A friction machine is a mechanical dynamometer that absorbs the output of a rotating shaft through a calibrated frictional contact — typically a brake band, brake block, or rope wrapped around a flywheel — and measures the resulting torque arm reaction on a spring balance or dead weights. James Watt's contemporary Gaspard de Prony built the original version in 1821 to rate steam engines. The purpose is to load the shaft at a controlled level and convert the friction torque directly into brake power. Workshops still use them today to verify rated output of stationary engines and small industrial gearboxes within ±2% accuracy.
Friction Machine Interactive Calculator
Vary shaft speed, brake force, and torque arm length to see torque and brake power for a Prony-style friction dynamometer.
Equation Used
The friction machine reads a brake force W at a known arm length L, giving torque T = WL. Brake power is torque times angular speed, so BP = 2*pi*N*W*L/60 when N is in rpm.
- SI units: force in newtons, arm length in metres, speed in rpm.
- Brake force acts perpendicular to the torque arm.
- Friction load is steady and the spring balance reading equals the tangential reaction force.
How the Friction Machine Works
A friction machine works by deliberately wasting the engine's output as heat through a controlled brake. The shaft turns a flywheel or brake drum, and a band, rope, or set of wooden blocks clamps around that drum with adjustable tension. Tighten the band and the engine has to work harder to maintain speed — that extra work shows up as a tangential friction force at the drum surface, which tries to drag a torque arm around with the rotation. The arm is restrained by a spring balance or hanging dead weights at a known radius. You read the load, multiply by the arm length, and you have torque. Multiply torque by angular speed and you have brake power.
The design only works if the friction is stable. If the band heats up unevenly the coefficient of friction drifts during the test, and your reading wanders by 10% over a few minutes — useless for a rated power test. That is why every serious friction machine has water cooling on the drum, either a hollow rim filled with circulating water or a continuous trickle from above. The drum surface itself must stay clean and dry on a Prony brake, but soaking wet on a rope brake dynamometer — mixing the two contaminates the friction coefficient and your numbers go to nonsense.
Tolerances matter more than people expect. The torque arm length must be measured from the centre of the shaft to the centre of the spring balance hook within ±1 mm on a 1 m arm — a 5 mm error gives you a 0.5% torque error before you have even read the scale. Brake band tension must be applied through a single adjusting screw or lever, never by hand, because pulse loading the drum creates stick-slip oscillation that you can hear as a low growl and see as a needle jumping ±15% on the spring balance.
Key Components
- Brake Drum or Flywheel: The cylindrical surface keyed to the engine output shaft that the friction element grips. Diameter is typically 300-1000 mm depending on engine size, with a machined surface finish around Ra 1.6 µm. On rope brake dynamometers the drum has flanged sides to keep the rope tracking, and a hollow rim for cooling water.
- Friction Element: Either a flexible steel band lined with brake material, a series of wooden blocks bolted into a hinged hoop (classical Prony design), or a hemp rope wrapped 1-3 turns around the drum. The element converts shaft rotation into a tangential friction force without slipping uncontrollably.
- Torque Arm: A rigid lever extending radially from the friction hoop, usually 0.5-1.5 m long. The arm transmits the friction reaction torque to the load-measuring device. Arm length L is the moment arm in the torque equation, so it must be calibrated and stamped, not estimated.
- Spring Balance or Dead Weights: Measures the force needed to hold the torque arm horizontal against the friction reaction. Spring balances read 0-50 kg or 0-500 N typically, with 1% accuracy. Dead-weight setups are slower but more accurate — used at heritage sites like the Anson Engine Museum for rope brake tests.
- Cooling System: Water jacket, drip tray, or internal water-filled rim that carries off the heat dumped by the brake. A 10 kW engine being braked dumps 10 kW into the drum — that is enough to boil 4 litres of water every 16 minutes, so cooling flow must be sized accordingly or the friction coefficient drifts and the test is invalid.
- Tachometer: Measures shaft speed N in RPM. Modern setups use an optical pickup or hand-held laser tach reading to ±1 RPM. On older Prony brakes a mechanical Veeder counter and stopwatch gave the speed to about ±0.5%.
Industries That Rely on the Friction Machine
Friction machines turn up wherever someone needs a cheap, robust, transparent way to measure shaft power without buying a hydraulic or eddy-current dynamometer. They are slow, hot, and physical — but the reading is unambiguous because every term in the formula is something you can measure with a tape and a spring balance. Engineering schools, heritage workshops, small engine manufacturers, and offgrid pump testers all keep them in service.
- Engineering Education: Undergraduate IC engine labs use a rope brake dynamometer on Kirloskar AV1 single-cylinder diesel engines to demonstrate brake power, indicated power, and mechanical efficiency in a single afternoon test.
