Why is it so hard to measure HP?

This is a serious question.

A horsepower is a unit of measurement. No different from an inch or a pound or a liter. Yet we get vastly different power measurements using a Mustang Dyno, Dynojet, or Dyno Dynamics setup.

People never say "Well, my car has a 100-inch wheelbase using a Mustang ruler, but a 105-inch wheelbase using a Rulerjet." But that's EXACTLY what you hear regarding power numbers.

How has this not been resolved? Even if the different dynos use different methods of calculating power, a horsepower is still a standardized unit.

The problem is... WHICH dyno gives the standardized unit? It's all in how it's calibrated, the operator, etc etc etc...

However, i do agree with you, and this frustrates me horribly as well. I have a suspicion that one of my cars dyno'd easily 60hp optimistically, and while i know that horsepower is simply a number and i care more about what the car does than what the paper says... it's still frustrating.

In the future, i'll be taking my cars to the dynos that are typically called "heartbreakers." I'd rather be surprised by the actual performance of my cars than disappointed relative to paper.

Dynos vary as said before by calibration and operator etc. They are also varied in how they are constructed and measured.

In the end look at it as a good guess or approximation. Just use the same dyno, that is maintained and calibrated for comparisons.

Will wrote: Even if the different dynos use different methods of calculating power, a horsepower is still a standardized unit.

It's not, though.

Part of the problem is that dynos measure power in different ways and are affected by different variables. A hub-mount dyno is going to be unaffected by tire losses, an electric dyno is going to be unaffected (or affected differently!) by lightweight rotating assemblies, compared to a purely inertial dyno. Calculations are made to compensate for this but it can never be equal from one species to the next.

That they all read even close to each other is pretty nice IMO.

When I want power numbers, I back-calculate from injector pulsewidth. When making changes, I see if I need to add fuel or remove fuel to see if it was beneficial or not. EFI is awesome.

Say, did you know that I found something like 65hp on a 389-based Pontiac engine once by doing a decreasing-RPM pull instead of an increasing-RPM pull, on an engine dyno? Figures don't lie but liars can figure.

Will wrote: No different from an inch or a pound or a liter.

Totally wrong. Also, weight is not a standard either. Weight is given at a certain altitude if it is to be given accurately (generally @ sea level).

However distance, weight and volume are all physical quantities you can look at. Any error in measuring tools is pretty small, because its one measurement.

Power is a force over a distance in a given amount of time. Basically 3 sources of error with no direct measurement.

Distance is physically how far one point is from another. Pretty cut and dry, you measure it directly. Power is simply energy per unit time. A few different ways to come at this one, each with slightly different results. each method will have multiple sources of error with knowing exact weights, speeds, time.... It is simply not the same as directly measuring length.

it doesn't matter how much power you make- it only matters how much you get to the ground.. so the only really accurate measure of HP is the timing setup on a drag strip. dyno's can be calibrated wrong and the reading can be affected by something as simple as a few more or less clicks on the ratchet strap that holds the car in place, but a drag strip measures how fast you get from one set of lights to the other.

One other thing is the simple limitations of the dyno.

One guy who recalibrated a dyno noted that the unit in question used an 8-bit ADC on the load cell. In theory, then, if the dyno could meaure up to 1000ft-lb of torque, then the smallest change it could measure would be about 4 ft-lb, and THAT assumes no in-use sensor drift, resistance, noise, etc. Feel confident in those Bob's Discount Liquor & Dyno Shop dyno readings that give numbers to the tenth or a hundredth, now?

Because power is calculated, not measured. Dynos measure torque, then calculate hp (work done) based on many assumptions. An easier example is at a job, it's easy to count how many ditches I dig in a day (= torque), it's much harder to assess how hard I worked (=hp), which depends on lots of things like how hot it was, whether the shovel had a short or long handle, etc.

If a dyno is repeatable and reproducible, then it WILL accurately measure differences in HP, and that's all a dyno is really good for anyway. Remove hp/tq from the y-axis of a dyno plot and two overlaid torque curves still tell you most of what you're running on the dyno for in the first place.

i'll take an accurate dyno over a precise dyno any day.

It's important to separate chassis dynos from engine dynos. The latter are a lot more repeatable as they have fewer variables. There are standards as to how power/torque should be measured on them, including what accessories are attached at the time.

You can also measure using a sweep (run from x RPM to y RPM and measure the amount of torque, either directly or by how fast a flywheel of a different mass is accelerated) or at a fixed RPM. The latter is more consistent but also puts more of a load on the engine - an artificially high load - and heat management can be a real challenge.

Chassis dynos vary partially because there are so many types. The Dynojet is a pure inertial dyno, accelerating a big heavy drum. It can't really do fixed RPM tests. Because torque is figured from the acceleration of the drum, any change in the rotating masses will add some error into the measurement. That's why a set of heavy wheels or flywheel will "cost" horsepower. Wheel slippage can also be a problem on high horsepower cars.

