Originally Posted by magnamx-5

I've noticed about 12-15% on average.

Originally Posted by Keith

There is no set percentage - it's a mix of fixed losses and percentage. If you want to be conservative, use 25 hp. If you're looking to sell parts, use 25%.

Originally Posted by Corky Bell

Suppose your motor is rated at 120hp, and the chassis dyno suggests 100rwhp. Then 20 hp can reasonably be assumed to have been lost in the tranny, driveshaft joints, diff gears, halfshafts, wheel brgs, brake drag..........

If one raises output to 200rwhp by whatever, were the gears changed? The diff gears, the halfshafts...... and so on....... Essentailly, the same 20 hp will be lost in the driveline, because nothing changed. Did you add bigger bearings? Larger U-joints? Molassas for gear oil? Probably not.

Essentially then, the drive line losses are 20 hp, rather than a percentage.

Originally Posted by Lance Schall

The SAE is an industry standards organization, not a regulatory body. The SAE has no capability or authority to verify claims or levy fines. A company may say it has an engine with a certain SAE horsepower. This implies a certain rigor and method for that determination. It does not confer any authority to the result beyond the honesty of the company. The FTC or Attorney General's office would have to dispute these claims. That said, I suspect that manufacturer's FWHP claims are anywhere from almost right to 15% optimistic.

Gearboxes and differentials get hot. The energy to create this heat comes from the gasoline. What is the mechanism? We have convection of hot air (through radiator and along the outside of the gear case) and conduction from the engine block. These two factors are already waste heat that is not doing any work to turn the crankshaft anyway. This heat does, however, confuse our attempts to accurately measure and quantify the energy lost in the driveline. Now let's look directly at what happens to the power that makes it to the input shaft. Inside a gearbox/differential we have seal friction, bearing friction (ball bearings rolling in races), windage (gear teeth spinning in air), stirring of oil (gear teeth spinning in oil), rolling tooth face friction, sliding gear tooth contact (especially the hypoid pinion gear in the differential), oil shearing (synchros and gear tooth squeeze out).

Of these, seal friction and stirring of oil are proportional to gear velocity only. These would lead to a constant drivetrain loss. Windage is probably small enough to ignore completely. Bearing friction and gear tooth contact forces, however, are proportional to the torque being transmitted. These sources of energy loss would seem to dominate the others and lead to a percentage loss in the drivetrain as a better approximation.

So, if it is a percentage, how do we calculate it? There is probably not enough accuracy in a DynoJet to get a good answer. Something on the order of 10% is not too bad of a guess for a Miata. So a 110 hp car will show 100 hp at the wheels. And if you turbocharge your Miata to double the engine output to 220 hp, you'll see 200 at the wheels. I would guess this drivetrain loss percentage to be similar for all cars. A 450 hp car would expect to lose 45 hp. While I'm pulling numbers out of the air, I would think this percentage would perhaps vary from 5-15% for different cars.

Consider the differential in a turbocharged Miata. A component designed to dissipate 10 hp is now dumping 20 hp or more. Temperature rises. I'd use good oil and change it often. In high power reduction drives (not necessarily automotive, but I have seen it on race cars and heavy trucks) oil can be pumped through a radiator. A typical point to return the oil to the gear set is to spray it directly at the gear tooth interface.

By the way, the type of differential does not effect the situation much, particularly on the DynoJet. If the differential is not differentiating, it is just along for the ride.

Then there are the problems with the inertial dynamometer itself. An engine dynamometer typically uses a mechanism not completely unlike a torque converter to load the engine. To get a FWHP vs RPM graph out of this system, the engine is held steady state with a wide open throttle at various RPM. The energy goes to pumping and heating the water in the torque converter. Water is used because it is cheap and plentiful. A large reservoir can be provided and/or hot water dumped down the drain. On the engine dynamometer, the torque supplied by the rotating engine is directly measured with a scale and the RPM is recorded. These two values fit the fundamental definition of HP which is easily calculated.

In addition to conventional FWHP plot, the engine dynamometer’s variable load capability can be used to do all kinds of interesting experiments with varying load at constant throttle settings, constant controlled engine acceleration rate at WOT, and varying both throttle and load to simulate road testing.

