Chassis Dyno Guide - Frequently Asked Questions About Dyno Tests - Hot Rod Magazine (2024)

Over the past decade, chassis dynos have proliferated across the country, in theory making it practical for the masses to evaluate the performance of their engine as installed in the vehicle and compare the results against their buddies' cars. But this assumes that results obtained at a given facility can be directly compared to those from other facilities, and that the results accurately reflect the true power output of the test vehicle. There's also the question of degree of accuracy, or whether the average dyno shop can adequately evaluate the true effect of small, incremental changes, particularly on late-model, computer-controlled cars.

To dig deeper into these issues, HOT ROD contacted leading manufacturers, including Dynapack, Dynojet, Mustang, and SuperFlow; consulted Jeff Burt of TMR, a leading independent dyno expert and test consultant; and took a car to several local dyno facilities for some real-world testing. Read on as we strip the mask off chassis dyno-testing. The truth? Jack Nicholson not withstanding, we can handle the truth.

What are the basic types of chassis dynos in use today?

Today's chassis dynos fit into one of three broad categories: inertia dynos, water-brake or hydraulic dynos, and electric dynos. The type of dyno affects test results and also the type of test that can be run. A pure inertia dyno can only do full-throttle acceleration runs, but the best modern load-bearing hydraulic and electric dynos with sophisticated control software can do step-tests, constant-speed pulls, and even part-throttle testing to the point that full vehicle road-load simulations can be conducted right on the dyno.

How does an inertia dyno work?

Inertia dynos extrapolate horsepower output by analyzing the dyno drum's acceleration rate using a sophisticated accelerometer and computer software. An inertia dyno works only when the car is accelerating. It uses heavy roller drums of known mass mounted on bearings that allow them to freely rotate. A vehicle is placed in position on the dyno with the drive wheels sitting on the rollers. The car is placed in gear and accelerated at wide-open throttle. It takes a certain amount of time and force for the tires to accelerate the heavy rollers. The laws of physics state that acceleration rate is directly proportional to how much power the tires place on the heavy roller to get it to rotate. The dyno software monitors roller velocity and the time it takes to arrive at a rate of acceleration and estimates power at the rear wheels. Using data from an engine-mounted inductive probe, the software then graphs the power and gear-compensated engine torque against engine rpm. Some inertia dynos also attempt to estimate flywheel power and torque numbers based on mathematical models and data from additional sensors.

A pure inertia-only dyno can only calculate power by measuring the rate of change in acceleration (that's why it's called an inertia dyno), so it can't do loaded tests or step-tests. The inability to perform load tests makes it difficult to accurately establish optimum timing and fuel curves for use under varying driving conditions. On today's inertial dynos, the static roller weight or resistance isn't adjustable to match vehicle weight; depending on the dyno software and whether the inertial test function can be combined with a strain gauge to effectively change the rollers' inertia trim-as is possible on higher-end multifunction dynos-extremely light or heavy vehicles might fall off the curve, and turbocharged engines won't build boost as they do in the real world. On the other hand, with mainstream vehicles weighing around 3,500 pounds, the results are repeatable with minimal setup time. Dynojet is the most common pure-inertia dyno in use today; however, some of its newer models also have an eddy-current option (see below).

How does a hydraulic dyno work?

Similar in operation to common engine dynos, the hydraulic or water-brake dyno relies on a constant speed brake or absorber coupled to a rotor that has a rotating element and a stationary element. While on an engine dyno, the flywheel is directly coupled to the dyno's rotor-on a chassis dyno, the tires turn rollers that in turn couple to the rotor. Although the large rotor mass stores energy and provides an inertial mass that must be overcome, modern dyno software in theory adjusts for all this. An exception is the unique Dynapack hydraulic dyno in which the load brake attaches directly to the axle, taking the tires and rollers out of the equation.

Either way, braking force generated by the rotor elements is used to absorb or match the powertrain output. The braking force is provided by water or hydraulic fluid. A strain gauge may be used to measure the torque reaction between the rotor's rotating element and a stationary element, then (just like an engine dyno) calculate the horsepower output. However, some hydraulic dyno manufacturers like Dynapack don't use a strain gauge and instead measure fluid (hydraulic) power required to hold a vehicle at a constant rpm (the basis is: horsepower = [gallons per minute of flow through the rotor x psi] 1,713). Theoretically, the fluid horsepower should equal vehicle power.

