Everybody knows about engine dyno testing. But, multiple factors play into the published numbers which may have escaped your attention. Torque and horsepower are the numbers that everybody knows, but there are also useful tuning values that rarely receive the attention they deserve. This story takes a look at those details which can give the sharp tuner a significant advantage over his less-informed colleagues.
The best place to start is with how a dynamometer produces its numbers. There are several different styles of dyno, but the two most common are water-brake and eddy-current electric dynos. A third type — AC electric dynos — are highly accurate and can actually drive an engine to measure rotating power, but access to those dynos is limited. So we’ll focus on the far more common water-brake versions, like the SuperFlow SF-902 series used by many shops.
A typical water-brake absorber is housed in a large circular steel case using water pumped under pressure into the housing. Inside is a set of curved vanes on a rotor that is connected to the engine. This rotor spins inside a set of fixed stator vanes. As the engine rotates, the water accelerates away from the rotor and the pressure is forced against the stator vanes where the twisting motion is measured.
Measuring Power From Water
The load from the water creates a force that is measured by an electronic load cell called a strain gauge. Of course, the gauge must be calibrated to ensure the readings are correct. This usually involves hanging a calibrated weight from the end of a precise-length arm.
If the arm is two feet long and 150 pounds (including the weight of the arm) is hung on the end of the arm, this creates 150 x 2 = 300 lb-ft of force on the gauge. That figure is then entered into the electronics package in the dyno. This calibrates the strain gauge. If a dyno facility does not calibrate its equipment on a regular basis, this can lead to erroneous data.
With the dyno calibrated, we can then attach an engine and accurately measure the torque exerted by the engine. Let’s say our test engine makes 400 lb-ft of torque at 4,000 rpm. The formula for computing horsepower is HP = (Torque x RPM) / 5,252. We won’t go into how this 5,252 number originated in this article, but it can be traced back to James Watt who devised a way to equate the power of his steam engines to the power of a draft horse.
With 400 lb-ft of torque at 4,000 rpm we have:
This formula also reveals that if we make the same amount of torque at 6,000 rpm (1.5 times more RPM), we’ll also make 1.5 times the horsepower because we’re exerting the same force (torque) more often over the same period of time:
These are observed power numbers, which means they have come directly off the dyno, based on existing atmospheric conditions. To compare power readings from different days with different weather conditions, engineers created a correction factor. This factor takes into account the three most critical atmospheric variables: pressure, temperature, and vapor pressure — also known as humidity.
Comparing Numbers
The correction factor most magazine stories and performance engine builders use is what is called Standard Temperature and Pressure (STP). The conditions for this correction include sea-level air pressure of 29.92 inches of mercury (inHg), an ambient temperature of 68 degrees Fahrenheit, and zero-percent humidity.
This set of conditions will make an engine look strong, but as long as all engines are corrected to this same standard they can be accurately compared. An engine running on a 100-degree day with a pressure of only 29.12 inHg, and 70-percent humidity will produce far-lower observed power than one measured at STP.
Most dyno sheets will indicate the percentage of correction factor applied to the observed numbers. In our example dyno test, the STP correction factor came to 1.066 which will add 6.6 percent to the observed numbers. However, corrected horsepower is not quite as simple as multiplying the observed numbers by the correction factor.
The Society of Automotive Engineers (SAE) has established a friction factor based on the size of the engine, with special attention paid to stroke, since piston travel represents a large percentage of engine friction. Since friction is not affected by atmospheric changes, the accepted procedure is to add the frictional power loss to the observed power, multiply by the correction factor and then remove the friction power from the result. SuperFlow performs this calculation automatically using this formula:
This might look like a way to produce inflated horsepower numbers, but it really is not; all engine dynos correct horsepower in this fashion. We had to go through this explanation in order to show precisely how the corrected numbers are created. Otherwise, somebody might think the dyno operator was cheating, because if you merely multiply the observed torque times the correction factor, the result is a lower number.
The important part is that correction factors allow us to compare different combination tests on various days with a reasonable assurance of comparable numbers, despite atmospheric changes. It’s worth noting that the Society of Automotive Engineers (SAE) changed the production engine correction factor in 1993 to reflect normal operating conditions more closely. The new standard is a pressure of 29.234 inHg, 77 degrees Fahrenheit, and zero-percent humidity. This reduces the corrected power numbers by roughly four percent compared to the STP version.
More To A Dyno Than Horsepower and Torque
To investigate some of the other parameters, let’s start with fuel flow, which is recorded in pounds per hour (lbs/hr). If the engine is carbureted, a good dyno report will separate the primary and secondary sides so the tuner can evaluate each individually.
Once we have the lbs/hr of fuel consumed, this can be divided by observed horsepower to create what is called brake specific fuel consumption (BSFC) expressed in pounds of fuel per horsepower, per hour (lbs/hp/hr). This number is the amount of fuel used to make one horsepower for one hour at wide-open throttle (WOT).
As power has improved over the years, so has engine efficiency. This reveals itself as less fuel required to make the same power. Roughly 40 years ago, the BSFC standard for a naturally aspirated pump-gasoline engine was 0.50.
This means a typical engine would burn a half-pound of fuel to make one horsepower for one hour. BSFC numbers have dropped continually since then and now. According to Westech’s dyno guru Steve Brule, he now sees BSFC numbers for street engines down in the 0.41 to 0.42 range.
BSFC is a measure of efficiency but is often misconstrued as a measure of whether the engine is “rich” or “lean.” An engine with a BSFC of 0.52 does not necessarily run richer than one with a BSFC of 0.42; it just means that the latter engine requires less fuel to make the same power.
