An aeolipile is a hollow rotating sphere fitted with two bent nozzles that vent steam tangentially, generating reaction thrust that spins the sphere on its bearings. Hero of Alexandria described it around 60 AD in his Pneumatica, making it the earliest documented steam-powered device. It converts thermal energy in pressurised steam into rotational kinetic energy without pistons, valves, or gears. A small classroom build with a 60 mm sphere and 2 mm nozzles will hit 1,500 RPM unloaded — useful as a teaching tool, not as a power source.
Aeolipile or Hero's Steam Engine Interactive Calculator
Vary steam jet flow, velocity, sphere diameter, and nozzle angle to see thrust loss and spin torque.
Equation Used
This calculator applies the aeolipile reaction equation F = mdot v to two identical nozzles, then reduces the useful tangential thrust using the article's teaching approximation that a 10 deg nozzle error causes about 15% thrust loss. The resulting useful force times the sphere radius gives the ideal spin torque before bearing friction and air drag.
- Mass flow is entered per nozzle and the aeolipile has two matched nozzles.
- Nozzles are intended to be tangent to the rotation circle.
- Angle loss uses the article teaching approximation: 10 deg off tangent gives about 15% thrust loss.
- Bearing friction, windage, pressure drop, and acceleration inertia are not included.
How the Aeolipile or Hero's Steam Engine Actually Works
The aeolipile works on Newton's third law. You heat water inside a sealed hollow sphere, the water boils, steam pressure builds, and the steam escapes through two L-shaped nozzles mounted on opposite sides of the sphere's equator. Because the nozzles point tangentially — not radially — the escaping jets push the sphere in the opposite direction, exactly like a lawn-sprinkler arm. No pistons, no valves, no gears. Just steam jet thrust acting on a rotating sphere boiler.
The geometry matters more than people think. The nozzle exit must sit perfectly tangent to the rotation circle — if you angle it 10° off tangent you bleed roughly 15% of the thrust into a useless radial component, and the sphere barely turns. The two nozzles must also point in opposite tangential directions; if a student solders one backwards (a classic teaching-lab mistake) the jets cancel and the sphere just sits there hissing. Bearing friction is the other killer. The sphere typically pivots on two hollow trunnions that double as the steam-feed path from an external boiler, or sits on simple pin bearings if it's self-contained. Any binding here drops top speed dramatically — a sphere that should run at 1,500 RPM will struggle past 400 RPM with a sticky bearing.
Failure modes are simple but unforgiving. Insufficient water and you boil dry, the sphere overheats, and the solder joints on the nozzles let go. Too much water and there's no steam volume — you get a few weak puffs and nothing. Nozzle bore inconsistency between the two sides causes the sphere to wobble and precess on its bearings instead of spinning cleanly. The aeolipile is mechanically trivial but thermally fussy.
Key Components
- Hollow sphere (boiler): The pressure vessel that holds water and generated steam. Typical demonstration builds use a 50-80 mm copper or brass sphere with wall thickness of 0.8-1.2 mm. Wall thickness below 0.6 mm risks deformation under pressure; above 1.5 mm and thermal mass slows steam generation noticeably.
- Tangential nozzles: Two bent tubes — typically 2-3 mm bore — soldered through opposite sides of the sphere and bent 90° so the exit points tangent to the rotation circle. Both nozzles must have matched bores within ±0.1 mm or the sphere wobbles. Bore mismatch is the single biggest cause of poor performance in classroom builds.
- Trunnion bearings: Two hollow pins that support the sphere and let it rotate freely. In tethered designs the trunnions also carry steam from an external boiler into the sphere. Friction here directly caps top speed — even light gummy oil cuts RPM by half.
- Heat source: An open flame or alcohol burner under the sphere, or steam delivery from a separate boiler. A typical 100 ml-water sphere needs roughly 200-400 W of heat input to reach steady-state spin within 60-90 seconds.
- Support frame: Holds the trunnion bearings rigid and isolates them from the heat source. Misalignment of the bearing axis by more than about 0.5° introduces wobble that drops RPM and accelerates bearing wear.
Real-World Applications of the Aeolipile or Hero's Steam Engine
The aeolipile never did useful work in antiquity and it doesn't do useful work today — its torque output is far too low and its efficiency sits below 1%. But it earns its keep as a demonstration steam engine, a physics teaching aid, and a historical artefact. You'll see aeolipiles in science museums, university thermodynamics labs, steampunk maker projects, and the occasional working-replica garden ornament.
- Science education: Thermodynamics lab demonstrations at universities — the MIT Edgerton Center has used aeolipile demos to show reaction-thrust principles before introducing actual turbine theory.
