RR98ITR

It seems that there is enough ignorance of what makes a car go that it leaves room for sophistry in the marketing of horsepower.

There is an idea floating around some of the bulletin boards that I visit that requires some attention. Here's the pitch as delivered by Larry Widmer of Endyn:



Among those who I'm aware of having written about "Transient Response" are John Grudynski of Hytech Exhaust, Journalist Jim McFarland, and even a little bit from Paul Van Valkenburg in RaceCar Engineering magazine.

Typical of good sophistry, there is an element of truth upon which this creative marketing rests.

Larry refers to conventional dyno testing, by which he means the single peak horsepower number. Although it doesn't take a rocket scientist to know that the peak number IS relatively unimportant compared to the area under the curve in the useful powerband, some readers will take this to mean that he is saying that dynos and horsepower curves are relatively unimportant, when of course they are very useful. Larry and John, and probably others, appear to cultivate this sort of misunderstanding in their promotion of their products and services.

Primitive discussion on such topics often amounts to little more than this: "torque accelerates the car, horsepower gives you your top speed." In a debate between a drag racer and a Salt Flats racer this might make a little sense. That would simply be a case of different horses for different courses.

Let's approach this from the fundamentals. An engine produces torque (rotational force). This is absolutely worthless unless we actually rotate the crank. Horsepower is Torque X RPM, divided by 5252. The constant (5252) is for conversion of units - nothing magical.

So we run our motor on an engine room dyno, or we run our car on a chassis dyno, and we wind up with torque and horsepower curves. These curves tell the whole story of our motors WOT (wide open throttle) performance. What accelerates the car is the amount by which the instantaneous surplus of power exceeds the instantaneous load. That's it, there is nothing else.

So is there anything at all to this idea of Transient Response? Well, yes.

How did we get here? Because a single peak horsepower number is "an easy answer" - that's why. That single number is easy to convey and useful for marketing purposes. And of course it is the real answer to one specific question - an intuitive question at that: "how much power does it make". Other such questions are: "how fast does it go", and "how quick is it".

Most of us operate our cars in a window of performance in which that last question is the most relevant one. How Quick Is It means how fast does it accelerate. One way to answer that question is with time in the Quarter Mile. But is that all there is to it? No, not really.

Let's suppose that we have a powerband and gear ratios that are reasonably well matched. Now realize that the shape of the torque curve dictates the shape of the horsepower curve.

Picture a linear horsepower curve running from 100hp to 200hp over 2000rpm. It should be easy to imagine that a concave up parabolic horsepower curve with the same endpoints over the same range will simultaneously feel faster and yet be slower. And it should likewise be easy to imagine that a concave down parabolic curve with the same endpoints will feel like it runs out of breath badly on top but will be quicker.

This is part of where the torque versus horsepower argument originates - when it pulls better in the bottom half of the powerband we call it torque, and when it pulls better in the top half we call it horsepower. Bad language again leads to bad thinking. And speaking of bad thinking, you cannot trust your butt and it's not a dyno. Your butt likes a steep hp curve, and in the absence of timing information IT WILL LIE TO YOU.

So that's what makes a car go - the area under the horsepower curve in the useful part of the rev band. And how do you convey that to somebody? That's the problem. And that's not all there is to it.

The proper application of the word Transient to the discussion of horsepower is in this context: the area under the horsepower curve and it's general shape are what matter most in accelerating and controlling the car.

It should be easy to see that a smooth powerband is desirable for it's contribution to the ease of managing traction. One of the most interesting articles I've found on control of transients was written by Peter Wright of Lotus in RaceCar Engineering about Drive-by-Wire in Formula One, in which he relates the value of controlling throttle gain to ensure a flat torque response to throttle regardless of rpm (not quite Traction Control per se). For little guys like me the best that can be worked toward is a nice flat torque curve.

Less immediately useful, but interesting is Recovery of power output to the WOT curve during the shifting sequence. One of the Race Engineering magazines did an article on WOT Shifting and made some reference to differences in recovery time owing to keeping the throttle open. These differences are very small, and are of extremely limited interest for most applications. That is not to say that a very badly set up car that bogs noticeably whenever the throttle is cycled is acceptable, but rather that it is not the standard on which improvements in this mode of operation should be judged.

Lastly, the idea that Transients, as it's used by some of the Snake Oil guys, is about who makes more power at part throttle is...well, wacky. Again, unless your standard of comparison is a car with engine management problems. On a decent working car with a reasonably flat torque curve the throttle and output are yours to control, and the reason you're at part throttle is because you don't have enough traction to use the rest. Making relatively more or less power than the other guy at part throttle is really not the place to stake a claim for your product.

Well, that's about all I've got to say about that.

