An Aeolipile is a hollow rotating sphere fitted with two opposed bent nozzles that vents steam tangentially, producing reaction torque that spins the sphere on its axis. Demonstration builds typically run between 200 and 1500 RPM depending on boiler pressure and nozzle area. The device exists to convert thermal energy into rotary motion using pure jet reaction, with no pistons, valves, or gears. Hero of Alexandria described it around 70 AD, and modern versions appear in physics classrooms and steam-museum exhibits like the Kew Bridge Steam Museum collection.
Aeolipile Interactive Calculator
Vary jet force, lever arm, nozzle alignment, and pivot drag to see aeolipile reaction torque and wasted radial force.
Equation Used
The aeolipile torque calculation uses the article relationship tau = 2 x F x r, because two opposed tangential steam jets both add torque about the pivot. This calculator also shows how a nozzle angle error reduces useful tangential force by cos(alpha), while the off-tangent component becomes wasted bearing load.
- Two opposed nozzles produce additive torque.
- Nozzle angle alpha is the error away from perfectly tangential flow.
- Jet force is per nozzle.
- Pivot drag is subtracted from generated torque.
The Aeolipile in Action
Boil water inside a sealed sphere, let the steam escape only through two L-shaped nozzles pointing in opposite tangential directions, and Newton's third law does the rest. The steam jets push outward, and an equal and opposite reaction torque spins the sphere on its bearings. There is no crankshaft, no piston, no valve timing — the only moving part is the sphere itself. That is why the Aeolipile is the cleanest physical demonstration of jet reaction you can build in a workshop.
The geometry matters more than people expect. The nozzle exit must point exactly tangential to the sphere — not 5° off, not angled upward. Any radial component of the jet wastes thrust as a force pulling on the bearing instead of producing torque. The nozzle bore also has to be small relative to the boiler outlet, typically 1-3 mm for a desktop unit, so the boiler maintains pressure rather than depressurising in one gulp. If you drill the nozzles too large you get a brief puff of steam and the sphere barely twitches. Drill them too small and the boiler over-pressures before any rotation starts.
Failure modes are simple but unforgiving. The bearing or pivot — usually a needle in a cup, or a brass tube on a polished pin — has to run almost frictionless. Any drag at the pivot eats your torque budget directly because the available reaction force is small, often well under 1 N for a classroom build. Steam leaks anywhere except the nozzles drop the effective exhaust velocity and kill RPM. And if the seal between the boiler and the rotating sphere is not steam-tight, you lose pressure faster than the nozzles can convert it to thrust.
Key Components
- Hollow sphere or rotor body: Acts as the pressure vessel and the rotating mass. Wall thickness of 0.8-1.5 mm in copper or brass is typical for a 60-80 mm diameter demonstration unit. The sphere must be balanced — even a 2 g offset at 50 mm radius produces noticeable wobble above 600 RPM.
- Tangential nozzles (2 off, opposed): Bent tubes with bore 1-3 mm that direct steam exit exactly perpendicular to the sphere's radius. The two nozzles point in opposite tangential directions so their reaction torques add. Bore tolerance matters — if one nozzle is 1.5 mm and the other 2.0 mm, the torques unbalance and the sphere wobbles laterally on its bearings.
- Pivot and bearing: Supports the sphere with the lowest possible friction. A hardened needle in a polished cup is the classic Hero arrangement. Bearing drag torque must stay below roughly 5% of the reaction torque or rotation never starts.
- Steam supply / boiler: Provides saturated steam at 30-150 kPa gauge for desktop units. The boiler-to-sphere coupling must be steam-tight while still allowing free rotation — typically a hollow shaft with a small clearance gap, or a fixed boiler feeding into the pivot itself.
- Heat source: Spirit lamp, gas burner, or electric element delivering 200-800 W to the boiler. Below about 200 W a small unit cannot generate steam faster than it leaks out, and the sphere never reaches stable RPM.
Who Uses the Aeolipile
The Aeolipile is not a power plant — it never was. The thermal efficiency is fractions of a percent because the steam jet exits the nozzle at a speed far above the rim speed of the sphere, so most of the kinetic energy in the jet is dumped to atmosphere unused. What it is good for is demonstrating jet reaction, teaching Newton's third law, and serving as a working historical artefact. You see it today in physics classrooms, science museums, steampunk art pieces, and the occasional engineering-history reproduction.
- Education: PASCO and Arbor Scientific sell classroom Aeolipile kits used in high-school physics labs to demonstrate reaction propulsion.
- Museums: Working reproductions on display at the Thinktank Birmingham Science Museum and the Kotsanas Museum of Ancient Greek Technology in Athens.
- Engineering history reconstructions: University mechanical engineering departments — including projects at Stony Brook and the Technical University of Crete — build Aeolipiles as part of ancient-technology coursework.
