A Torsion Dynamometer measures the power transmitted through a rotating shaft by reading the small angle of twist between two points along its length. The instrument was developed and refined by engineers like Bevis and Hopkinson in the late 1800s for measuring marine shaft horsepower at sea, where absorption-style brakes were impractical. It works because torque is proportional to twist angle through the shaft's modulus of rigidity, so a known shaft geometry plus a measured twist gives you torque, and torque times angular speed gives shaft power. Modern versions use strain gauges and rotary telemetry to log shaft horsepower continuously on ships, turbines, and gearboxes.
Torsion Dynamometer Interactive Calculator
Vary shaft power, speed, diameter, and gauge length to see torque, twist angle, stiffness, and sensor resolution update on a rotating shaft diagram.
Equation Used
The torsion dynamometer treats the shaft as a torsion spring. Torque comes from shaft power and speed, then twist follows the proportional relationship theta = T L / GJ. This calculator scales the article worked example using J proportional to d^4, so changing power, RPM, length, or diameter shows how the measured twist changes.
- Uniform circular shaft section within the elastic range.
- Polar moment scales with d^4 for the shaft diameter.
- Twist scaling is calibrated to the article worked example: 200 mm, 5000 kW, 100 RPM, 3 m, theta = 0.4 deg.
- Sensor step shown is theta/400 to match the worked example resolution statement.
How the Torsion Dynamometer Works
A Torsion Dynamometer treats the shaft itself as the spring. When you transmit torque through a circular shaft, the shaft twists by a small angle θ over a measured length L, and that twist angle is linearly proportional to the applied torque as long as you stay below the material's elastic limit. Mount two flanges or two optical discs at a known distance apart on the shaft, measure the relative angular displacement between them, and you have torque without needing to interrupt the drivetrain or absorb the power into a brake. Multiply by shaft speed and you have shaft power in real time.
The geometry has to be right or the reading drifts. The gauge length L must be long enough that the twist angle is large enough to resolve — for a 200 mm diameter steel propeller shaft transmitting 5,000 kW at 100 RPM, total twist over a 3 m gauge length is only about 0.4°, so your angular sensor needs to resolve at least 0.001° to give you 1% accuracy. The shaft must also be uniform along the gauge length: keyways, step changes, or a coupling inside the gauge length will throw off the modulus of rigidity assumption and give you a low reading. We typically tell customers to keep the gauge length on a clean, constant-diameter section at least 5 shaft diameters away from any flange or coupling, because stress concentration at those features distorts the local twist field.
Failure modes are usually wiring or geometry. Strain-gauge based torsion meters fail at the slip ring or at the wireless telemetry battery long before the gauges themselves drift. Optical or magnetic-toothed-wheel types fail when one of the toothed wheels slips on its taper or when shaft runout exceeds the sensor's air gap. If you see torque oscillating at exactly 1× shaft speed, your toothed wheel is eccentric. If you see DC drift with temperature, the shaft modulus of rigidity is shifting — steel drops about 3% per 100 °C, so a hot gearbox shaft reads high if you don't temperature-compensate.
Key Components
- Gauge-length section of shaft: The portion of shaft between the two angular reference points acts as the torsion spring. Length is typically 1-3 m on a marine shaft, chosen so the twist at rated torque is at least 0.2° to give comfortable signal-to-noise. Must be uniform diameter with no keyways or section changes.
- Angular reference flanges or toothed wheels: Two precision-machined discs clamped to the shaft at each end of the gauge length. Concentricity to the shaft axis must be within 0.05 mm TIR, otherwise eccentricity shows up as a 1× rev torque ripple that masks the real signal.
- Angular displacement sensor: Measures the relative twist between the two discs. Older Hopkinson-style instruments used a stroboscopic mirror; modern units use two magnetic pickups reading toothed wheels and measuring the phase shift, with resolution down to 0.001° at 100 RPM.
- Strain-gauge bridge (modern variants): A 4-arm Wheatstone bridge bonded directly to the shaft at 45° to the axis, where principal strain from torsion peaks. Output is typically 2 mV/V at full-scale torque, with bridge excitation at 5-10 V.
- Rotary signal coupling: Either slip rings, rotary transformer, or wireless telemetry transmits the bridge signal off the rotating shaft. Wireless modules with onboard battery are now standard above 200 RPM where slip-ring brush life becomes a problem.
- Speed pickup: An independent magnetic or optical sensor reads shaft RPM, since power requires both torque and angular velocity. Usually integrated into one of the toothed reference wheels to save a sensor mount.
Who Uses the Torsion Dynamometer
Torsion Dynamometers earn their keep wherever you need to measure delivered shaft power on a machine that you cannot stop, cannot disconnect, and cannot load with an absorption brake. Ships, wind turbines, helicopter drivetrains, and large industrial gearboxes all fit this description. The instrument also gives you fuel-burn-versus-shaft-power efficiency curves that absorption dynos simply cannot produce, because it reads the actual delivered power under real operating conditions — propeller cavitation, blade pitch, gearbox losses and all.