The drawback of a hydraulic dyno is the same as similarly configured engine dynos. There's a slight lag while the impeller cavity fills up with fluid that in turn translates to a slight lag when you first snap the throttle wide open. This causes an initial inaccuracy until the mechanism stabilizes. Dynapack gets around this with an optional "start-point stabilize" option that holds rpm steady at the start of the test until the engine stabilizes. Hydraulic dynos require periodic calibration, as accuracy can degrade over time if the device isn't properly maintained.

How does an electric dyno work?

How do chassis-dyno readings compare to engine-dyno readings? Is the amount of power lost through the drivetrain predictable with any degree of accuracy?

Many factors that influence test accuracy are common to all dynos, including engine dynos; these include temperature, airflow, barometric pressure, and torque calibration. But on chassis dynos, many additional factors can affect the results, factors much harder to control than those typically encountered on an engine dyno. Drivetrain losses vary according to gear selection (testing should usually be performed in the transmission's 1:1 gear to minimize this factor), fluid temperatures, acceleration/ load factors, drivetrain inertia, brake drag, the vehicle tie-down method, the weight over the axle, and tire selection, growth, and slippage.

As TMR's Jeff Bert puts it, "An engine dyno is like weighing yourself with no clothes on; a chassis dyno is like trying to weigh yourself wearing clothes-sometimes you'll have shoes, sometimes pants, sometimes wet clothes, sometimes dry clothes." Although some of the latest dyno software adds sophisticated formulas and assumptions in an attempt to correlate rear-wheel numbers with flywheel numbers, there is really no way to measure, predict, or otherwise determine engine flywheel power from a chassis-dyno test with any repeatable certainty, particularly when using a common acceleration test.

People frequently claim that drivetrain loss is about 10 percent with a stick or 15-18 percent with an automatic. But SuperFlow's Harold Bettes says, "That's a percentage of what number? Obviously 15 percent of 400 hp is different than 15 percent of 700 hp." Also, flywheel horsepower minus 15 percent is different than rear-wheel horsepower plus 15 percent. Late-model drivetrains are often more efficient (they suffer fewer losses) than classic musclecar drivetrains. So assigning a fixed number is a guess, at best-and in a sense, it's really irrelevant: "The goal should be to end up going faster," says Mustang Dyno's Michael Caldwell. "That means ending up with more usable force at the wheels in every gear, which means knowing where you are-what works and what doesn't-and how to get to where you want to be. Better tires, more air, more fuel, and more efficient drivetrain parts, gear selection, and tuning will make it faster. Comparing engine power to wheel power is a distraction."

If you must have a comparison, the most accurate, repeatable way to compare results from an engine dyno and a chassis dyno is to conduct a loaded step test on each system, stabilizing the engine at each selected test rpm point. Ideally you'd do both tests in identical atmospheric conditions, making sure coolant and oil temperatures are the same. Any remaining difference would yield a fairly accurate estimate of the parasitic losses through the drivetrain.

What are the primary factors affecting chassis-dyno test accuracy? Do they vary depending on the type of dyno used?

There are two ways to look at accuracy: accuracy of the dyno itself, and the effects of introducing other variables outside the measurement system that can affect what the car is actually doing. As long as the dyno is properly used and calibrated, then all dynos are accurate for the conditions. The pitfall is trying to compare results across different platforms.

Tire characteristics become an important factor on those chassis dynos relying on a roller-to-tire contact patch. Inlet air temperature differences can have a big influence on the numbers. Overall shop environment and operator consistency play a big role. For example, is the exhaust routed efficiently out of the test facility, or is the car inhaling its own exhaust? The latter negatively affects power. Does the operator have a consistent touch on the throttle to achieve repeatable results?

As drivetrain fluids heat up, internal parts get more slippery. Bearings in the trans get warmer. On computer-controlled engines, the computer automatically changes the way the engine runs based on atmospheric conditions and other factors. Some of the newest vehicles require that all four wheels spin at nearly the same speed. If they don't, the ECM cuts the power to prevent wheel slip. These vehicles require an AWD dyno even if they are two-wheel drive.

To the extent possible, you want to run multiple tests within a consistent shop environment, including similar ambient temperature and pressure. The more rigorous the setup, the more exacting the dyno calibration (on those load-sensitive dynos relying on brakes and absorbers), and the more controlled the overall shop environment, then the more repeatable the results, and the more suitable the facility is for measuring the effect of small incremental changes. Still, once you get down in the 5 percent range on the average chassis dyno in the average facility, it is difficult to evaluate whether the change is real or due to environmental transients.

