A Tread Mill is a large wheel turned by humans or animals walking on its inner or outer rim, converting body weight and stride into rotary motion. The Roman polyspaston crane used twin treadwheels to lift stones up to 6,000 kg on Trajan's Forum. The wheel exists to turn muscle endurance into torque at a useful shaft speed for grinding, pumping or hoisting. A trained ox on a 4 m treadwheel produces around 0.5 kW continuously — enough to run a small grist mill all day.
Tread Mill Interactive Calculator
Vary walker mass, wheel diameter, and off-center offset to see treadwheel torque, rim force, and how much of the wheel radius is being used.
Equation Used
The treadwheel torque is the walker's weight multiplied by the horizontal offset from the axle. A larger mass or larger offset gives more shaft torque; dividing by wheel radius estimates the equivalent tangential force at the rim.
- Torque is produced by the walker's weight acting at a horizontal offset from the axle.
- Gravity is 9.80665 m/s^2.
- Offset is measured horizontally from the axle to the walker's center of mass.
- Losses from bearings, slip, and gearing are not included.
How the Tread Mill Works
A Tread Mill works on a simple gravity-and-leverage principle. The walker stands off-centre on the wheel rim, their weight produces a torque about the axle equal to body weight times the horizontal offset from the axle, and the wheel rotates to keep them from falling. As the wheel turns, the walker steps forward to maintain that offset, and the torque stays roughly constant for as long as they keep walking. You would be amazed how little the walker actually 'climbs' — over an hour of work the body rises and falls maybe 50 mm net, while the wheel completes hundreds of revolutions. That's the trick of the treadwheel: continuous rotary output from what feels like flat walking.
The geometry matters more than the muscle. If you build the wheel too small, the walker's offset is limited and torque drops — a 1 m diameter dog wheel maxes out around 50 N·m even with a 30 kg dog, while a 4 m human treadwheel crane easily delivers 800 N·m with two operators. If the rim is too narrow or the cleats too shallow, the walker slips on the descending side and the wheel runs away — a runaway treadwheel crane killed and maimed crane operators across medieval Europe, which is why the brake band on the axle is not optional. Cleat spacing should match the walker's natural stride: 350-450 mm for humans, 200-280 mm for dogs, 500-700 mm for oxen. Get this wrong and the walker either stutter-steps or over-strides, both of which collapse efficiency.
The most common failure modes are bearing seizure on the axle journals (wooden bearings need pitch lubrication every few hundred hours), cleat shear at the rim joint, and brake-band glaze when the descent rate exceeds what friction can absorb. If the wheel starts squealing under load, you're hearing the journal — stop and re-pitch before it scores.
Key Components
- Rim and tread surface: The walking surface that carries the load. Width typically 600-1200 mm for humans, 300-400 mm for dogs. Cleats or cross-battens are spaced to match stride length and prevent slipping — for a human wheel, 400 mm spacing is the standard.
- Axle and journal bearings: The shaft transmits torque to the load — gears, a crank, or a rope drum. Historic wheels used iron-shod wooden journals running in greased oak bearings, replaced every 2-5 years. The axle must be sized so deflection under full eccentric load stays below 1/1000 of span.
- Spokes or wheel webs: Connect rim to axle and carry the eccentric load. Compression spokes (tangential bracing) handle the torque; radial spokes carry the radial weight. A 4 m diameter wheel typically uses 8-12 spokes of 80-100 mm oak.
- Brake band or pawl: Stops a runaway. A friction band on the axle drum, hand-operated, dissipates the kinetic energy if the walker stops or falls. A pawl-and-ratchet on the drive drum prevents reverse rotation under load — mandatory on any hoisting treadwheel.
- Drive output (drum, pinion, or crank): Couples the wheel shaft to the load. A rope drum for hoisting, a lantern pinion for milling, or an offset crank for pumping. Gear ratio is set so the walker's natural pace (about 3 km/h) produces the load's required input speed.
Where the Tread Mill Is Used
Treadwheels powered the western world for nearly 2,000 years before the steam engine. They show up wherever continuous low-speed shaft power is needed and a steady supply of muscle is cheaper than fuel. You still see them in living-history museums, in some niche modern off-grid setups, and as engineering teaching tools. The mechanism shines when you need torque rather than speed and when the duty cycle is long but not punishing — a draft animal can walk for 8 hours but cannot sprint, and a treadwheel turns that endurance directly into useful work.