Other types such as a Rototest or a Dynapak will bolt on to the hub, eliminating all the ugliness that comes from having a pneumatic rubber interface in the dyno mechanism. Instead of a big flywheel, the engine is pulling against a resistance that can be varied. I don't know the specifics of the Dynapak or other setups, but in the case of the Rototest torque is measured directly with a strain gauge - quite accurate. The operator can hold the engine at a given RPM and get a very accurate torque number - again, with the problems of rapidly rising heat and an artificially high load on the engine. The operator can also emulate a Dynojet-type sweep run, which will add in some inertial losses. Then, as Knurled as discovered, you can mess with the sweep and affect the power delivered.

Then, of course, you get into environmental factors. Humidity, air pressure, temperature all play a part in how much power an engine will make. That's why SAE correction factors exist, but they have limitations.

Interestingly, we used to have a Dynojet at the shop. When we got our Rototest. customers wanted to see the "traditional" sweep dyno chart. So we looked at how long the typical turbo Miata pull took on the Dynojet and programmed the Rototest to emulate it. And whaddya know, the numbers came out almost exactly the same.

Drag strips are no substitute for a dyno. There are waaay too many other factors involved to make them a repeatable measurement tool. That's why bracket racing is actually a form of racing

We have a dyno at FM for development purposes. It lets us do terrible things to cars and engines. It's very precise. We never load a car on for the typical "three pulls and a dyno chart" deal anymore, but we do know how to use it to deliver very consistent numbers - consistent from car to car and also consistent with official manufacturer SAE ratings. So it's not impossible, but it's more difficult than you might think.

Be sure to use the same dyno every time. It may be off, but it'll be off by the same amount each time.

And if you're competing in NASA TT events your dyno can be verbal, as in "It makes precisely 240hp and 230ft/lb torque, just like the factory brochure says"

Keith Tanner wrote: Drag strips are no substitute for a dyno. There are waaay too many other factors involved to make them a repeatable measurement tool. That's why bracket racing is actually a form of racing

ET depends a lot on driver but MPH changes very little. That's why bracket racing is ET based, too

Like any other tool, it has its uses.

I'd almost go the other way... a dyno is no substitute for a drag strip.

In a performance application, we use dynos to get a rough estimate of how much power we're making, ergo... how "fast" the car is.

So... why not just cut to the chase?

The problem being that a dyno is a very useful tuning tool, but in terms of numbers.... meh.

If I take the windshield off my Locost, it has a faster ET. Does that mean it has more power?

One thing I want to tack in here: not only are dynos calculating hp from torque and rpm, they're also trying to "standardize" the numbers based on things like ambient temperature and barometric pressure. That's why even though by the math the hp and torque lines should cross at 5252 rpm, they just about never do.

Keith Tanner wrote: If I take the windshield off my Locost, it has a faster ET. Does that mean it has more power?

No, but does it matter? It was faster.

It does matter if I'm trying to determine if the most recent change to the engine was good. Or, assuming I'm smart enough not to change the bodywork, what if there's a headwind?

Reverend, the correction factors for atmospheric conditions affect both power and torque. So as long as you're using horsepower and lb-ft of torque and the same scale for both, the lines will cross at 5252. It's a mathematical certainty.

If they're not crossing at that point, I'll bet the dyno chart shows different scales for each line.

1: you cannot measure HP you can only calculate it 2: your engine does not make a set HP at all times in all conditions 3: different methods to calculate return different results 4: HP is just an number and a relatively useless one by itself

You have to realize that your cars "HP" is going to vary with atmospheric conditions as stated previously (ambient temperature, barometric pressure, humidity, oil temp, oil viscosity, engine temp, fuel temp) then remember that the dyno is also effected by those factors.

A dyno can read differently on a cold run vs. the 30th run in a row as well. To be as accurate as possible you would have to state a given HP, list ambient conditions, and all other variable factors, and that would be honestly a little absurd.

IMO i would only use a dyno to show progress on modifications and changes, forget absolute numbers as they don't really DO much.... I would only compare my HP numbers with my own car and I would know what im looking for as well not just peak numbers.

HP sells cars torque wins races but in the end neither makes a car fast, it may contribute, but its only 1 piece of the puzzle...

Keith Tanner wrote: Reverend, the correction factors for atmospheric conditions affect both power and torque. So as long as you're using horsepower and lb-ft of torque and the same scale for both, the lines will cross at 5252. It's a mathematical certainty. If they're not crossing at that point, I'll bet the dyno chart shows different scales for each line.

Given that the constant 5252 is used in the formul for HP i agree, you would HAVE to have a different scale for the two lines, only way possible... or i suppose if the correction factors were different for the two... but that make no sense, as the dyno should get a TQ reading, apply corrections, calculate for HP, and be done, no reason to reapply the same OR different correction factores once it has been done once...