- Maker and steampunk art: Hand-built brass Aeolipiles featured at Maker Faire Bay Area and on builds documented by channels like NightHawkInLight as standalone kinetic sculptures.
- Aerospace pedagogy: Used as the introductory example in jet-propulsion textbooks — including Hill and Peterson's Mechanics and Thermodynamics of Propulsion — to introduce reaction thrust before scaling to actual turbojets.
- Film and theatre props: Spinning steam props for period stage productions and museum dioramas where a visible, working steam device adds authenticity.
The Formula Behind the Aeolipile
The torque produced by an Aeolipile is the reaction force from each nozzle multiplied by its lever arm to the rotation axis, summed over both nozzles. The reaction force itself is the mass flow rate of steam through the nozzle multiplied by the exhaust velocity. At the low end of a typical desktop operating range — say 30 kPa boiler gauge pressure with 1 mm nozzles — you get torques in the 0.5-1 mN·m range and the sphere lazily turns at maybe 200 RPM. At nominal 80 kPa with 2 mm nozzles, torque rises to several mN·m and you see 600-900 RPM. Push the boiler to 150 kPa with larger nozzles and the sphere screams past 1500 RPM, but you start losing pressure faster than the boiler can replace it and the run-time per fill drops to seconds. The sweet spot for a teaching demo is the middle band — visible, audible, sustained.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| τ | Reaction torque on the sphere | N·m | lbf·ft |
| ṁ | Mass flow rate of steam through one nozzle | kg/s | lb/s |
| ve | Steam exhaust velocity at the nozzle exit | m/s | ft/s |
| r | Radial distance from rotation axis to nozzle exit | m | ft |
Worked Example: Aeolipile in a brass classroom Aeolipile demo
You are building a brass Aeolipile for a university physics demo. The sphere has a 70 mm outer diameter, two opposed nozzles with 2 mm bore exiting at r = 40 mm from the spin axis. The boiler runs at 80 kPa gauge (saturated steam at roughly 117 °C). You want to predict the reaction torque so you can pick a pivot bearing whose drag is well below the available torque budget.
Given
- Dnozzle = 2 mm
- r = 0.040 m
- Pgauge = 80 kPa
- ve (estimated for saturated steam through short nozzle) = 450 m/s
- ρsteam at exit = 0.6 kg/m³
Solution
Step 1 — at nominal 80 kPa, calculate the nozzle exit area for one 2 mm bore nozzle:
Step 2 — calculate mass flow through one nozzle, ṁ = ρ × A × ve:
Step 3 — apply the torque formula τ = 2 × ṁ × ve × r:
That is the nominal value at 80 kPa. At the low end of a typical classroom run, 30 kPa gauge, exhaust velocity drops to roughly 280 m/s and steam density at exit drops with it, giving:
At 7.9 mN·m the sphere starts turning slowly — maybe 150-250 RPM — and you can see individual jet pulses if the boiler bubbles unevenly. Push to the high end, 150 kPa with the same 2 mm nozzles, and exhaust velocity climbs toward 580 m/s with denser steam:
That puts the sphere at 1200-1500 RPM, but the boiler now drains in 30-60 seconds because mass flow has more than doubled. The 80 kPa nominal is the sweet spot for a sustained demo — enough torque to overcome any sane pivot drag, with run-times measured in minutes rather than seconds.
Result
Nominal reaction torque is approximately 31 mN·m at 80 kPa boiler pressure. That is enough to spin a 70 mm brass sphere at 600-900 RPM against a typical needle-in-cup pivot, producing a clearly audible whoosh and a visible blur of motion across a lecture hall. The low-end 7.9 mN·m at 30 kPa gives a slow, almost ceremonial rotation around 200 RPM where individual nozzle pulses are audible, while the high-end 76 mN·m at 150 kPa runs hard and fast but empties a 200 ml boiler in under a minute. If you measure noticeably less torque than predicted, check three things in order: nozzle alignment off-tangent (a 10° error costs roughly 1.5% torque per degree squared), pivot drag from a corroded or out-of-round bearing cup, and steam leaks at the boiler-to-sphere coupling that drop effective exhaust velocity below the saturated-steam estimate.