- Construction (historical): The Roman polyspaston crane used a pair of human treadwheels to lift columns and architraves on the Pantheon and Trajan's Column — typical lift capacity 3,000-6,000 kg.
- Cathedral construction: The surviving treadwheel crane in the bell tower of Beverley Minster, England, dates to the 14th century and was still used to lift bells into the 1700s.
- Domestic kitchen (historical): Turnspit dog wheels — the 'Canis Vertigus' breed was specifically used in English inn kitchens from the 16th to 19th century to turn meat over a fire on a spit drive.
- Agriculture: Horse engines and dog-powered treadmills drove butter churns, threshers, and feed cutters on American farms through the 19th century — the Wheeler Patent dog power was a common 1880s product.
- Prisons (historical): Sir William Cubitt's prison treadwheel, introduced at Brixton in 1817, used inmate labour to grind grain or simply pump air against a brake — ran at about 50 steps per minute.
- Education and demonstration: Working replica treadwheel cranes at Guédelon Castle in France and at Bardi Castle in Italy are used today both for genuine construction and as engineering demonstrations.
The Formula Behind the Tread Mill
Power output from a treadwheel comes down to walker weight, the horizontal offset from the axle, and walking pace. The formula tells you the continuous shaft power for a given walker on a given wheel, but the interesting question is where in the operating range you actually want to sit. Run a walker at the high end of their pace and they'll fatigue in under an hour. Run them at the low end and you're wasting wheel diameter. The sweet spot for a human is roughly 50-60 steps per minute on a 3-4 m wheel, which lines up with a comfortable downhill walk. Push faster and the walker slides on cleats; go slower and torque ripple becomes audible at the drive drum.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| P | Continuous shaft power output | W | ft·lbf/s |
| m | Walker mass | kg | lb |
| g | Gravitational acceleration | 9.81 m/s² | 32.2 ft/s² |
| d | Horizontal offset of walker from axle (effective lever arm) | m | ft |
| D | Wheel diameter | m | ft |
| vwalk | Walker's tangential speed along the rim | m/s | ft/s |
Worked Example: Tread Mill in a small craft cidery hoist
You are building a working treadwheel hoist for a craft cidery in Somerset that lifts apple bins from the wagon yard up to a second-floor crusher. The wheel is 3.6 m in diameter, the operator weighs 75 kg, and they walk roughly 0.6 m off the axle vertical. You need to know shaft power across the realistic walking pace range so you can size the rope drum and gearing.
Given
- m = 75 kg
- g = 9.81 m/s²
- d = 0.6 m
- D = 3.6 m
- vwalk,nom = 0.9 m/s
Solution
Step 1 — at the nominal walking pace of 0.9 m/s (about 54 steps per minute on a 3.6 m wheel), compute the torque-equivalent term:
Step 2 — multiply by walking speed to get nominal continuous power:
Step 3 — at the low end of the practical range, 0.5 m/s (a slow amble, about 30 steps/min), the same lever-arm term applies:
That's barely enough to lift a 40 kg apple bin at any useful rate — the operator will feel like they're walking on flat ground but the bin will crawl, and rope-drum starting torque becomes marginal because static friction in the journals eats a meaningful chunk of the input.
Step 4 — at the high end, 1.4 m/s (a brisk 80 steps/min):
In theory that's a 56% boost, in practice the operator can sustain it for maybe 10-15 minutes before lactate builds up and they have to stop. Above about 1.2 m/s the walker also starts catching cleats with their toes on the descending side, so the wheel actually loses efficiency to stutter.
Result
Nominal continuous power output is 110 W — about a seventh of a horsepower, which lifts a 40 kg apple bin at roughly 0. 25 m/s through a 1:1 drum, or 0.5 m/s through a 2:1 drum. At the low end (61 W) the hoist feels sluggish and friction losses dominate; at the high end (172 W) the operator can move bins quickly but only in short bursts before fatigue forces a rest. Sit the design at 0.9 m/s and gear the drum so a full-bin lift takes 30-45 seconds. If you measure shaft power 25% below the predicted 110 W on the bench, the most likely causes are: (1) the walker drifting toward the axle vertical so effective d shrinks below 0.6 m, (2) green or undried oak journals binding the axle bearings (you'll hear it before you measure it), or (3) cleat depth worn below 15 mm letting the foot slip on the descending arc, which kills tangential speed without showing up as obvious slippage.
