Horsepower: Making it Add Up
The O Pine




We recommend Internet Explorer set to 1024x768.

© 2003 Brian F. Schreurs
Even we have a disclaimer.

If you're gonna bench race, at least start your brain.
Considering how much energy car people expend talking about horsepower and comparing how much they and their rivals have, there is a surprising amount of misunderstanding about how and where those horsepower numbers come from. While it's true that horsepower is horsepower, the circumstances under which it has been measured over the years has varied. To make a valid comparison, it's important to make sure you're using the same numbers for all cars.

From the Horse's...

Manufacturers cause most of the problems in that there are several standards by which they rate their cars when new. Without knowing exactly how much difference there is from one standard to another, it's easy to get into the mentality that "horsepower is horsepower" (mathematically, it is always the same, after all) without taking into consideration the circumstances under which it was measured.

SAE Net Horspower
In 1972, American manufacturers phased in SAE net horsepower. This is the standard on which current American ratings are based. This rating is measured at the flywheel, on an engine dyno, but the engine is tested with all accessories installed, including a full exhaust system, all pumps, the alternator, the starter, and emissions controls. Both SAE net and SAE gross horsepower test procedures are documented in Society of Automotive Engineers standard J1349. Because SAE net is so common, this is the standard we will use to compare all others.

SAE Gross Horsepower
This is the old process that American manufacturers used as a guide for rating their cars. It was in place until 1971. SAE gross also measures horsepower at the flywheel, but with no accessories to bog it down. This is the bare engine with nothing but the absolute essentials attached to it; little more than a carb, fuel pump, oil pump, and water pump. Because the test equipment on the engine is not the same as in SAE net, it is impossible to provide a mathematical calculation between SAE net and SAE gross. As a general rule, however, SAE net tends to be approximately 80% of the value of SAE gross. SAE J245 and J1995 define this measurement.

DIN Horsepower
This is a standard, DIN 70020, for measuring horsepower that very closely matches SAE net. The conditions of the test vary slightly, but the required equipment on the engine and the point of measurement (flywheel) remains the same. Because the test conditions are so similar, it is safe to divide DIN horsepower by 1.0139 to arrive at SAE net. This value is so close to equal that for all but the most technical purposes DIN and SAE net are interchangeable.

Brake Horsepower
Often road test magazines will list horsepower as "bhp". This is just another way to talk about SAE net horsepower.

Kilowatts
Kilowatts, or kW, is not a different way of measuring engine power; it's just a different unit of measure. Countries that use kilowatts instead of horsepower typically use a rating system very close to SAE net horsepower (usually DIN). To convert kW to SAE net hp, divide the kW value by 0.7457.

Advertised Horsepower
Surprise! Those horsepower numbers presented in advertising and brochures aren't always accurate. Though manufacturers are supposed to base their horsepower ratings on SAE net standards, they are not completely beholden to it. They often fudge the numbers. Ford and Mazda both recently got in trouble with the Mustang Cobra and the MX-5 Miata, respectively, when they delivered a car that had less horsepower than what they advertised. Ford ended up doing considerable warranty work to bring the numbers up where they belonged, and Mazda re-rated their car and offered to buy back any offended customers' cars. General Motors regularly underrates their engines, most notably the GM LS1 5.7L engine as installed in the F-body (Camaro and Firebird) cars. Mechanically almost identical to the engines installed in the Y-body car (Corvette), the engine mysteriously "lost" 40 advertised horsepower in the F-body chassis. Although this technically is as fraudulent as selling a car with less than the advertised horsepower, no one seems to complain when they get a car with more horsepower than what appears on the spec sheet.

Testing, Testing

Factory ratings are all well and good, but many enthusiasts modify their cars and then want to see how much of an improvement they got from their labors. The problem is that most of the time people are not interested in ripping the engine out of their car to have it tested on an engine dyno; no, they're going to be testing on a chassis dyno. The most common chassis dyno, the inertial dynamometer (popularized by DynoJet), measures the horsepower as delivered at the power wheels -- whether front or rear.