Keith Tanner wrote: It does matter if I'm trying to determine if the most recent change to the engine was good. Or, assuming I'm smart enough not to change the bodywork, what if there's a headwind? Reverend, the correction factors for atmospheric conditions affect both power and torque. So as long as you're using horsepower and lb-ft of torque and the same scale for both, the lines will cross at 5252. It's a mathematical certainty. If they're not crossing at that point, I'll bet the dyno chart shows different scales for each line.

Heh, i get what you're saying Keith... but i don't think that's really the point of the original question, and i may be guilty in bringing this off-track somewhat.

If the numbers on the dyno change for the better, that'll be good enough to tell you if the most recent change to the engine was good.

But without a true reference point, a number by which all dynos are measured, the number is just a number, and there's no way to tell which numbers from which dynos are more accurate than the others.

bigbens6 wrote: A dyno can read differently on a cold run vs. the 30th run in a row as well. To be as accurate as possible you would have to state a given HP, list ambient conditions, and all other variable factors, and that would be honestly a little absurd.

That's effectively what the correction factors are intended to do, take a number of variables into account (temperature, pressure, humidity). It's up to the operator to minimize the others. Of course, some operators are better than others and some dynos might suffer more from run-to-run variation than others. For example, one that uses hydraulic pumps to provide the resistance will be sensitive to fluid temperature unless something's done to compensate.

92CelicaHalfTrac wrote:

Keith Tanner wrote: It does matter if I'm trying to determine if the most recent change to the engine was good. Or, assuming I'm smart enough not to change the bodywork, what if there's a headwind? Reverend, the correction factors for atmospheric conditions affect both power and torque. So as long as you're using horsepower and lb-ft of torque and the same scale for both, the lines will cross at 5252. It's a mathematical certainty. If they're not crossing at that point, I'll bet the dyno chart shows different scales for each line.

Heh, i get what you're saying Keith... but i don't think that's really the point of the original question, and i may be guilty in bringing this off-track somewhat. If the numbers on the dyno change for the better, that'll be good enough to tell you if the most recent change to the engine was good. But without a true reference point, a number by which all dynos are measured, the number is just a number, and there's no way to tell which numbers from which dynos are more accurate than the others.

There is a fairly good reference point: factory numbers. A stock car should read pretty much the same on any chassis dyno and their power ratings are set under much more controlled conditions. The variation isn't as big as some people make it out to be, I don't think. Not when the tool is set up and used properly.

We were accused of not knowing how to run our dyno by an Integra Type R owner. He'd added up the stickers and our dyno was showing that his car wasn't even showing stock numbers. Obviously we were idiots and our dyno was inaccurate. Of course, when he pulled the motor apart a year later and found the stock pistons had been replaced with low compression ones, he didn't come back to apologize.

ReverendDexter wrote: One thing I want to tack in here: not only are dynos calculating hp from torque and rpm, they're also trying to "standardize" the numbers based on things like ambient temperature and barometric pressure. That's why even though by the math the hp and torque lines should cross at 5252 rpm, they just about never do.

Huh? The temperature & pressure calculations feed into the torque measurement, then you calculate the horsepower from torque & RPM. If it doesn't cross at 5252.something, then either the two curves are on different scales (stupid way to display it, but people do that sometimes) or someone's being playing photoshop with it.

As for why measuring horsepower is so hard -- measuring ANYTHING is hard. Measuring time doesn't seem that hard, does it? It took hundreds of years to invent a clock accurate enough to solve the longitude problem. How about weight? There's a hunk of platinum-iridium alloy in a vault in France that's part of the attempts to solve that problem.

Keith Tanner wrote:

92CelicaHalfTrac wrote:

Keith Tanner wrote: It does matter if I'm trying to determine if the most recent change to the engine was good. Or, assuming I'm smart enough not to change the bodywork, what if there's a headwind? Reverend, the correction factors for atmospheric conditions affect both power and torque. So as long as you're using horsepower and lb-ft of torque and the same scale for both, the lines will cross at 5252. It's a mathematical certainty. If they're not crossing at that point, I'll bet the dyno chart shows different scales for each line.

Heh, i get what you're saying Keith... but i don't think that's really the point of the original question, and i may be guilty in bringing this off-track somewhat. If the numbers on the dyno change for the better, that'll be good enough to tell you if the most recent change to the engine was good. But without a true reference point, a number by which all dynos are measured, the number is just a number, and there's no way to tell which numbers from which dynos are more accurate than the others.

There is a fairly good reference point: factory numbers. A stock car should read pretty much the same on any chassis dyno and their power ratings are set under much more controlled conditions. The variation isn't as big as some people make it out to be, I don't think. Not when the tool is set up and used properly.

Good point, but then how do we explain the wild variations?

Hopefully i'll have a data point to consider once i strap my car again. I have a feeling i'm looking at a 40-50hp variance with no changes to the car in any way.