should therefore burn as a candle would where there were no gravity at all. That has been done, in drop towers, aircraft flying parabolas and orbiting laboratories, and the flame does something no flame on the ground can: it becomes a sphere.

Buoyancy is a weight

Hot air rises because of buoyancy, and buoyancy is easy to misremember as a property of hot air. It is a property of the air around it. The weight of the water that is not there put it in its plainest form: a body in a fluid is pushed up by the weight of the fluid it displaces, because the fluid’s pressure increases with depth, and the pressure increases with depth because the fluid has weight. Remove the weight, and the pressure in the fluid is uniform, the push from below equals the push from above, and buoyancy vanishes. The balloon that leans the wrong way used this to explain why a helium balloon in an accelerating car moves forward: the car’s acceleration acts like an extra horizontal gravity, the air’s pressure gradient follows it, and the balloon is pushed the way the gradient points.

In a freely falling box there is no pressure gradient at all. The air in the box falls with it, at the same rate as the walls, and is weightless relative to them. A pocket of hot gas is no lighter than its surroundings in any way that matters, because “lighter” has nothing to act on. A flame lit in free fall heats the air around it and the hot air simply stays where it is, spreading only as fast as heat and molecules can diffuse.

Two ways to bring oxygen to a flame

A flame burns as fast as oxygen reaches it. On the ground, the plume does most of the delivering: a centimetre-sized flame whose gas is several times hotter than the room rises at a speed of order gℓ ΔT/T\sqrt{g\ell\,\Delta T/T}, about 0.7 metres a second for a flame a centimetre across and gas five times hotter in kelvin than the room, and fresh air is dragged in at that sort of speed. Without gravity, oxygen arrives only by diffusion — molecules wandering in from the still air at a rate set by the diffusivity DD and the size of the flame, roughly D/ℓD/\ell.

Below a ten-thousandth of g a flame breathes only by diffusion. Two ways oxygen reaches a flame 1 cm across, against the strength of gravity: the rise of its hot gas, about √(gℓΔT/T), 0.70 m/s on Earth, and diffusion, about D/ℓ = 1 cm/s whatever gravity does. They cross near 2.0 × 10⁻⁴ of Earth's gravity; below that the flame is fed by diffusion alone. A laboratory in orbit sits near a millionth of g, a hundred times below the crossover. Rough estimates for a candle-sized flame.
Fig. 1 Two ways oxygen reaches a flame 1 cm across, against the strength of gravity: the rise of its hot gas, about gℓ ΔT/T\sqrt{g\ell\,\Delta T/T}, 0.70 m/s on Earth, and diffusion, about D/ℓD/\ell = 1 cm/s whatever gravity does. They cross near 2 × 10⁻⁴ of Earth’s gravity; below that the flame is fed by diffusion alone. A laboratory in orbit sits near a millionth of g. Rough estimates for a candle-sized flame.

The diffusivity of oxygen in the hot gas round a flame is of order 10−410^{-4} square metres a second, several times its value at room temperature because molecules move faster and collide less effectively when hot, as how far a molecule gets followed. Over a centimetre that makes a delivery speed of about a centimetre a second, seventy times slower than the draught. The two are equal only when gravity is about two ten-thousandths of Earth’s. Below that, the buoyant draught is smaller than diffusion and the flame does not know which way is down. An orbiting laboratory, whose residual accelerations from air drag and crew movement are around a millionth of gg, is two hundred times below the crossover.

The crossover’s position follows from the square root. Buoyant speed falls only as the square root of gravity, so making a flame forget gravity takes a reduction of gravity by the square of the speed ratio: seventy squared is about five thousand. That is why low-gravity combustion needs real free fall rather than a weak field. The Moon, at a sixth of gg, is nowhere near; a flame on the Moon still has a plume.

A flame on the Moon is still mostly a terrestrial flame

The speeds are estimates. A better way to ask how much a flame’s burning depends on gravity is to use what engineers measure for any hot body in still air: the rate at which heat, or equally a diffusing gas, is carried to or from a sphere by the natural convection its own temperature drives, expressed as a Nusselt number — the ratio of the actual rate to the rate by diffusion alone. For pure diffusion round a sphere that ratio is exactly 2, from the steady solution of the diffusion equation; buoyancy adds a term that grows as the quarter power of the Rayleigh number, which is proportional to gravity.

