Take internet horsepower with a grain of salt
Often very high (sometimes unbelievable) horsepower numbers are published.
There are a few methods (wether applied intentionally or accidentally) to publish wrong results.
- miscalibrated dyno
- a dyno is (or at least should be) calibrated. It can print any result if calibration is wrong. Installers often favor dyno-s showing high HP instead of real HP, which keeps this trend live
- measuring WHP, and apply a presumed multiplier for crank HP. Often the multiplied result is published as WHP (which it is not, by far). The factor is usually * 1.2 .. * 1.25 depending on front WD/rear WD/4WD and rule of thumb (so they say 500 HP when they measure 400 HP on the wheel).
- the multiplier often assumes tire slip that is just wrong (eg. would melt the rubber) or not present (eg. using a dyno where you bolt a dyno to each hub: makes the conversion factor MUCH smaller).
When you compare engines, before deciding to believe or not, it's a very good idea to
- calculate the VE (eg. in a spreadsheet). Either real VE numbers, or just scale the crank-HP numbers to 1.0L 100kPa and 6000 RPM
- best to calc for (/scale) the whole torque curve if possible (often not easy, as the boosted MAP is not published for each RPM)
- Compare with factory numbers (like to a Honda VTEC NA engine) that are usually right
- they measure the crank HP (so without driving losses and fictive multiplier), on the best out of 10 engines in circumstances like -20C MAT that you will never be able to reproduce at home on the dyno, but at least those are not fictive numbers.
If the VE numbers for an engine at MAP=250 kPa and 9000 RPM (especially if it's 2 valve head :-) calculates to be higher than a factory variable valve Honda/Toyota at MAP=100kPa and 8000 RPM, start to doubt.
Consider 400m time
The measurements done on 2 different dyno's at different times deviate more than desirable for sane comparison. Even if there is no intent to publish bad data (from bad calibration or multipliers).
The best times usually tell much more.
TODO: find or build a program to closely estimate best times, using
- torque curve
- gearbox (gear ratios, differential ratio, tire size, etc...)
- time needed to shift gears is a parameter (appr 200..800msec: lower for a sequential gearbox, higher for a manual)
- from the above, the best points for switching gears can be calculated automatically
- air drag CW
- one would think this is hard, but not at all. The air-drag is easy to measure at 200km/h airspeed (in natural gear), knowing the weight. CW can be estimated quite well (knowing professional measurements of similar chassis), and cross-section can be measured with a digital camera and gimp.org
- don't forget altitude above see level
- tire traction
- the worst of all
Knowing the time and speed after 60 or 100m makes it all much more precise by largely eliminating the tire traction part.
Having RPM log of the whole run also helps a little (to find not only the area under - a certain RPM-range of - the torque-curve, but the shape of it too).