When to Use a Aeolipile and When Not To
The Aeolipile is the simplest possible reaction turbine, but simple is not the same as good. Compared to a real impulse or reaction steam turbine, it loses on every metric except cost, build complexity, and pedagogical clarity. Here is how it stacks up against the two mechanisms a builder might actually consider as alternatives for converting steam to rotation.
| Property | Aeolipile | De Laval impulse turbine | Reciprocating steam piston engine |
|---|---|---|---|
| Typical RPM range | 200-1500 RPM | 10000-30000 RPM | 100-500 RPM |
| Thermal efficiency | <1% | 20-40% | 5-15% |
| Useful shaft power | <5 W demo scale | kW to MW | 100 W to MW |
| Build complexity | Very low — sphere + 2 nozzles + pivot | High — precision blading, nozzle ring | Medium — piston, valve gear, crank |
| Cost (small-scale build) | $30-150 hobby | Not viable at hobby scale | $200-1000 model engineering kits |
| Best application fit | Education, demonstration, art | Power generation, marine propulsion | Vintage locomotive replicas, model boats |
| Maintenance burden | Pivot lube + descale boiler | Blade inspection, lube oil system | Valve timing, seals, lubrication |
Frequently Asked Questions About Aeolipile
That asymmetry almost always traces to one nozzle being slightly off-tangent. If both nozzles point exactly perpendicular to the radius, torque is symmetric. If one points 5° outward of tangent, its reaction force has a radial component that does no useful work and the sphere is effectively running on one and a half nozzles in that direction.
Quick check — pull the sphere off the pivot, sight down each nozzle along its axis, and verify the line passes tangent to the sphere surface. A jeweller's loupe and a steel rule are enough. If you bent the tubes by hand, expect to redo at least one of them.
At desktop scale the most common thief is the pivot bearing, but not in the way people expect. The classic failure is not friction in the cup — it is misalignment between the sphere's centre of mass and the pivot axis. Even a 1 mm offset on a 200 g sphere creates a gravitational torque that competes with your reaction torque on every revolution, and you see a lurching half-rotation instead of smooth spin.
Balance the sphere statically before you blame the bearing. Suspend it on the pivot with no steam and flick it — it should come to rest at random angles, not always the same one. If it always settles heavy-side-down, add solder to the light side until it doesn't.
For demonstration purposes, two opposed single nozzles with bore matched to your boiler output is the right answer. Doubling the nozzle count per side halves the per-nozzle mass flow at constant boiler pressure, which actually drops your exhaust velocity slightly because the boiler holds higher pressure for longer. You gain torque arm symmetry but lose peak RPM.
The exception is if you want longer run-time at lower torque — splitting into 4 smaller nozzles gives a slower, steadier spin from the same boiler charge. Pick based on whether you want the demo to be impressive (2 nozzles, big bore) or sustained (4 nozzles, small bore).
Real Aeolipile nozzles are short, sharp-edged tubes — not the converging-diverging profile that lets steam expand efficiently. The flow separates from the wall almost immediately, and a significant fraction of the available pressure drop goes into turbulence and condensation rather than directed kinetic energy.
Expect 60-75% of the isentropic exhaust velocity in practice. If you're using textbook 1-D compressible flow numbers, multiply by 0.7 before plugging into the torque formula. That single correction explains most of the gap between predicted and measured torque on first builds.
Strictly a demo, and the physics says why. The thermodynamic efficiency limit of any reaction turbine peaks when the rotor rim speed approaches half the jet exhaust velocity. A typical Aeolipile rim speed is maybe 3-5 m/s while jet velocity is 300-500 m/s, so you are running at less than 1% of optimal velocity ratio. Most of the jet's kinetic energy literally flies past the sphere into the room.
To extract useful work you would need either much higher rotor speeds (impossible without bearings designed for 30000+ RPM) or much lower jet velocities (impossible without pressure drops too small to matter). That's why no one ever scaled Hero's design — the math forecloses it.
Two likely culprits, and both relate to the boiler-to-sphere interface. First — water carryover. If the boiler boils too vigorously, droplets enter the nozzles and the two-phase flow has dramatically lower exit velocity than dry steam. The sphere accelerates while steam quality is high, then decelerates when slugs of water arrive. Watch for visible water spitting from the nozzles.
Second — the pivot seal heating up. Many small Aeolipiles use the pivot itself as the steam path. As that joint heats and expands, the running clearance can close to zero or open into a leak. If you see the sphere stall around 30-60 seconds in, time it against a thermal scan — coincidence with pivot temperature peak is the giveaway.
The boiler steam generation rate must exceed peak nozzle mass flow at operating pressure, with margin. Rule of thumb — design steam generation at 1.5× the calculated ṁ from the torque equation. If your two nozzles together pass 1.7 g/s at 80 kPa, your heat source needs to evaporate at least 2.5 g/s of water, which works out to roughly 6 kW of heat into the boiler ignoring losses.
Undersized boilers cause exactly the oscillation you describe — the sphere accelerates, depletes the steam reservoir, decelerates while pressure rebuilds, and surges again. A bigger heat source or a smaller nozzle bore both fix it. Smaller nozzles is usually the cheaper path.
References & Further Reading
- Wikipedia contributors. Aeolipile. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