- Commercial marine: Kyma Shaft Power Meters installed on Maersk container ships measuring delivered propeller shaft horsepower for IMO EEXI and CII fuel-efficiency reporting.
- Wind energy: Manner Sensortelemetrie torque telemetry rings fitted to the main shaft of Vestas V90 turbines for gearbox load validation during prototype testing.
- Helicopter drivetrains: Sikorsky S-92 main rotor shaft torque measurement using a strain-gauged torsion section feeding the cockpit Engine Torque indicator.
- Power generation: Acceptance testing of GE steam turbine LP rotors at customer sites using HBM T40B torque flanges installed in the gearbox-to-generator shaft line.
- Heavy haul rail: EMD SD70ACe locomotive traction motor shaft torque verification during dynamic brake commissioning, where a portable torsion meter replaces a load bank.
- Mining: Continuous shaft-power logging on FLSmidth SAG mill drives at copper concentrators, used to trim ball charge against motor draw.
The Formula Behind the Torsion Dynamometer
Shaft power from a torsion dynamometer comes from two measurements multiplied together: torque from the twist angle, and angular velocity from the speed pickup. The twist-to-torque conversion depends on shaft geometry and material, so the formula matters at every operating point. At the low end of a typical operating range — say 25% of rated torque — the twist angle gets small enough that sensor noise dominates and your power reading wobbles by several percent. At rated torque you sit in the sweet spot where signal is clean and the shaft is comfortably below yield. Push to 120% of rated and you start crowding the elastic limit on a steel shaft, where modulus of rigidity is still linear but fatigue life becomes the limit, not measurement accuracy.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| P | Shaft power transmitted | W | hp |
| T | Torque on the shaft | N·m | lbf·ft |
| ω | Angular velocity of shaft | rad/s | rad/s |
| G | Shear modulus (modulus of rigidity) of shaft material | Pa | psi |
| J | Polar moment of inertia of shaft cross-section | m4 | in4 |
| θ | Measured twist angle over gauge length | rad | rad |
| L | Gauge length between angular references | m | in |
| D | Shaft diameter (solid circular) | m | in |
Worked Example: Torsion Dynamometer in a tugboat sea-trial shaft-power test
A naval architect at a Vancouver shipyard is verifying delivered shaft horsepower on a newly built 32 m harbour tug fitted with a Caterpillar 3516C main engine and a Reintjes WAF reduction gearbox. The propeller shaft is solid AISI 4140 steel, 180 mm diameter, with a Manner wireless torsion telemetry ring installed over a 2.0 m clean gauge length between the gearbox output flange and the stern-tube bearing. Rated continuous output at the propeller is 1,800 kW at 220 RPM. They need to confirm delivered power at the bollard pull condition.
Given
- D = 0.180 m
- L = 2.0 m
- G (AISI 4140 steel) = 80 × 109 Pa
- θ measured at bollard pull = 0.46 ° (0.00803 rad)
- N = 220 RPM
Solution
Step 1 — compute the polar moment of inertia of the solid 180 mm shaft:
Step 2 — convert the measured twist angle to torque using the shear modulus of 4140 steel:
Step 3 — convert 220 RPM to angular velocity and compute nominal shaft power:
That is the bollard-pull operating point — about 42% of the engine's rated 1,800 kW, which is normal for bollard pull where the propeller is fully loaded but RPM is held down.
Step 4 — at the low end of the typical operating range, harbour idle at roughly 80 RPM with light load, twist angle drops to about 0.08°:
At 0.08° of twist you are sitting at about 4× the noise floor of a good telemetry ring — the reading is usable but jitters by ±2 kW. Step 5 — at the high end, full-speed continuous at 220 RPM and rated torque, twist climbs to about 1.08°:
This is the sweet spot for accuracy — the twist signal is over 100× the sensor noise floor and the shaft is still at less than 50% of yield shear stress for 4140.
Result
Delivered shaft power at the bollard-pull point is 762 kW (1,022 shp). That tells the trial engineer the engine is producing about 42% of rated MCR at this propeller-load condition, which is exactly where a properly matched bollard-pull tug should sit — fully loaded prop, engine governor pulling RPM down. Across the typical operating range you see the dynamometer span 48 kW at harbour idle to 1,800 kW at continuous full power, and the reading gets cleaner as you climb because twist angle scales linearly with torque while sensor noise stays fixed. If your measured power comes out 5-10% below predicted, the most common causes are: (1) the telemetry ring's zero offset has drifted because the shaft warmed up 30 °C between calibration and the trial, shifting G by roughly 1%, (2) the gauge length has a hidden taper or weld repair inside the 2.0 m section that softens local stiffness, or (3) the toothed reference wheel has slipped 0.5 mm axially on its taper fit, throwing the phase reference off by a fixed angle that looks like a torque bias.
