What drivetrain loss is, and what it is not
Drivetrain loss is the power that never reaches the road. It has three physically distinct parts, and confusing them is the source of most arguments about the subject.
Friction and churning. Gears mesh, bearings turn, and oil is thrown around inside the gearbox and differential. This part is largely a fixed drag: it depends on speed, temperature and oil viscosity far more than on how much torque is passing through.
Load-dependent losses. Gear teeth deflect and slide under load, so a small fraction of the torque genuinely does turn into heat in proportion to what is being transmitted. Torque converters add slip, which can be substantial and varies with the converter and the operating point.
Inertia. On an inertia dyno the wheels, tyres, driveshaft and clutch have to be accelerated as well as the roller. That absorbs power for as long as the engine is accelerating, and none of it is a steady-state loss at all.
What matters for using this calculator is that only the middle group scales cleanly with power. A single percentage is a convention, not a measurement, and it is why two reputable shops can disagree by 40 hp about the same car.
The conversion, and the denominator people get wrong
Loss is conventionally quoted as a share of crank power:
L = (crank hp − wheel hp) ÷ crank hp
Rearranged, that gives the two conversions this calculator performs:
crank hp = wheel hp ÷ (1 − L) wheel hp = crank hp × (1 − L)
The denominator is the trap. Dividing by (1 − L) is not the same as multiplying by (1 + L), and people do the second because it is easier mentally. At 15% the correct factor is 1 ÷ 0.85 = 1.1765, while the shortcut gives 1.15 — so 400 whp becomes 470.6 hp correctly, or 460 hp if you take the shortcut. The 10.6 hp gap is entirely arithmetic, and it grows with the loss percentage: at 22% the two factors are 1.2821 and 1.22.
The other thing to understand is what the percentage assumption does to your answer. Take 400 whp. At a 10% assumption the crank figure is 444 hp; at 22% it is 513 hp. That 68 hp spread comes from nothing but the assumption, which is more than most bolt-on modifications produce. If someone reports a crank figure from a chassis dyno, the number tells you as much about their loss assumption as about their engine.
Worked example: 400 wheel horsepower through an automatic
A rear-wheel-drive car with a torque-converter automatic reads 400 hp at the wheels. Take the conventional 15% for that layout.
- Loss fraction. 15 ÷ 100 = 0.15.
- Crank horsepower. 400 ÷ (1 − 0.15) = 400 ÷ 0.85 = 470.6 hp.
- Power absorbed. 470.6 − 400 = 70.6 hp.
- Check the definition. 70.6 ÷ 470.6 = 0.15, so the loss really is 15% of crank power, as intended.
- In kilowatts. 470.6 ÷ 1.341022 = 350.9 kW.
Now run it the other way. The manufacturer rates the same engine at 500 hp at the flywheel. What should it put down?
- 500 × (1 − 0.15) = 425 hp at the wheels.
So a car reading 400 whp on a 500 hp rating is 25 whp below the expectation the 15% assumption sets — which could be a tired engine, a different dyno, a hotter day, or simply that 15% is the wrong figure for this car. That is the honest reading, and it is why the number is a conversation starter rather than a verdict.
How much loss should you actually assume?
The bands in the layout selector are the ones tuners use, and they are rules of thumb rather than measured constants. Manual rear-wheel drive sits lowest because the path is short and there is no fluid coupling. Front-wheel drive adds a transaxle. Torque-converter automatics add slip, which is why they are quoted higher. All-wheel drive adds a transfer case, a second differential and a second pair of driveshafts, and heavy part-time systems sit highest of all.
Here is the more useful way to think about the number. Suppose the true loss on a given car is roughly a fixed Lhp of parasitic drag — the friction and churning group above — rather than a fixed percentage. Then the true percentage is Lhp ÷ crank, which falls as crank power rises. Concretely: if a drivetrain absorbs a steady 50 hp, that is 25% of a 200 hp engine and 6.25% of an 800 hp engine. Applying one fixed 15% instead would credit the 200 hp engine with only 30 hp of loss and the 800 hp engine with 120 hp — understating the low-power case and overstating the high-power one. To the extent the fixed-drag model holds, a single percentage is least wrong in the middle of the range it was derived from.
This is why big-power builds that quote 25% AWD losses are usually inflating their crank figures. It is also why the only defensible way to know a car's real number is to measure the same engine on an engine dyno and then in the car, which almost nobody does.
Use crank estimates for comparison against factory ratings and nothing more. For tuning decisions, compare wheel horsepower against wheel horsepower on the same dyno, same day, same operator. If you want a power figure that is independent of dyno house style altogether, work backwards from a timed run with the trap speed horsepower calculator or check the result against power to weight ratio.
Crank horsepower from wheel horsepower at common loss assumptions
| Wheel hp | 10% loss | 12% loss | 15% loss | 18% loss | 22% loss |
|---|---|---|---|---|---|
| 200 | 222.2 | 227.3 | 235.3 | 243.9 | 256.4 |
| 300 | 333.3 | 340.9 | 352.9 | 365.9 | 384.6 |
| 400 | 444.4 | 454.5 | 470.6 | 487.8 | 512.8 |
| 500 | 555.6 | 568.2 | 588.2 | 609.8 | 641.0 |
| 600 | 666.7 | 681.8 | 705.9 | 731.7 | 769.2 |
At 400 whp the assumption alone spans 444.4 to 512.8 crank hp — a 68 hp range from a single choice of percentage.
Correction factors are a separate question
Drivetrain loss and atmospheric correction are two different adjustments and they are often confused. SAE J1349 correction adjusts an observed reading to a standard atmosphere — 99 kPa of dry air at 25 °C — so that runs on different days are comparable. Drivetrain loss converts between measurement locations on the same car. A quoted figure can be corrected, uncorrected, at the wheels or at the crank, and comparing two numbers means knowing which of the four each one is. Work out how much the weather alone is moving your readings with the density altitude calculator.
Why two dynos disagree about the same car
- Different dyno types measure different things. An inertia dyno infers power from how fast a known mass accelerates; a load-bearing dyno holds a steady speed and measures torque directly. They respond differently to tyre slip, ramp rate and inertia in the rotating parts.
- Correction factor choice. SAE J1349, DIN, JIS and uncorrected readings differ by several percent on the same run. The correction is usually printed on the sheet.
- Tyre pressure, strap tension and tyre construction. Sidewall flex on a roller absorbs real power. Higher pressures and a stiffer tyre commonly read higher.
- Gear selection and ramp rate. A slower sweep gives heat time to build in the intake and the engine time to fall out of its best state; a faster sweep flatters an inertia measurement.
- The loss assumption itself. If either shop reports a crank figure, the percentage they used is part of the result and should be stated alongside it.
When the crank figure matters and when it does not
Convert to crank power for exactly two purposes: comparing against a factory rating, and comparing against an engine-dyno figure from a builder. Both of those are quoted at the flywheel, so a wheel figure cannot be compared with them directly.
For everything else, stay at the wheels. Wheel horsepower is what the car actually delivers, it is what changes when you tune, and it is measured rather than inferred. Tracking a build on wheel horsepower removes one estimated quantity from every comparison you make.
Two related figures are worth having alongside it. Power-to-weight ratio explains far more about how a car performs than either horsepower number on its own, and torque tells you what the engine does away from the power peak — convert between the two at any rpm with the horsepower from torque calculator. If the intention behind the crank number is to predict performance rather than to settle an argument, those are the tools to reach for.