A derivative of this fuel-efficiency rating is a related term called brake specific air consumption (BSAC). This is where airflow (in cfm) is divided by observed horsepower. For naturally aspirated engines, it’s common to use around six pounds of air to make one horsepower for one hour. At 500 horsepower, this means the engine is consuming approximately 3,000 pounds of air per hour.
We won’t spend much more time on BSAC because it’s not typical to see the air turbine used on most dyno tests. According to data in Harold Bettes’ HPBooks’ Engine Airflow, you can expect a properly built naturally aspirated engine to use around 1.25 cfm of air per horsepower at peak torque, with that number moving up to 1.4 cfm per horsepower at peak horsepower. So a 500-horsepower engine at peak power would be moving roughly 700 cfm of air.
Pressures Matter, Even For N/A Combinations
Brake Mean Effective Pressure (BMEP) is another calculation you will see on a typical SuperFlow report. This is strictly a theoretical calculation based on a factor derived for a four-stroke engine. It contends an engine will make one lb-ft of torque per cubic inch (lb-ft/ci). The factor used equates to 150.8 psi of average (mean) cylinder pressure per cubic inch pushing down on the piston throughout the entire stroke. This factor is used as a way to evaluate how efficiently the engine makes power.
BMEP uses observed torque multiplied by 150.8 and then divided by the engine displacement. In a 6.0L (346ci) LS engine example:
BMEP is used as a way to compare engines and does not represent the actual cylinder pressure. Naturally aspirated street engines on pump gas that come in at 180 psi or more generally make good power. BMEP numbers exceeding 200 psi are generally only found in top-tier naturally aspirated racing like NHRA Competition Eliminator, Pro Stock, and NASCAR, for example.
Another, perhaps simpler, way to look at average cylinder pressure is to divide corrected peak torque by the displacement for lb-ft/ci. For pump-gas engines, a number at or above 1.25 lb-ft per cubic inch is considered good. BMEP uses observed torque, but the lb-ft/ci formula uses corrected torque. At that same 4,800 rpm, the corrected torque is 477.3/364 = 1.31 lb/ft per cubic inch which is good.
Engine builder Jon Kaase tells us that competitive Engine Masters engines will commonly make around 1.55 lb-ft per cubic inch. To illustrate this point, we did the math on Kaase’s Mercury-Edsel-Lincoln Vintage-class-winning engine from a few years ago that made 630 lb-ft of torque from 400ci, which comes to 1.57 lb-ft per cubic inch! These generally run on gasoline that’s closer to 100 octane.
Rate of Acceleration Plays a Part
Another factor that rarely gets attention is the dyno acceleration rate. In the pre-computer dyno testing days, each RPM-point was tested separately. This was usually done at 500 rpm intervals. But, with computer servo control of the water brake, a SuperFlow, for example, will run through an entire 2,500 to 6,500 rpm test range in 100-rpm increments in one continuous sweep at a pre-programmed acceleration rate.
The rate is usually set at 300 rpm per second. If the engine is tested at a faster rate, say 600 rpm/sec., power numbers will be lower because it requires horsepower to accelerate the engine more quickly. Conversely, a slower test rate will produce slightly better numbers.
Despite all these standards, there are many ways to affect the output of a given engine dyno test. A simple tweak in testing procedures can have a significant effect on displayed power output. For example, coolant temperature affects power.
We’ve performed back-to-back tests and witnessed a naturally aspirated, 400-horsepower street engine gain ten horsepower just by reducing the starting coolant temperature from 180 degrees to 135 degrees. That’s free horsepower.
Another trick that can go unnoticed is the configuration of the headers used in testing. For example, Westech Performance prefers to use its own dyno room headers, not only because they allow easy access to spark plugs, but also because the primary tubes extend almost horizontally from the engine. This straight shot from the exhaust port tends to improve peak horsepower with no cost to low-end power. It’s a small thing but it all adds up.
Most dyno cells also use an electric water pump to circulate water with the engine off to quickly stabilize the engine coolant temperature. But this pump also eliminates the parasitic loss of driving a water pump with the crankshaft. On a 6.0L LS engine, we’ve seen six- to eight-horsepower improvements after replacing an entire LS truck accessory drive (water pump, alternator, and power steering pump) with an electric water pump.
Street engines require an air filter, yet very few dyno sessions are performed with an air cleaner in place. However, we discovered (by accident) an eight-horsepower loss on a 550-horsepower EFI small-block when a one-inch spacer was used underneath an air cleaner base to provide clearance for throttle linkage on a four-barrel throttle body.
The spacer was necessary for the throttle linkage to clear the filter base. By adding a gentle radius to the base, we eliminated the spacer and the power came right back. Had we not tested the engine exactly the way it was going to be run in the car, we would not have discovered that power loss.
Enthusiasts are often confused when their real-wheel power numbers are much lower than the flywheel numbers — even when subtracting out the equivalent power loss through the drivetrain. This can often be explained by simply adding up examples of the above situations because the engine was not engine-dyno tested in its complete and accurate street configuration.
A six-bladed, engine-driven cooling fan can easily devour eight flywheel horsepower. We’ve done that test too. Add up the above examples and it would not be surprising to account for a 40- to as much as a 50-horsepower loss between the dyno and the engine compartment.
If nothing else, this story should open up ideas about how paying careful attention to all the engine details can affect final results. The devil, they say, is in the details.