Is Your Dyno Lying?

Measuring your horsepower depends on whose yardstick you use.

Horsepower is like good luck. It can't be seen, touched, or tasted, but you know when the inventory is low. How much do you have? Go measure it. The best place to put your luck to the yardstick is in Las Vegas. The best place to measure your horsepower is on a dynamometer. In both cases, results have been known to vary.

Turns out that different horsepower dynamometers will spit out different results on the same car depending on their design and how carefully the tester manages the variables. When ambient air temperature and pressure were all that yesterday's carburetors cared about, factoring in the variables was easy. But modern cars are gradually becoming too smart for what was once a simple test. Engines trying to squeeze out more horsepower are run by watchful computers with better sensors that are able to differentiate between speeding down a freeway with a cooling headwind and spinning the dyno drums in a stuffy room. The technology of today's cars needs to be factored into your readings along with variations from dyno type to dyno type, or your horsepower numbers may be hazy, too.

Recall that horsepower is a unit measuring the rate at which an engine performs work. By current industry standards, one horsepower equals 550 pounds lifted at the rate of one foot every second. Automakers figure a car's oomph by hitching the engine to a machine that puts a controlled load generated by an electric or hydraulic brake on the flywheel. They plumb the intake and coolant to external fixtures that simulate on-road airflow and then run the engine from idle to redline. The engine dyno measures torque at the crankshaft using a variety of means and then uses a math formula—torque times rpm divided by 5252 (a sort of gearhead's constant, derived by dividing . . . oh, never mind). Horsepower comes after the equal sign. Thousands of hours of testing produce a peak horsepower number that prints in the dealer's brochure.

For the rest of us, rather than unbolt the engine and rent a multimillion-dollar dyno room, it's much easier to simply drive the whole car onto a chassis dyno that measures horsepower where the tires meet the tarmac. Throughout the U.S. are shops with such machines, some even capable of handling four-wheel-drive cars. Ordinary citizens pay anywhere from $35 to $100 to run the car up to redline three or four times, the tires turning rollers monitored by the dyno's computer. Eventually, the computer spits out a horsepower graph that you can take to your next bar-stool drag race.

But before you pop the clutch on your tongue, consider that a chassis dyno doesn't measure horsepower at the flywheel but at the tires after various driveline losses have subtracted their drag. Friction from rubbing gear faces, inertia from heavy shafts, and the stirring of oatmeal-like gear lube all reduce the advertised horsepower reaching the tires and, hence, the dyno.

"There's no really accurate way to get engine horsepower from a chassis dyno," says Matt Harwood, marketing coordinator for Mustang Dynamometer, a major supplier of chassis dynos in Twinsburg, Ohio. Some tuners use the so-called 15/20 rule, which assumes a 15-percent driveline loss for manual transmissions and 20 percent for automatics. But, says Harwood, "I've seen losses as high as 35 percent." So unless it's printed in the brochure or was measured on a true engine dyno or by a tuner with tons of experience with your particular brand of car, any flywheel horsepower number quoted by a hot rodder under the shade tree is most likely just a calculated guess.

More important, did your car run on the dyno as it would on the street? If it's the latest model, chances are good it may not have, says BMW tuning wizard Steve Dinan. An afternoon spent at his Bavarian speed emporium in Morgan Hill, California, convinced us that cars are gradually becoming too computerized for the simple dyno test.

Dinan's cars are wired with a battery of sensors that report when the airflow over the bumper is too little, when the inlet air is too hot, and when the water temperatures in the block and radiator are too close together (most turbocharged and supercharged cars also "know" when their intercoolers aren't cooling enough). The computer reacts by backing off the spark and turning up the richness—and as a result, turning down the power—to prevent catastrophic engine meltdowns.

To prove his point, Dinan bolts to his Dynopack one of his 2003 Dinan M5s, heavily tweaked to make a claimed 470 horsepower at the crank (he expects about 415 at the wheels). With the hood closed and no external fan blowing air into the radiator, the car wheezes out just 334 horsepower at the wheels. An LCD data logger on the dashboard reveals the air-fuel ratio from the engine computer. Approaching redline, the BMW's computer richens the mixture all the way to 9.5:1 as the underhood temperatures soar.

That's one thick mix, practically charcoal briquettes blowing out of the tailpipe. But then, with the M5 running in fifth gear (the 1:1 gear ratio with the least friction, preferred by dyno testers), the computer expects 159 mph worth of cooling wind blast around the horsepower peak. It's getting nothing, and it knows.