- Museum exhibits: The Thessaloniki Technology Museum's Ancient Greek Technology collection includes a working aeolipile reconstruction operated for visitors several times daily.
- Maker and hobby builds: Steampunk artists and hobbyists build tabletop aeolipiles using copper plumbing fittings — typically a 1.5-inch copper end-cap sphere with brass tube nozzles.
- Physics classroom kits: Aeolipile demonstration kits sold by educational suppliers like American Science & Surplus and Edmund Scientific, used to show Newton's third law in a more dramatic form than balloon rockets.
- Historical reconstruction: Working replicas built for documentaries on ancient Greek and Roman engineering, including BBC and Discovery Channel productions on Hero of Alexandria's Pneumatica.
- Engineering education: Introductory courses on turbomachinery use the aeolipile as the conceptual ancestor of modern reaction turbines like the Parsons steam turbine.
The Formula Behind the Aeolipile or Hero's Steam Engine
The aeolipile is a pure reaction device, so the torque on the sphere comes from the mass flow of steam leaving each nozzle multiplied by the exit velocity and the nozzle radius from the rotation axis. This formula tells you the theoretical torque, and from there you can estimate unloaded RPM. At the low end of the practical range — small 40 mm spheres with 1.5 mm nozzles and a weak alcohol burner — you'll see torque in the single-digit µN·m range and modest RPM. At the high end, a 100 mm sphere fed by an external boiler at 2 bar gauge can push toward 3,000 RPM unloaded. The sweet spot for classroom demos sits around a 60-80 mm sphere with 2 mm nozzles, producing visible, audible, fast spin without the thermal hazards of higher-pressure setups.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| τ | Torque on the sphere | N·m | lb·ft |
| ṁ | Mass flow rate of steam through one nozzle | kg/s | lb/s |
| ve | Steam exit velocity at the nozzle | m/s | ft/s |
| rn | Radius from rotation axis to nozzle exit | m | ft |
| 2 | Factor for two nozzles acting symmetrically | dimensionless | dimensionless |
Worked Example: Aeolipile or Hero's Steam Engine in a classroom aeolipile demonstration
You are building a tabletop aeolipile for a high school physics demonstration. The sphere is a 70 mm copper plumbing cap soldered into a hollow ball, with two 2 mm-bore brass nozzles bent tangentially at a radius of 35 mm from the rotation axis. An alcohol burner delivers enough heat to vaporise water at roughly 0.15 g/s per nozzle, and the steam exits at approximately 180 m/s at atmospheric pressure.
Given
- ṁ = 0.00015 kg/s per nozzle
- ve = 180 m/s
- rn = 0.035 m
- Sphere moment of inertia I = ≈ 4.0 × 10-5 kg·m²
Solution
Step 1 — compute the nominal torque at design conditions, both nozzles flowing 0.15 g/s of steam at 180 m/s exit velocity:
That's 1.89 mN·m — small, but acting on a sphere with very low moment of inertia, so RPM builds fast against bearing friction. In practice, an aeolipile of this scale settles at roughly 1,200-1,800 RPM unloaded once thermal steady-state is reached, around 60-90 seconds after the burner lights.
Step 2 — at the low end of the practical operating range, say a weaker burner producing only 0.08 g/s per nozzle and exit velocity dropping to 140 m/s as steam pressure falls:
That's roughly 40% of nominal torque. The sphere still spins, but visibly slower — maybe 500-700 RPM — and any small misalignment in the trunnion bearings can stop it entirely. This is what you see in most first-attempt classroom builds: it spins, but anaemically.
Step 3 — at the high end, an external pressurised boiler delivering 0.25 g/s per nozzle at 240 m/s exit velocity:
Now you're looking at over twice the nominal torque and theoretical RPM well above 2,500. Audibly louder, visibly faster, and you start seeing the sphere precess on its bearings if the trunnions aren't dead-true. Above this point you also need to start thinking about pressure-vessel safety on the sphere itself — 70 mm thin-wall copper is not rated for serious gauge pressure.
Result
Nominal torque comes out to 1. 89 mN·m, which drives the 70 mm sphere to roughly 1,200-1,800 RPM unloaded — a fast, audible spin you can clearly see and hear from across a classroom. At the low end (0.78 mN·m) the sphere creeps along at 500-700 RPM and looks underwhelming; at the high end (4.2 mN·m) it screams past 2,500 RPM and starts wobbling if the bearings aren't precise. If you measure significantly less spin than predicted, three failure modes dominate: nozzle bore mismatch greater than 0.1 mm between the two sides, which throws the thrust off-balance and dumps energy into wobble; non-tangential nozzle alignment where one or both exits are angled even 5-10° off the true tangent, costing 10-15% of useful thrust each; and trunnion bearing drag from carbon buildup or solder splash, which can halve top RPM in extreme cases.
