gripe about horsepower numbers

[cubicleslave]

I'm relatively new to the subject of tuning, but I have this question that has been on my mind for a while. People put their cars on the dyno, and they remember mainly one number from the experience (actually two): the peak horsepower and the peak torque. Then they go out on the track and try to get their best quarter mile time. From what I can tell, the dyno number does not necessarily predict how they are going to do at the track, because the car is not continuously operating at the peak rpm. I see some people say that a particular tune or mod made their car feel more powerful, although the peak HP and Torque did not improve, or went down slightly, or shifted to a different rpm value. I also see some cars being labeled as "dyno queens", obviously meaning that they do well on the dyno but not on the track. So, it appears that what really matters for a car's performance is not being captured in the way we dyno. I have this idea: it is the total area under the graph on a dyno chart that really matters, not the peak. I believe if we look only at that, we'd get much better correlation between dyno results and track results. And one convenient way to measure the area under the graph is to simply take its integral. This should not be hard to do, since dyno charts are printed on a computer, and a computer is good at taking integrals. Even my pocket calculator takes integrals. One horsepower is defined as 745.7 Watts, or 745.7 Joules per second. If we take the integral of the dyno chart, we would therefore end up with just one number: Joules of energy. Of course, a standard range needs to be agreed upon to take the integral, otherwise we would not be able to compare two different dyno runs. Something like from 2000rpm to 6000rpm might be a good range. I would bet anything that if this system were used, the track performance of a car would correlate very, very well with the dyno result (Joules of energy). And instead of bragging about the horsepower their car puts out, people could brag about how many "standard joules" it delivers. We still get one number from a dyno pull, but a number that matters more! What does everyone think of this idea? Has anything similar ever been proposed?

[Rez90]

too much free time.

[fostytou]

You are right, we should do this....
The problem starts coming to how you figure your defined area. One car may have a 5kRPM redline, while another car with upgraded springs/retainers or a desmodromic valve system may be able to rev to 16kRPM.
You also have to consider where you are using the powerband. At the drag strip you may never dip below 5k or even higher in some cars, at the road course you may use a wider, but lower band.

So this is where human intelligence comes in. If we had a simple system to calculate this for a given RPM range (or range per particular application) it would be just as easy to overlay a dyno chart on top of another and compare the two. Being intelligent humans we could figure out the rest on our own.

There are also problems with dynos reading differently, the amount of time a dyno run actually takes (the numbers/graph aren't everything) and plenty of other problems that make putting this into objective, scientifically comparable data nearly impossible.

[cubicleslave]

You have good valid points there. I just wanted to suggest something better than taking the peak HP and torque number from a dyno pull. But maybe there could still be some useful application of the integration technique. For example, somone doing a "before vs after" dyno pull to gauge the impact of his mods could use the integration method over a range of his choosing. Then he wouldn't have to do a "visual integration" by overlaying the two dyno charts. Or maybe some standard integration ranges could be agreed upon and defined by some standards group, say, one for road use, one for drag, etc. Thanks again for your input!

Originally Posted by fostytou

You are right, we should do this....
The problem starts coming to how you figure your defined area. One car may have a 5kRPM redline, while another car with upgraded springs/retainers or a desmodromic valve system may be able to rev to 16kRPM.
You also have to consider where you are using the powerband. At the drag strip you may never dip below 5k or even higher in some cars, at the road course you may use a wider, but lower band.

So this is where human intelligence comes in. If we had a simple system to calculate this for a given RPM range (or range per particular application) it would be just as easy to overlay a dyno chart on top of another and compare the two. Being intelligent humans we could figure out the rest on our own.

There are also problems with dynos reading differently, the amount of time a dyno run actually takes (the numbers/graph aren't everything) and plenty of other problems that make putting this into objective, scientifically comparable data nearly impossible.

[fostytou]

Originally Posted by cubicleslave

You have good valid points there. I just wanted to suggest something better than taking the peak HP and torque number from a dyno pull. But maybe there could still be some useful application of the integration technique. For example, somone doing a "before vs after" dyno pull to gauge the impact of his mods could use the integration method over a range of his choosing. Then he wouldn't have to do a "visual integration" by overlaying the two dyno charts. Or maybe some standard integration ranges could be agreed upon and defined by some standards group, say, one for road use, one for drag, etc. Thanks again for your input!

Agreed, it would be great for a person comparing their own numbers. Probably wouldn't be very difficult to program either.

[Flyin EVO 2]

There are even more factors to consider, final drive, tire size(height), aerodymanics, speed of shifting, and recovery of boost are just a few. On the larger turbo cars driving style is just as important as the power curve. In a small turbo, large power curve car you can be lazy with the shifts and it dosn't affect the mph and et too much. If you do the same thing on a 42 r you'll be running 35r times if your lucky.

[Bebo]

How about you just make a program for drive cycles? I'm sure F1 does this. You couldn't completly simulate a 1/4 mile run but, you could get it pretty close. You could then tune it for that instead of just power #'s. Just my $0.02