In any case, the SAE methods for classical FWHP can be found in SAE J 1349 and SAE J 1995 (These standards are not available individually from the SAE, but rather bound together in several enormous tomes that can be had for something like $350). Accessories, intercooler airflow, fuel temperature, radiator cooling water flow, exhaust system, energy necessary to drive the fuel pump, and a host of other parameters are accounted for. It is important to note that the basic scientific practice of controlling all of the variables in your experiment except the data you are after is demonstrated in the SAE standards. If SAE methods are used, power will be comparable from one engine to another.

The disadvantage to this method, of course, is the requirement to remove the engine from the car for testing and the frightful expense of a proper engine dyno setup.

Hence, the development of the inertial dynamometer. Instead of loading the engine directly, we accelerate heavy rollers with the rear wheels. The raw data that comes from the rollers is rotational speed at a given time. The slope of this curve (roller rotational acceleration) and the inertia of the rollers can be combined to determine the tangential force applied to the roller surface. Dynojet simplifies the calculation by stopping here and calculating RWHP and graphing against engine RPM.

Given the rolling diameter of the tire, tire rolling friction, inertia of the engine, inertia of the drivetrain, rear end ratio, and transmission ratio (usually 1:1 gear) the engine torque would be calculated. Overlay this with engine RPM in the same time scale and FWHP could be calculated. By asking whatever rear wheels that are presented to the Dynojet to accelerate the same rollers, we have sort of simulated an instantaneous engine swap into a standard weight car where we then devise RWHP by comparing 1/4 mile times. How does this method differ from the SAE engine method? Where is error introduced? Why doesn't the Dynojet software just spit out a FWHP instead of making us guess?

1. ENGINE LOAD DIFFERENT. The inertial dynamometer can not measure steady state output. The resistance comes from the energy necessary to accelerate the rollers. The test is complete after a single pass to the engine’s RPM limit. And this acceleration rate can not be controlled by the experimenter, it is determined by the torque and rpm of the engine under test. Too light rollers yield smaller numbers and rollers that are too heavy give higher numbers but are more representative of road use.

2. EXPERIMENT NOT CONTROLLED. Few of the engine operational parameters specified by the SAE are monitored or controlled on the inertial dyno. Results are not comparable from one dyno to another.

3. DRIVETRAIN INERTIA UNKNOWN. The energy necessary to accelerate the drivetrain must be estimated by the software. Since the inertia of the rollers is constant, as power is increased or decreased the acceleration varies. The inertia of the engine is also constant. The faster the engine is accelerated, the less power is available at the flywheel. Insignificant compared to the heavy rollers? Have you calculated the energy in a 20 lb flywheel rotating at 7000 RPM? The top speed of the rollers is only about 800 RPM. At the end of a dyno run, 2% of the total stored energy is in the flywheel. Add the clutch, crankshaft, transmission, driveshaft, ring gear, and wheels. We might have a 5% error here. A 50% reduction in flywheel weight would erroneously show an increase in HP on an inertial dynamometer.

4. DRIVETRAIN FRICTION UNKNOWN. The drivetrain friction must be estimated by the software. A major component of drivetrain friction is tire rolling friction. This depends on tire pressure, tread, tire temperature, and speed. This item and the one above can be estimated by a clutch disengaged coastdown. However the effects of the two are co-mingled. Coastdown data is not collected and Dynojet does not attempt this calculation. Drivetrain friction and inertia do not both add linearly with engine rpm or roller speed, so they must be estimated.

5. ROLLER INERTIA ESTIMATED. The moment of inertia of the roller itself should be measured. This is not as easy as it sounds. They might suspend the roller from one end on a rotational pendulum. I don’t know how Dynojet does it, but I’d be surprised if it was like this. An easier solution might be to estimate the inertia by weighing the roller and considering its geometry. Since an error at least 5% would be expected, perhaps this is the method Dynojet used to get the 175 to 185 number they list on their website. A better method that comes to mind is to “motor” the roller assembly on an engine dyno, and measure data to determine the roller inertia and the behavior of its bearings. By the way, the Dynojet website lists its roller inertia in units that are not moment of inertia (kilograms force times meters squared instead of the correct kilograms mass times meters squared). Where’d they go to school? The Downing/Atlanta Sebring Institute of Technology?

6. ROLLER'S BEARING FRICTION UNKNOWN. Bearing heat in both the car’s drivetrain and the dyno will effect the friction. It would be useful to measure the temperature of the roller bearings, but only if the software and scientific data were present to know what to do with the information. Dynojet does not collect this data. For repeatability it would be advisable to get the dyno bearings up to operating temperature and keep them there.