How much does the transmission type change the results? What about the torque converter?

A lot! Let's look at just one variable: trans-fluid temperature. Fluid often get so hot in performance cars that a separate, auxiliary trans cooler is required to keep the trans alive. What heats the fluid? Power! What generates power? The engine! Engine power that heats up the fluid is wasted energy-it's going into the cooler instead of the rear wheels. Depending on who you're asking, this could be 3-5 hp alone lost just from heating the fluid.

Loose torque converters heat up more, turning power into heat; however, converter slippage itself doesn't actually cost power. A converter multiplies torque, but as it does so, output rpm decreases (from slippage) as compared to engine rpm. Because rpm and torque are inversely related in the calculation of horsepower, net power output in theory should remain the same. But due to this converter slippage-particularly with a high-stall converter-the values at a given rpm point may differ from those seen if the same engine were tested on an engine dyno, and they may be down slightly from an otherwise-identical vehicle equipped with a manual trans. With an automatic, especially one with a loose converter, the rpm at peak power will also change compared to what an engine dyno may reveal, and the curve may have peaks and valleys as the converter comes close to its stall speed. Some sophisticated dynos use various methods to compensate for this effect. For example, on the SuperFlow chassis dyno, one can measure the engine speed (rpm), the wheel slip (with the non-contact optical tach), and the roll speed to derive the converter slip ratio.

Will different final drive ratios affect the dyno readings?

This one's tricky. First, there are potential discrepancies because different gears have different inertia values, generate more friction, and change the amount of tire slip. Higher numerical gears tend to be more inefficient, so as gear ratios increase numerically, power levels tend to slightly drop, particularly on an inertia dyno. When torque is multiplied by steeper gears, tire slippage also tends to increase.

However, there's another, often overlooked, factor in the brew: rpm and torque are inversely related to calculating horsepower, so changing the rear axle ratio or testing in other than a 1:1 transmission gear seemingly shouldn't change the horsepower numbers. But this doesn't take into consideration the fact that changing gear ratios changes the engine's rate of acceleration. For example: We know that on an engine dyno, if you change a sweep test's acceleration rate from, say, 300 rpm/second to 600 rpm/second, the flywheel power number (bhp) drops due to the faster rate of acceleration. As an engine accelerates at a higher rate, the power required to accelerate the engine increases, and a greater portion is consumed before it gets to the flywheel. Going to numerically higher gear ratios-whether in the trans (testing in a lower gear) or in the rearend-is like increasing the rate of acceleration in a sweep test. Whether this actually changes a given chassis dyno's reported results depends on how the specific dyno manufacturer does its math. For the most consistent results, always test in the same trans gear (generally 1:1) and rebaseline the vehicle after a rear-axle ratio change.

How do the tires affect the results?

The tires are part of the overall drivetrain and can be considered a test condition. "If you change a test condition, you may see a different result," says Mustang Dynos. "A tire change can result in more or less force at the wheels (tire diameter), more or less tire inertia, and in some cases more or less traction (coefficient of friction). All of the above affect power at the wheels and where it occurs in the speed range." And according to SuperFlow, "The tire is a power absorber, so tire losses will vary with drive-axle weight, inflation pressure, tread pattern, and carcass design, but generally account for 1 to 3 percent of the total wheel power."

Tires will grow as wheel speed increases and the rubber heats up. Different tires have different coefficients of friction, which could impact the amount of slippage on the rollers. Tire changes affect an inertia dyno most, as it changes the effective rotational mass and overall gear ratio. As Dynojet puts it, "Going to a smaller-od tire is like increasing gear ratio; you lose horsepower."

What methods can be used to fudge or cheat chassis-dyno power numbers? What is a good giveaway that the numbers may be bogus or otherwise unreliable?

There are many ways to produce inaccurate results. Some methods are malicious; some are from poor test procedures. We've discussed how myriad environmental factors can affect the results. Obtaining consistent, accurate results in the first order requires controlling these variables to the extent possible. "If nothing is controlled, it's all bull," says Jeff Burt. "The more things are controlled, the more accurate the test will be." At a minimum, the facility should accurately correct for atmospheric conditions, which is standard practice for engine dyno-testing.

Varying the tie-down method or tension from run to run can significantly alter the results. Other ways to skew results include changing tire pressure, testing the car when the engine is very hot or cold, lying to the software about estimated wheel slip, hacking the software in general, moving the external dyno cooling fans closer or farther from the air inlet, and placing the temperature sensor in unusually cold or hot air so it skews the SAE correction factor.