Choosing the Aeolipile or Hero's Steam Engine: Pros and Cons
The aeolipile is the simplest possible steam engine, and that simplicity is both its appeal and its limitation. It's worth comparing against the two devices people most often confuse it with — the piston steam engine and the modern reaction turbine — on the engineering dimensions that actually matter for choosing one.
| Property | Aeolipile | Piston steam engine | Reaction steam turbine |
|---|---|---|---|
| Typical operating speed | 500-3,000 RPM unloaded | 60-600 RPM | 3,000-30,000 RPM |
| Thermal efficiency | < 1% | 5-15% | 30-45% |
| Useful shaft power | Negligible (mW range) | kW to MW | MW to GW |
| Mechanical complexity | Trivial — 4-5 parts | High — pistons, valves, crankshaft, governor | Very high — precision blades, multiple stages |
| Manufacturing cost (demo unit) | $20-50 from copper fittings | $500-5,000 model engine kits | Not available at hobby scale |
| Primary application fit | Education, demonstration, art | Locomotives, traction engines, model engineering | Power generation, marine propulsion, aviation |
| Lifespan in continuous use | Tens of hours before solder fatigue | Decades with maintenance | Decades with maintenance |
Frequently Asked Questions About Aeolipile or Hero's Steam Engine
This is almost always uneven nozzle output. When the sphere is cold, condensation inside one nozzle can partially block it, so the other nozzle dominates and you get net rotation. As steam fully clears the blocked nozzle, the two thrusts equalise and net torque drops to zero. If one nozzle's bore is slightly larger than the other from manufacturing variation, you can even see net thrust reverse direction because relative flow shifts.
Diagnostic check: pull the sphere off, blow through each nozzle with the same source pressure, and compare flow rates. If they differ by more than ~10%, ream both to a matched bore.
Realistically, no. The aeolipile's reaction-jet geometry is enormously inefficient because the steam leaves the nozzle at high velocity relative to the ground, carrying most of its kinetic energy away with it instead of transferring that energy to the sphere. Real reaction turbines solve this with multiple stages and shaped blades that extract energy across a velocity gradient.
If you want hobby-scale shaft power from steam, build an oscillating-cylinder piston engine instead. Even a small one will deliver 100× more useful torque than an aeolipile of similar size.
Three usual suspects beyond nozzle mismatch: the centre of mass is offset from the rotation axis (uneven solder buildup on one side adds surprising mass at this scale), the trunnion bearings aren't coaxial within about 0.5°, or the water inside is sloshing because there's too much of it. The sloshing case is sneaky — a half-full sphere's centre of mass shifts as it rotates, creating a forced precession that looks exactly like bearing misalignment.
Fix order: drain to roughly 30% fill, check trunnion alignment with a straight-edge across both bearing centres, then weigh-test by spinning the sphere slowly by hand on its bearings cold and watching for a preferred resting orientation.
Rule of thumb from working classroom builds: nozzle cross-sectional area should total roughly 0.05-0.15% of the sphere's internal surface area. For a 70 mm sphere (≈15,400 mm² internal surface), that means total nozzle area of 8-23 mm², or two nozzles of 2.3-3.8 mm bore.
Go too small and pressure builds but mass flow is choked, producing a quiet, weak spin. Go too large and you can't maintain pressure — the sphere just hisses and barely turns. The 2-3 mm bore range hits the sweet spot for typical alcohol-burner heat input.
You've boiled off most of the water. Aeolipiles have very small water charges (often under 50 ml) and once water level drops below the nozzle's internal pickup point, steam generation falls off a cliff because you've lost most of the wetted heat-transfer area on the sphere's inner wall. Steam quality also drops — you start getting superheated low-density steam instead of saturated steam, which means lower mass flow and lower thrust.
If you need a longer demo run, use an external boiler with a continuous steam feed through the trunnions, or simply use a larger sphere with more water. A 100 mm sphere holding 200 ml will run several minutes at full spin.
The honest answer is we don't know. Hero's Pneumatica (circa 60 AD) is the earliest surviving description, and Vitruvius mentioned a similar device about a century earlier without crediting an inventor. There's no evidence aeolipiles were used for anything beyond temple curiosities and demonstrations in the ancient world — no agricultural, military, or industrial application has been documented.
For a working engineer this matters because it explains why the design never evolved: nobody in antiquity had a problem an aeolipile could solve, and the metallurgy and machining needed to build a useful steam engine wouldn't exist for another 1,600 years.
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.