7. TIRE SLIP NOT MEASURED. The inertial dynamometer does not capture tire slip, which depends on tractive load, speed, and weight transfer. At the dragstrip, weight transfers to the rear wheels upon acceleration. On the inertial dynamometer there is no weight transfer although the effect is minimized by running the test in 4th gear. Tire slip is disregarded.

8. ROLLER HAS WRONG INERTIA. The "correct" moment would not be a constant. By the time a Miata gets to redline in 4th gear, the Dynojet roller has about 2.5E6 Joules rotational energy. The car driving on the road would have 1.3E6 Joules linear kinetic energy. The roller is almost twice as heavy as it should be for a Miata. The advantage to a heavy roller is that it swamps the effect of drivetrain inertia making the computer calculations more consistent. Are we effectively measuring the power of a Miata by sticking its engine in a lard assed Expedition, comparing race horses while they are doing the work of draft horses? Only at low PRM. Aerodynamic forces increase with the square of velocity. Lacking aerodynamic drag, the Dynojet does not provide enough resistance at higher speeds (the important half of your dyno graph). So even as the inertial dynamometer has problems predicting FWHP, it is equally challenged to simulate road testing. These load differences effect volumetric efficiency impacting ignition and mixture settings. The perfect Dynojet adjustments will not be the perfect road adjustments. By the way, power necessary increases as the cube of velocity. To go twice as fast requires 8 times the power. Put that in your 200 mph Miata and smoke it!

9. ALTITUDE AND AIR TEMPERATURE CORRECTION IS INADEQUATE. Altitude and temperature correction included in the SAE standards does not lend itself to Dynojet application. The equations are more complicated than expected. Although trivial for a computer, some obscure data input would be required. Even after turbochargers and intercoolers are installed and provided with the appropriate air flow, correction equations are still applied. Along with the expected variables of ambient air temperature and absolute pressure are correction exponents selected from a table in the SAE standards. These parameters are determined by forced induction type, charge cooling method, and the engine's specific fuel consumption. This later correction figure is determined by yet another function. I don't know what Dynojet software does with the their temperature probe, but it isn't an SAE standard dynamometer correction. Not to say there isn't anything else constructive to do with temperature, I just don't know what it is.

10. FUEL CORRECTION FACTOR IGNORED. In addition to the air correction factor there is a fuel correction factor, largely based on the specific gravity of the fuel used in comparison with reference fuel.


Some of these error sources are avoided by Dynojet when the computer spits out RWHP. Most of them introduce uncertainty in Dynojet numbers. All of them must be considered if we wish to estimate FWHP from inertial dynamometer RWHP. I doubt that Dynojet's source code is available to be dissected and potentially modified by their customers. Dynojet's target is not rigorous scientific research. It does the job it was constructed to do. A more scientific piece of similar equipment is made by Superflow. It cost three times the price of a dynoJet. Would you pay three times as much for each pull? That's why Dynojet is humping all the inertial dynamometers out their back door to every Tom, Dick, and Harry in the tuner world.

The SAE recognizes that even SAE HP is not a reliable predictor of road performance. SAE HP gives scientists a controlled environment for experimentation as well as providing industry standardization. There is an entire subset of SAE study and technical papers concerned with "road load" testing. That is, how do we model real world performance in the lab? The basic road load equation used is a multi-term second order polynomial where factors are lumped into constants depending on the exponent that best describes their behavior.

Given all this confusion, why use an inertial dynamometer? Is it just because it is cheap and easy? Why indeed? Inertial dynamometers are dead simple repeatable. With a engine dyno, we would have to fiddle with uncountable apparatus adjustments to make sure we had accurate and repeatable SAE numbers. With an inertial dynamometer, you may not be able to compare your results with anybody else on the planet and you may never know your actual FWHP. But from one day to the next, in your shop, you’re going to quickly and easily get repeatable and comparable results.

Originally Posted by Prospero

Most common I have heard over the years is 15% for a manual and 20% for an auto.

It's incorrect as hell but does well enough for me.

Cheers,
Prospero

Originally Posted by jayc72

Fixed.

Originally Posted by UrbanSoot

by common sense and basic physics knowledge i posess, it HAS to be percentage rather then static number.