I’ve always been told that its between 10- 12%
I talked to my tech and he said 15% for drivetrain and for apples to apples dyno numbers. The low-inertia dyno (brake dyno) is different than dynojet and consistently gives lower (more accurate) numbers at all throttle positions.
Here is some more info on this topic that I found helpful:
I get that question all the time and this is the best explanation I have for it:
Eddy-Current Chassis Dynamometer Information
To get the most horsepower and performance out of your engine, AF1 Racing utilizes the latest in dynamometer technology. We exclusively use the Factory Pro EC997a eddy-current low-inertia dynamometer. An eddy-current style of dynamometer is far superior to an inertia-style dynamometer (most DynoJet dynes) in terms of its accuracy and capabilities. An inertia-style dyno is a glorified accelerometer, while an eddy-current dynamometer provides real-world loading of the engine. Only a true eddy-current style of dynamometer is capable of step-tests, where the eddy-current brake holds the engine at specific RPM data points and measures HP and torque in real-time steady-state. Testing steady-state, where drum acceleration equals zero, is the only way to get accurate EGA data, True horsepower readings, and real-world loading. Having the capability to tune steady-state is the only way to accurately tune EFI-equipped motorcycles and flatslide carburetors.
HP = (Torque x RPM) / 5252 and Torque = (HP x 5252) / RPM
An inertia style dynamometer measures HP and torque by calculating the acceleration of a heavy drum during a rapid sweep test. You cannot hold steady-state on an inertia dynamometer (accelerometer) making EFI and flatslide tuning impossible at best. How does one test highway cruise and part throttle when you have to accelerate the drum just to get a reading? An eddy-current dynamometer measures rear wheel torque with a load cell while the engine is loaded steady-state at any specific RPM we choose. Think of it like a giant torque wrench where the brake provides resistance against the force of the tire. After measuring torque with the load cell on an eddy-current dynamometer, calculating HP is easy. Horsepower is a simple function of torque and RPM.
Our dynamometer also features a super light roller that is quick to react to load changes and power outputs, which enables us to very precisely tune EFI-equipped bikes, and flatslide carburetors. Our Factory Pro eddy-current dynamometer is not affected by gearing changes, wheel inertia changes, tires sizes, or tire pressures. It reads True HP corrected to SAE atmospheric conditions. It is 0.2 HP accurate every single day, every single run ever done at any Factory Pro tuning center. We can accurately compare any two sessions any Factory Pro dynamometer has ever done. No other machine comes close to its repeatability and accuracy. Expect your HP numbers to be around 12-17% lower with our dynamometer when compared to inertia-type dynes commonly used in magazines and at other shops. Our machine can run anything from scooters, cruisers, HDs, dirtbikes with knobbies, two strokes, and AMA Superbikes.
I have an eddy-current style dyno with extremely accurate load sensors and climate sensors that calculate everything automatically so there isn't a chance for "human error" corrections that would show an inaccurate/higher output reading.