Gravity roughly doubles what a small flame can burn. The rate at which oxygen reaches a body 1 cm across in still air, relative to the rate by diffusion alone, against gravity, from the standard correlation for natural convection round a hot sphere. In free fall it is 1 by definition; on Earth 2.18, on Mars 1.92, on the Moon 1.75. The effect fades slowly, as the quarter power of gravity, so a flame on the Moon still has most of a terrestrial draught, and only near orbit does it lose it all.
Fig. 2 The rate at which oxygen reaches a body 1 cm across in still air, relative to the rate by diffusion alone, against gravity, from the standard correlation for natural convection round a hot sphere. In free fall it is 1 by definition; on Earth 2.18, on Mars 1.92, on the Moon 1.75. The effect fades as the quarter power of gravity, so only near orbit does a flame lose its draught entirely.

By this measure a centimetre-sized flame on Earth gets about 2.2 times the oxygen it would get by diffusion alone, and so can burn about 2.2 times as fast. That is a smaller factor than the speeds suggested, because a plume does not deliver oxygen as efficiently as its speed implies — the oxygen still has to diffuse across the thin layer of hot gas next to the flame — but it is a factor that depends on gravity weakly. On Mars a flame would get 1.9 times its diffusion rate, on the Moon 1.75. Fire on another planet is not very different from fire here. Fire in free fall is.

The flame in its own exhaust

Diffusion has a property that makes a flame without a draught behave strangely: it is slow to reach far, and it never stops reaching. A sphere that consumes all the oxygen arriving at it draws down the oxygen around it in a region that grows with time, and in the steady state the oxygen fraction rises towards the room’s value only as 1−rf/r1 - r_f/r — a dependence on distance that falls off so slowly that half the oxygen is missing at twice the flame’s radius.

In still air the starved zone keeps growing. The oxygen round a spherical flame 1 cm across that consumes all the oxygen reaching it, fed only by diffusion, 0.1, 1 and 10 s after it lights, and in the steady state. The depleted region spreads outward as the square root of time and settles to 1 − R/r, R the flame's radius, which approaches the room's value only slowly: half the oxygen is missing at twice the flame's radius, a fifth at five times. With no rising plume to carry products away and fresh air in, the flame sits in a shell of its own exhaust.
Fig. 3 The oxygen round a spherical flame 1 cm across that consumes all the oxygen reaching it, fed only by diffusion, 0.1, 1 and 10 s after it lights, and in the steady state. The depleted region spreads outward as the square root of time and settles to 1−rf/r1 - r_f/r: half the oxygen is missing at twice the flame’s radius, a fifth at five times.

The same slow fall-off governs the flame’s products. Carbon dioxide and water vapour diffuse away from the flame by the same rule, so the flame sits inside a thick shell of its own exhaust, diluting the oxygen that has to reach it. On the ground the plume sweeps both away within a fraction of a second. In free fall the shell persists, and a flame in a closed box gradually starves itself. Candle flames in orbit burn as dim blue hemispheres — the candle body blocks the lower half — and in the candle experiments flown on orbiting stations they were small and faint, burned slowly, and in some cases flickered and went out as their surroundings filled with their own products.

The blue is itself a consequence of the slowness. A candle’s yellow is the glow of soot, tiny carbon particles that form in the fuel-rich interior of the flame and are heated to incandescence before they burn, the light the glow that says nothing about the surface found to be blackbody emission from small hot particles. A diffusion-fed spherical flame is cooler and its fuel spends its time differently; it makes less soot, and what remains is the blue emission of excited molecules in the reaction zone, which on the ground is visible only at the flame’s base.

What free fall made measurable: the burning droplet

The cleanest flame free fall allows is not a candle but a single droplet of liquid fuel, ignited while hanging on a fibre or floating free. In still air without gravity it burns as a perfect sphere of flame round a spherical drop, and the theory of that arrangement, worked out in the 1950s for the fuel sprays in engines, makes a clean prediction: the fuel’s evaporation rate is proportional to the droplet’s diameter, so the droplet’s area shrinks at a constant rate.