- Heritage Engine Preservation: The Anson Engine Museum verifies the rated output of a 1948 Lister CS 6/1 stationary diesel using a hemp rope brake on the engine's original flywheel.
- Agricultural Machinery: Small workshop testing of rebuilt Lister D and Petter A1 stationary engines used for pumping and grinding in off-grid agricultural sites — a Prony brake on the flywheel confirms rated 1.5-3 hp output before shipping.
- Marine Auxiliary Power: Bench testing of rebuilt Yanmar L100 and Kubota EA300 single-cylinder diesels for fishing vessel auxiliaries, where a workshop-built rope brake confirms power within 5% of the manufacturer curve.
- Wind and Hydro Microgeneration: Field measurement of small Pelton wheel and Banki cross-flow turbine output at remote micro-hydro sites in Nepal and the Andes, where electrical loadbanks are unavailable and a Prony brake on the turbine shaft is the only practical method.
- Pump and Compressor Manufacturing: Verifying input shaft power of belt-driven irrigation pumps at small Indian and African manufacturers, where IS 11724 calls for shaft power measurement and a calibrated Prony brake satisfies the standard at a tenth the cost of an eddy-current unit.
The Formula Behind the Friction Machine
The brake power formula is the heart of every friction machine reading. What changes across the operating range is which terms dominate the uncertainty. At the low end of the range — light load, low RPM — the spring balance reading is small and a 0.5 N zero error swamps your 5 N reading, so accuracy collapses. At the high end the friction is stable and the reading is large, but heat dissipation becomes the limit and the drum starts cooking the brake band. The sweet spot sits at roughly 60-80% of the engine's rated load and rated speed, where the spring reading is well above noise but the drum stays below 80°C with normal cooling.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| BP | Brake power absorbed by the friction machine | W | hp |
| N | Shaft rotational speed | RPM | RPM |
| W | Net load on spring balance (dead weight minus spring reading on a rope brake) | N | lbf |
| L | Effective torque arm length, centre of shaft to centre of load measurement | m | ft |
| π | Mathematical constant (3.14159…) | — | — |
Worked Example: Friction Machine in a small hydro turbine field test in Nepal
A micro-hydro field engineer in the Annapurna foothills is rating a locally-fabricated 5 kW Pelton wheel driving a workshop-built rope brake dynamometer. The brake drum is 400 mm diameter with a 25 mm hemp rope wrapped twice around it. The torque arm — measured from shaft centreline to the spring balance hook — is L = 0.812 m. At nominal flow the turbine settles at N = 720 RPM with a dead weight of 8.0 kg suspended on one side and a spring balance reading of 1.2 kg on the other side.
Given
- Nnom = 720 RPM
- Dead weight = 8.0 kg
- Spring reading = 1.2 kg
- L = 0.812 m
- g = 9.81 m/s²
Solution
Step 1 — compute the net load W on the rope. Net load is dead weight minus spring reading, converted to newtons:
Step 2 — at nominal 720 RPM, plug into the brake power formula:
That sits at about 82% of the turbine's 5 kW nameplate — exactly where you want a Pelton wheel to run. The drum will be warm to the touch but not steaming, and the spring balance needle will sit rock-steady because the rope is well-loaded.
Step 3 — at the low end of the typical test range, the engineer throttles the inlet nozzle so the wheel runs at 360 RPM with the same rope tension. Spring reading rises slightly to 1.8 kg because the rope grips harder at low speed:
At 1.86 kW the turbine is loafing — the jet is half-open and most of the available head is wasted as splash. Useful for plotting the lower end of the power curve, useless as a rated-output test.
Step 4 — at the high end, opening the nozzle fully drives the wheel to 880 RPM, but the rope starts smoking after 90 seconds and the spring reading bounces between 0.8 and 2.0 kg as stick-slip sets in:
The number looks plausible but it is not trustworthy — a bouncing spring balance means the friction coefficient is no longer steady, and the heat is overrunning the rope's ability to shed it. Wet the rope, slow the wheel, and rerun the nominal test.
Result
Nominal brake power is 4. 08 kW at 720 RPM with a 66.7 N net rope load. That number tells the engineer the Pelton wheel is delivering roughly 82% of its 5 kW nameplate at the current head and flow — a healthy result for a locally-cast bronze runner. The low-end reading of 1.86 kW at 360 RPM and the unstable high-end reading near 4.83 kW at 880 RPM bracket the operating window, with the sweet spot clearly around 700-750 RPM. If a measured value comes in 15-20% below predicted, suspect three things first: rope wrap angle has slipped from 720° to under 540° because the rope walked off the drum flange, the dead weight pan is swinging instead of hanging still and shedding 10% of the recorded load, or the torque arm L has been measured to the rope contact point instead of the shaft centreline — a common 25-50 mm error on a 400 mm drum that biases every reading by 3-6%.