But testing rear-wheel horsepower (rwhp -- obviously, front drivers would be measuring fwhp) makes it difficult to convert from what the dyno says to what the manufacturer says. The manufacturer, remember, measures horsepower at the flywheel. All that equipment between the engine and the wheels -- the transmission, driveshaft, differential, and axles -- introduce friction and inertial losses summarized as "powertrain loss" or "parasitic losses". The efficiency of the driveline can greatly affect the amount of the powertrain loss: Ford's AOD transmission, for example, is notoriously inefficient. As a very general rule, rear-wheel horsepower on a manual-transmission car is about 15% less than SAE net, and rear-wheel horsepower on an automatic-transmission car is about 20% less than SAE net.

Even looking at dyno numbers, though, it's important to exercise some caution. Dynos measure horsepower under the conditions of the day, then apply a mathematical conversion to bring the numbers in line with SAE J1349. The raw numbers can vary substantially. In one dyno test of a 1998 Firebird conducted several days apart, the same car ran a raw number of 284 horsepower one day, and 299 horsepower on a rather colder day. Corrected, both numbers were within half a horsepower of each other. The corrected numbers are useful for comparing this car to other cars, or the same car after different modifications spanning a long time, but in the real world a car's horsepower isn't corrected: on a dragstrip, the Firebird would have been about a tenth quicker on the day it was making 299 horsepower than on the day it was only making 284.

Vapor Horsepower?

For people in the habit of thinking about SAE net horsepower, or old musclecar enthusiasts accustomed to SAE gross numbers, looking at real-world rear-wheel horsepower can be quite a wake-up call. This 1970 Charger makes an excellent example. Its 318 was factory rated in 1970 at 230 horsepower (SAE gross). But on the dyno it came just short of 150 horsepower (corrected rear-wheel). Where did that 80 horsepower go?

Since that Charger is an automatic, roughly 20% of it went to turning the drivetrain. That puts it at somewhere around 188 SAE net horsepower (or to use American manufacturers' penchant for rounding up, 190). But since the factory number uses SAE gross, there's another 20% difference. And that puts us at 235 horsepower, just about where it needs to be. It all adds up, and the same engine can have an 80 hp difference through no other fault than the means by which the power is measured.

Things get real interesting when the numbers don't add up. Dyno testing proved that General Motors was lying about the low horsepower numbers in the F-body when compared to the same engine in the Y-body. Hot Rod magazine gathered a collection of performance cars and dyno tested them for the May 1998 issue. They found 292 rwhp for a Firebird Trans Am and 286 rwhp for a Corvette. The slight difference between the cars is likely due to varying build tolerances; certainly not enough to say one engine's design is notably different from the other's. Either way, the LS1 is looking at about 340 SAE net horsepower in 1998, nearly on the money for the Corvette's factory rating (345) but way aboveboard for the Firebird's (305). By comparison, the 1998 SVT Mustang Cobra was also rated at 305 horsepower but on the dyno it only delivered 257 rwhp -- just right for a 15% powertrain loss. And the chart on this Camaro page seems to support the underrating of the F-body cars by looking at the performance numbers it posts compared to other vehicles with higher rated horsepower (and higher price tags). In this case, the vapor horsepower is the power loss from when the SAE net horsepower was converted into ad copy.

Dyno Racing

We've focused on the F-body quite a bit because it makes for such an interesting case study, but this is not to say that General Motors has a monopoly on fiddling with horsepower numbers. "Tweaking" the numbers a bit is a common practice, especially in segments of the market where power sells, and amongst enthusiasts who have modified their cars but never put them to the test. At the end of the day, arguing about exact horsepower numbers leaves you with nothing but a day wasted arguing. You can't race a dyno; what matters is how fast a car can get down the track. If it can run the number, then the amount of horsepower it took to get it there, or how the horsepower was measured, won't change the outcome of the race. Next time someone tries to bench race with horsepower numbers, ask for a timeslip. If he can't deliver, then cut the conversation short and move on. He's not really in the game anyway.