Although hydraulic and eddy-current dynos have the potential to be the most accurate, their myriad test regimens make them more vulnerable to the whims of unscrupulous operators. The strain gauges must be correctly calibrated. Vehicle weight, drag coefficients, and other variables must be correctly entered into the software if load testing is combined with vehicle simulation mode. Dynojet claims it's harder to fudge the numbers on a pure inertia dyno: "Since the mass is fixed, the actual measured results will be the same every time. If you looked at a dyno test as an experiment, the dyno would be the control," with the car, vehicle dynamics, and atmospheric conditions the variables. "On our dynos, the atmospheric conditions are sampled automatically, and there aren't any other user inputs that could skew the results one way or the other."

However, repeatability does not necessarily mean accuracy. Some experts maintain that inertia-dyno data may not actually represent the true power and torque produced by the vehicle being tested. For dead-nuts accuracy, the load-bearing dyno remains the standard. Just be suspicious of extremely favorable correction factors, power gains that don't make sense based on empirical evidence, or unusually high numbers. According to SuperFlow's Bettes, "A very good method of evaluation is to have the speed versus time plotted or to graph engine speed versus time. Another indication is the evaluation of how much fuel flow was used, and looking at brake specific fuel consumption (BSFC) numbers. If in doubt on the correction process, one should ask to see the correction factor and its arithmetic components."

Finally, as previously stated, each manufacturer has its own math formulas buried deep within the dyno software. A competent operator will know the vagaries of his machine and how to compensate for them. However, the average hot rodder should realize that the results of chassis-dyno testing should not be used to compare one car against another; they are best suited for evaluating the effects of incremental changes on the same vehicle.

In other words, dynos are for tuning, racetracks are for racing.

The Real WorldSo much for theory. What's actually happening in the real world? Staff Editor Stephen Kim took a brand-new 4.6L SOHC-engined, manual trans-equipped Mustang to four different dyno facilities for the same test regimen the average guy off the street would get. The car was tested on APEXi's unique Dynapack, which uses a hydraulic absorber and attaches directly to the axles; on MagnaFlow's Mustang eddy-current and Dynojet inertia rigs, and on a SuperFlow eddy current at Westech Performance.

Due to its ubiquity, the Dynojet inertia dyno has become the standard that many tuners compare their results against. The nature of an inertia dyno produces a more common yardstick for everyone, and to some extent even permits comparing results across different facilities, but it may not reflect a vehicle's true output. However, load-bearing dynos require more setup time, input variables, and operator skill to generate valid results; and after all the extra dialing in, their numbers are typically more conservative than inertia figures (although experts say they more accurately reflect the true power output).

Some dyno manufacturers have different definitions of the term load test, defining it as applying additional simulated loads beyond the standard loads imparted by a non-inertia measurement system. Our definition of a load test is any test where torque or power is primarily measured by some type of load-bearing turbine as opposed to extrapolated based on the vehicle's inertial rate of acceleration.

Using this definition, the Dynapack and Mustang tested in their standard load-test configurations, Dynojet tested in its standard inertia-only mode, and Westech's SuperFlow installation-even though its eddy-current dyno is capable of performing sophisticated load tests-elected to run in pure inertia mode only. Westech states it deliberately calibrates the inertia test cycle to duplicate the Dynojet "since that's what everyone around town is comparing their cars against." Westech's standard corrected inertia test on the eddy-current SuperFlow were within one number of the Dynojet facility's. Westech says it must charge extra to run full vehicle-load-bearing sims due to the much more extensive setup times required.

As predicted by the experts, the more conservative (accurate?) load-bearing dynos were down as much as 30 lb-ft and 18 hp in comparison to the inertia tests, but the deviation between Dynapack and Mustang was significantly more than the max five number deviation between the inertia tests on the Dynojet and SuperFlow.

MagnaFlow (on the Dynojet) and Westech (on the SuperFlow) also repeated the tests multiple times, allowing approximately five-minute intervals between each one. The variations at the peaks were less than two numbers, certainly answering the repeatability question, at least at those facilities.

Where the capability existed, the data was corrected to both traditional performance industry standards, as well as to SAE's net-power correction factor, the less-favorable standard used by most new-car manufacturers. As expected, the SAE numbers were down about 4 percent compared to standard corrected numbers. And of course these rear-wheel-derived figures were also down compared to Ford's official ratings, which were taken at the flywheel.