Friction Machine vs Alternatives
A friction machine is the cheapest way to measure shaft power, but it is not always the right way. The honest comparison is against the two methods that compete with it directly in the small-to-medium power range — the hydraulic water brake, and the eddy-current dynamometer. Each wins on different axes.
| Property | Friction Machine (Prony / Rope Brake) | Hydraulic Water Brake | Eddy-Current Dynamometer |
|---|---|---|---|
| Typical accuracy | ±2-5% | ±1-2% | ±0.25-0.5% |
| Useful power range | 0.5-50 kW | 5 kW-2 MW | 1 kW-500 kW |
| Maximum continuous speed | 1500-2000 RPM | 8000-10000 RPM | 8000-15000 RPM |
| Capital cost (relative) | 1× (baseline, can be shop-built) | 8-15× | 20-50× |
| Heat dissipation method | Air + water trickle on drum | Bulk water flow | Forced air or water jacket on stator |
| Response to rapid load change | Slow, prone to stick-slip | Medium, valve-limited | Fast, electronic control |
| Maintenance interval | Re-line band every 20-50 hours | Seal service annually | Bearing service every 2000+ hours |
| Best fit application | Heritage engines, micro-hydro, teaching labs | Production engine test cells | Powertrain R&D, EV motor test |
Frequently Asked Questions About Friction Machine
The friction coefficient of the brake material is dropping as the drum heats up. Cast-iron-on-wood and steel-on-leather both lose roughly 20-30% of their cold µ between 20°C and 150°C, and a 5 kW brake with no water cooling reaches that temperature inside 3-4 minutes.
Fix it by adding a water drip on the drum at about 0.5 L/min for every kW absorbed, or by switching to a rope brake with a hollow water-cooled rim. If the drift continues after cooling is added, check whether the brake band lining is glazing — a shiny black surface means the binder has melted and the lining must be sanded back to fresh material before the next test.
Rope brake wins for steady-state rated-power testing because the rope's elasticity damps the firing-stroke torque pulses of a single-cylinder engine. A rigid band-type Prony brake transmits those pulses straight into the spring balance, which then bounces ±20% and you cannot read it.
Band brakes win when you need to vary load quickly during a test — tightening a wing-nut adjuster takes seconds, while changing rope tension means swapping dead weights. For a 10 kW Lister or Petter type engine on a heritage rated-power verification, fit a 2-turn hemp rope on a water-cooled drum and accept the slow load adjustment.
Mechanical efficiency is BP divided by indicated power IP, so a value above 100% means either BP is overestimated or IP is underestimated. The usual culprit is the indicator diagram — a worn indicator spring or a leaky indicator cock gives a low-area diagram and a low IP, making the ratio impossible.
Check the indicator spring calibration first against a dead-weight tester. If the spring is fine, look at the planimeter trace on the indicator card — many students measure the area in mm² but forget to multiply by both the spring scale and the stroke scale, which can drop IP by a factor of 2.
Wet hemp grips the cast-iron drum harder than dry hemp — the coefficient of friction rises from about 0.25 dry to 0.35-0.40 wet. So for the same shaft torque, more of that torque is reacted by the rope-drum contact and less by the falling dead weight, which means the spring balance reading on the rising side increases.
This is not an error, it is exactly why rope brakes are run wet — the higher and more stable µ gives a steadier reading. Just make sure you record the wet-running reading consistently and never mix wet and dry data points on the same power curve.
Run time is limited by heat, not by mechanical wear. A simple rule: with no cooling, a dry Prony brake gives reliable readings for about 60 seconds per kW of brake power before the lining temperature exceeds 150°C and µ drops noticeably. So a 5 kW test gives you about 5 minutes of trustworthy data on a cold start.
With a water-cooled rim sized for the rated power, run time is effectively unlimited as long as cooling water keeps flowing. The diagnostic check is simple — touch the drum every 2 minutes with a wet rag. If the rag sizzles, you are over 100°C and the test is already drifting.
The brake hoop is hanging off-centre. On a properly balanced Prony brake the hoop's centre of gravity sits directly under the shaft centreline so gravity adds nothing to the torque reaction. If the hoop is hung lopsided the dead weight of the hoop itself shows up as a constant offset that adds to one direction of rotation and subtracts from the other.
Check by removing the engine drive belt and rotating the shaft slowly by hand in both directions — the spring should read the same small breakaway value either way. If it reads 5 N one way and 12 N the other, re-shim the hoop until the difference is under 1 N.
References & Further Reading
- Wikipedia contributors. Prony brake. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