A burning droplet shrinks by area at a steady rate. The squared diameter of a heptane droplet burning in free fall against time, for droplets 1, 2 and 3 mm across, with a burning constant of 0.8 mm²/s. Fed only by diffusion, a sphere's burning rate is proportional to its diameter, so its area falls at a steady rate and its life goes as the square of its size: 1.25 s, 5.0 s and 11.3 s. A 2.2 s drop tower can burn only the smallest out; the larger ones needed longer towers or orbit.
Fig. 4 The squared diameter of a heptane droplet burning in free fall against time, for droplets 1, 2 and 3 mm across, with a burning constant of 0.8 mm²/s. Fed only by diffusion, a sphere’s burning rate is proportional to its diameter, so its area falls at a steady rate and its life goes as the square of its size: 1.25 s, 5.0 s and 11.3 s.

That is the d2d^2 law: d2=d02−Ktd^2 = d_0^2 - Kt, with a burning constant KK that for heptane in air is about 0.8 square millimetres a second. On the ground a burning droplet distorts into a teardrop like a candle and the law is only approximate; in free fall it holds well over most of the droplet’s life, and experiments in drop towers and orbit have measured the constant for many fuels and found where the simple theory falls short. One of the surprises from orbiting experiments was that large heptane droplets, after their visible flame went out from radiating away too much heat, sometimes kept burning invisibly in a slower, cooler chemistry, a cool flame that on the ground would have been swept away by its own plume before it could be noticed.

The square law also sets the price of an experiment. A one-millimetre droplet burns out in 1.25 seconds, a two-millimetre one in five, a three-millimetre one in eleven. Doubling the size of the flame quadruples the time it needs.

Seconds bought by falling

Free fall on Earth is expensive in height. A body falls a distance 12gt2\tfrac12 gt^2 in time tt, so the time available from a height hh is 2h/g\sqrt{2h/g}: the height has to be quadrupled to double the time.

A drop tower buys seconds, and a flame needs its size squared. Time in free fall against the height of the drop, √(2h/g), with three working drop facilities marked: 2.2 s from 24 m, 4.7 s from 110 m and 5.2 s from 132 m. Against them, the time a diffusion flame needs to settle into its new shape, about r²/D: 1 s for a flame of radius 1 cm, 4 s for one of radius 2 cm. Small flames can be studied falling; a candle's needs an aircraft's twenty seconds or an orbiting laboratory.
Fig. 5 Time in free fall against the height of the drop, 2h/g\sqrt{2h/g}, with three working drop facilities marked: 2.2 s from 24 m, 4.7 s from 110 m and 5.2 s from 132 m. Against them, the time a diffusion flame needs to settle into its new shape, about r2/Dr^2/D: 1 s for a flame of radius 1 cm, 4 s for one of radius 2 cm.

The drop towers built for this work are shafts tens to over a hundred metres deep, some evacuated so that air drag does not slow the falling capsule, and they buy between two and five seconds — a little more by firing the capsule upward first and letting it rise and fall. A flame takes a time of order r2/Dr^2/D to settle into its diffusion-fed shape after the fall begins, which is a second for a flame of centimetre radius and four seconds for one of two. Small flames and small droplets fit in a tower; a candle’s flame barely does, and to watch one burn for minutes takes an aircraft flying parabolic arcs, which gives about twenty seconds of reduced gravity at a time and not very cleanly, or an orbiting laboratory.

There is a symmetry here worth noticing. The scales that work without weight found that weighing in orbit has to be done by inertia, by shaking, because weight has gone. Combustion in orbit has to be done by diffusion, because buoyancy has gone. In both, the equivalence principle removes a convenience that every terrestrial instrument relies on so completely that nobody notices it until it is gone.