Choosing the Torsion Dynamometer: Pros and Cons
Torsion Dynamometers compete with absorption dynamometers (water brakes, eddy current brakes) and with inline torque flanges. Each makes sense in a different operating window — what matters is whether you can shut the machine down, whether you can break the driveline, and how much shaft power you need to handle.
| Property | Torsion Dynamometer (shaft-mounted) | Absorption Dynamometer (water/eddy brake) | Inline Torque Flange (HBM T40B class) |
|---|---|---|---|
| Typical accuracy | ±0.5 to ±1.0% of reading | ±0.1 to ±0.25% of reading | ±0.05 to ±0.1% of reading |
| Power range | 10 kW to 100+ MW | 1 kW to ~50 MW | 1 kW to ~5 MW |
| Speed range | 1 to 30,000 RPM | 0 to 12,000 RPM | 0 to 25,000 RPM |
| Driveline interruption required | No — clamps onto existing shaft | Yes — replaces the load | Yes — must cut shaft and bolt in |
| Suitable for in-service measurement | Yes | No — test cell only | Yes, but installation is invasive |
| Installed cost (mid-range capacity) | $15,000 to $60,000 | $80,000 to $400,000 | $25,000 to $100,000 |
| Calibration interval | 12-24 months | 12 months | 24-60 months |
| Best application fit | Marine shafts, wind turbines, helicopter rotors | Engine test cells, transmission rigs | Gearbox acceptance, R&D test stands |
Frequently Asked Questions About Torsion Dynamometer
You are watching the shear modulus of steel drop with temperature. G falls roughly 3% per 100 °C rise, and the formula T = (G × J × θ) / L converts that directly into a 3% torque under-read for the same true twist angle. On a marine shaft running through a hot stern tube the shaft can climb 40-60 °C during a multi-hour trial, which is enough to put you outside class-society accuracy limits.
The fix is either a shaft-mounted RTD feeding a temperature compensation channel into the telemetry, or doing your zero and span calibration at the same shaft temperature you intend to measure at. Most modern Manner and Binsfeld systems have built-in temperature comp — check that it is enabled, because some installers leave it off by default.
Drive it backwards from twist-angle requirement. You want at least 0.3° of twist at rated torque so the noise floor of a phase-shift sensor (around 0.003°) gives you 1% resolution. Plug rated torque into θ = T × L / (G × J), solve for L, and round up.
For a typical 2 MW turbine main shaft at 18 RPM and 1,060 kN·m rated torque on a 250 mm solid section, you need around 1.4 m of clean gauge length to get 0.3° of twist. If your nacelle layout only gives you 0.6 m of clean shaft, you will need a strain-gauge type instead — strain gauges read the surface shear directly and do not depend on gauge length at all.
Pick the torsion dynamometer when you cannot break the driveline. Cutting a shaft on a 12,000 TEU container ship to install an HBM T40B is a six-figure dry-dock job; clamping a Kyma or Binsfeld telemetry ring onto an existing shaft is a one-day install with the ship still afloat.
Pick the torque flange when you need 0.05% accuracy for an R&D test stand and you are designing the driveline from scratch anyway. The flange is more accurate, more repeatable, and easier to calibrate, but only because you are building the system around it from day one.
Almost always the instrument. A 1× rev periodic signal is the classic signature of toothed-wheel eccentricity — one of the two reference wheels is mounted with 0.05-0.1 mm of TIR, so as the shaft turns the air gap to the pickup modulates and the apparent phase between the two wheels swings sinusoidally.
Diagnostic check: stop the shaft, rotate it by hand through 360°, and watch the static phase reading. If it varies by more than 0.01° you have eccentricity. Re-clock or re-mount the wheel on its taper, then re-check. Real engine torque ripple from a diesel shows up at firing frequency (half engine RPM × cylinder count for a 4-stroke), not at 1× shaft RPM.
It matters more than people expect. Shear modulus G varies from about 77 GPa for low-carbon mild steel up to 82 GPa for alloy steels like 4140 or 4340 — that is a ±3% spread before you even touch the instrument.
Class-society shaft-power meter installations require a material certificate for the gauge-length section so G is known to within 1%. If you only have 'medium-carbon steel' on the drawing, either pull a coupon and measure G in a torsion test rig, or treat your power readings as having a built-in ±3% bias. For trade compliance work like IMO CII reporting that bias is too large to ignore.
Hollow shafts work fine — you just swap J. For a hollow circular shaft, J = π × (Do4 − Di4) / 32. Everything else in the formula stays the same.
The catch is that you need an accurate measurement of the inner bore diameter, and on a finished propeller shaft that bore may not be perfectly concentric or perfectly cylindrical down its full length. A 2 mm bore eccentricity on a 200/100 mm hollow shaft shifts J by roughly 1.5%. If you cannot verify the bore geometry, either use a strain-gauge instrument that reads surface strain directly (independent of J), or accept the geometric uncertainty in your error budget.
That is the shaft and the telemetry ring reaching thermal equilibrium. Two things are happening at once: the shaft itself is warming (changing G), and the telemetry electronics are warming (changing internal offsets). Both walk in the same direction for about the first 15-25 minutes, then stop.
Standard practice on sea trials and turbine commissioning is a 30-minute warm-up at steady load before you start logging accuracy data. If you need to log during the warm-up, log the shaft temperature alongside torque and post-correct G in software.
References & Further Reading
- Wikipedia contributors. Dynamometer. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