Tread Mill vs Alternatives
A treadwheel is one option for low-speed shaft power. Before committing to one, weigh it honestly against the two mechanisms it most often competes with — a horse-drawn capstan (sweep) and an undershot waterwheel — on the dimensions that actually matter for a working installation.
| Property | Tread Mill (treadwheel) | Horse capstan / sweep | Undershot waterwheel |
|---|---|---|---|
| Continuous power output | 50-800 W (human), 0.4-1 kW (ox) | 0.5-2 kW (horse on a 6 m sweep) | 1-5 kW depending on stream flow |
| Shaft RPM at output | 3-8 RPM at the wheel, geared up as needed | 2-4 RPM at the sweep | 4-12 RPM at the wheel |
| Capital cost (small-scale build) | £3,000-£8,000 in timber and ironwork | £1,500-£4,000 for a sweep and gear pit | £8,000-£20,000 including weir and headrace |
| Site requirement | Indoor floor space, 4-5 m headroom | Open level ground, 8-12 m diameter clearance | Year-round flowing stream with head |
| Duty cycle | 1-2 hours per operator before rotation | 4-6 hours per horse with rest breaks | 24/7 while water flows |
| Lifespan of main structure | 50-200 years (the Beverley Minster wheel is 600+) | 20-40 years before sweep arms need replacement | 15-30 years before paddles and shaft replacement |
| Best application fit | Intermittent hoisting, demos, off-grid hand-power | Continuous low-speed milling, threshing | Continuous milling where water is available |
Frequently Asked Questions About Tread Mill
You're seeing torque ripple from the lever arm changing as the walker's centre of mass swings forward and back during each stride. On a small wheel, every step the walker takes shifts d by 100-200 mm in either direction, and since power scales linearly with d, you get a sinusoidal pulse on the output shaft.
The fix is either a heavier flywheel on the drive shaft to absorb the pulses, or a larger-diameter wheel where stride length becomes a smaller fraction of d. Below about 2.5 m wheel diameter the ripple is genuinely audible at the load — above 3.5 m it smooths out without a flywheel.
Outside-rim (caged) gives you more usable diameter for a given building height and is safer because the walker can't fall in. Inside-rim (drum) is mechanically simpler — the rim itself is the floor — but requires a wheel diameter at least 2.4 m so a person can stand upright, and you need a substantial side wall to keep the walker contained.
For human-scale builds under 3 m, go inside-rim if floor space is tight and outside-rim if headroom is tight. For animal treadmills, almost always outside-rim with an inclined platform variant — the dog or sheep walks on a sloped belt rather than entering the wheel.
Yes, but the gear ratio is brutal. A typical permanent-magnet alternator wants 400-1500 RPM at the rotor, and your wheel is turning at 4-6 RPM. That's a step-up ratio of around 1:200 to 1:300, which means a two-stage gearbox or belt-and-pulley system, and every stage costs 5-10% efficiency.
Realistically a 3.5 m human treadwheel with a healthy operator delivers 80-120 W at the alternator output after losses. Useful for charging a battery bank slowly, not for running a household. If electrical generation is the goal, a bicycle generator is mechanically far simpler and cheaper.
Aim for a wheel diameter at least 4× the walker's standing height for humans, and at least 6× the shoulder height for animals. Below that ratio the walker either has to crouch (humans) or struggles to find a comfortable gait (animals).
For mass, the wheel rim and structure should weigh no more than 50% of the walker mass for the wheel to start easily from rest. A 4 m human treadwheel built in heavy oak might mass 250 kg — a single 75 kg walker can start it but only just. This is why most historical cranes used two operators side by side.
It's not running backwards from the load — the load is overhauling the wheel. If you've got a heavy mass on the drive drum and the walker pauses or steps off-centre toward the axle, the load weight overcomes the walker's torque and pulls the wheel back. This is the classic runaway condition that maimed medieval crane operators.
The mandatory fix is a pawl-and-ratchet on the drive drum that engages automatically and prevents reverse rotation regardless of walker input. Add a hand-operated brake band on the axle as a controlled-descent option. Never operate a hoisting treadwheel without both.
A windlass is cheaper, smaller, and good for short lifts under 50 kg where the operator can put their full body weight on the crank handle. Above that, the human arm fatigues fast — sustained crank power tops out around 60-80 W.
A treadwheel uses leg muscles and bodyweight, which sustains 100-200 W for an hour or more. The break-even is roughly: lifts over 100 kg, or lift heights over 5 m, or duty cycles longer than 10 minutes — pick the treadwheel. Below all three thresholds, the windlass wins on simplicity and cost.
References & Further Reading
- Wikipedia contributors. Treadwheel. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