Why this is a statement of the equivalence principle

The claim that a flame in a falling box burns as if there were no gravity is not a theorem of combustion. It is the equivalence principle, applied to a gas. It requires that the cold air, the hot air, the fuel vapour and the flame’s products all fall at the same rate, regardless of their composition and temperature; if hot gas fell even slightly faster or slower than cold, a residual buoyancy would appear in the falling frame and the flame would have an up. The fall that does not depend on what is falling followed the tests showing that bodies of different composition fall together to parts in 101510^{15}; a candle tests the same thing at a precision of perhaps a part in five thousand, set by how much residual draught would be noticeable, and it passes.

What a freely falling box cannot remove is the tidal part of gravity, the slight difference in gravity’s pull across the box that the term free fall cannot remove identified. In an orbiting laboratory that difference is a few tenths of a millionth of gg per metre, far below the crossover in the first figure, so tidal buoyancy is irrelevant to a flame. The residual accelerations that do matter in orbit come from the station’s own motion and air drag, a millionth of gg, and they too are far below the crossover. A candle in orbit is as good a falling box as anyone has needed.

The principle runs the other way as well. A station that spins to give its crew artificial gravity gives its candles a floor too: in a rotating frame the air near the rim is pressed outward and has a pressure gradient, hot gas is pushed towards the axis, and a flame grows a plume pointing inward, towards the station’s centre. The floor that cannot be told from gravity said that an accelerated floor is locally indistinguishable from a gravitating one, and a flame agrees. It also notices, as nothing on Earth does, that the rotating floor carries a Coriolis force, which deflects the rising plume sideways in proportion to its speed; a candle on a slowly spinning habitat would lean, slightly, in the direction of spin, because gas moving towards the axis carries more sideways speed than the air it moves into.

What the estimates cannot show

The speeds, the Nusselt correlation and the burning constant are rough or borrowed. The correlation is for a heated solid sphere in still air, not for a flame, which releases heat throughout a region and changes the composition of the gas it heats; it gives the right dependence on gravity but not a precise burning rate for a candle. The diffusivity used is a single mean value over gas whose temperature varies from 300 to 2,000 kelvin. The oxygen profile assumes a flame that consumes oxygen perfectly at a fixed radius and ignores the outward flow of the flame’s own products, which in a real diffusion flame pushes against the oxygen coming in. A candle is not a sphere: its wick, its pool of wax and its body make the flame a hemisphere at best.

None of the figures attempts the chemistry: the soot formation that makes a terrestrial flame yellow, the radiative losses that make large diffusion flames go out, or the cool-flame chemistry that can persist after they do. Those are the subjects of the combustion experiments that free fall made possible, and their results are not deducible from a picture of how oxygen arrives.

Still open: how a flame that radiates its heat away decides to go out

A spherical diffusion flame grows as its fuel source grows, and its heat release grows with its radius while its radiative loss grows with its volume. Beyond some size it radiates heat faster than it releases it and goes out. That radiative extinction limit is unobservable on the ground, where buoyancy keeps every flame small and quick, and it has been measured only in orbit; how it depends on the fuel, on the oxygen concentration, and on the soot the flame forms, and what then happens in the slower chemistry that sometimes continues after the visible flame goes out, are still being worked out. The answers matter for fire safety in spacecraft, where a smouldering fire in still air behaves unlike any fire a terrestrial detector was designed for.

A flame on the ground is fed by its own plume, a draught of about 0.7 m/s for a centimetre-sized flame that exists only because hot gas is buoyant, and buoyancy is the weight of displaced air; in free fall, where the equivalence principle makes everything fall together, the draught vanishes and oxygen arrives only by diffusion, about 1 cm/s — the two cross near 2 × 10⁻⁴ g, so the Moon’s sixth of g barely touches a flame while orbit’s millionth removes its up entirely, leaving a dim blue sphere in a shell of its own exhaust that burns at about half the terrestrial rate, and a burning droplet whose area shrinks at a constant 0.8 mm²/s. The candle’s teardrop is not the fire’s shape. It is gravity’s.

Part 9 of 9

This essay is one argument about Equivalence principle. The others:

The objects named here

The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.

BuoyancyDiffusionDiffusion flameEquivalence principleFree fallNatural convection