Scott, who won't surprise you by telling you that this is all a big joke...I'm not that kind of guy....




[Modified by RR98ITR, 11:51 PM 1/27/2003]

Stock R

wow.

still kinda confused. you brought up the question, but where is the answer?




[Modified by Stock R, 7:57 AM 1/28/2003]

Louie

my head hurts.

Louie~ who has a feeling the butterfly just flapped its wings.

Mac

iTrader: (1)

2 plumbers start up a new business. One gets all the latest gadgets, the van is new and he's got all the business cards and flyers one can buy. Plumber 2's got his dad's belt and a rackety old van and knocks on peoples doors.

Who's still in business?

2 things can be drawn from this. 1) Numbers sell. This isn't the fault of the manufacturer, but you'ra not gonna sit there and prattle on to Joe from Australian about transient load to sell him an exhaust. Which leads me to point 2. 2) The market dictates whats in demand. You can have all the technical terms, all the know how and the most overwhelming data - but the market aint gonna buy that.

Yes its enlightening, and we should damn well care about the finer details, but the majority of the population doesn't.





[Modified by Mac, 8:28 AM 1/28/2003]

Hash Browns

Having read your post, and your quotation of Mr. Larry Widmer of Endyn, I can say this:

From what I understand, The 'Transient Response' is still an important factor to consider.

An engine power curve, displaying horsepower and torque, will display power values at given points. Supposing our engine is like a mathematical function, using 'rpm' as the independant variable, while assuming a wide open throttle, power produced is the dependant variable. Torque is what is measured using the dynamometer, and using the appropriate conversions, we get horsepower.

I see many people quoting the maxims "a flat torque curve is best" and "maximize area under the curve". The mathematical and geometrical representations of these statements need to be considered and thought out properly, which is often forgotten. For many it may seem perfectly clear that these statements hold their value quite well in many instances, but it is also this group of persons that neglects an important aspect of the measurements we get from something such as a dynamometer.

As I noted before, it's perhaps obvious, that a dyno sheet gives us power as a function of engine speed. Engine speed, measured in rpm, is a measurement of rate. Rate is not a measurement of time, but a measurement of frequency per unit of time. This idea is, hopefully, what the "Transient Response" advocates were trying to get at.

To return to the notion of areas and maximizing them, we calculate areas using integrals given a lower and upper bound. If our engine were indeed producing power accord to some function of rpm, we would be calculating total power from one starting engine speed point 'A' to an ending speed point 'B'. How do we pick these values? It certainly isn't arbitrary. It's dependant upon the application of the engine and its coupled components, ie: transmission and gearing. We'll skip that part for now, but we should remember that our engines are mated to transmissions, and those transmissions have many gears. When making applications of our engines, we certainly don't stay in one gear, this would defeat the purpose of having different gear ratios.

I'd like to return again to the fact that our area calculation, or integral, from 'A' to 'B' tells us nothing about time. We could have two engines that have idential dynoplot measurements. They'd have the same power production and have the same engine speed range. Let's call those engines 'Engine X' and 'Engine Y'. What if Engine X were capable of accelerating its engine speed from A to B in a discrete time 't1'. And what if Engine Y were capable of accelerating its engine speed from A to B in a discrete time 't2'. If Engines X and Y were mated to identical transmissions, which is faster? If t1 is less than t2, than the Engine X configuration is faster. If it were the opposite where t2 is less than t1, the Engine Y configuration is faster. This is what I believe the Transient Response advocates have been attempting to explain.

We can take the total power from the range A to B, and they're the same for both engines. Let's call that power P. We understood that we wanted to maximize the total area between those points, but what's happening, and why did we want that? Flat torque curves produce a linearly predictable power delivery. Horsepower is a function of the torque and engine speed. HP = (RPM) * (Torque / 5252) Flat torque curves are for the most part, constant across the engine speed range, and we assume it to be a constant, horsepower simply becomes the product of engine speed, torque and that conversion factor. Horsepower would basically become a function of RPM. But still what does this all mean?

If our torque production were constant, or fairly constant, that means that the number is same across the entire range of engine speed. Geometrically, that's not very useful. If you could visualize the line produced by the the product of (RPM) * (Torque / 5252), you'd get the increasing and very linear horsepower function. This is much more useful to us, because (and here comes the marketing speak), at a given engine speed, we are producing torque that many times. As engine speed increases, we produce that torque even more.

So, returning to the main point, when we calculate the area across A to B, we're taking the sum of all the power that we've been putting down at all points from A to B. That's 'P'. If Engine X did get from A to B in less time than Engine Y, Engine X put down that P in less time than Engine Y, it would be the better of the two. This becomes more apparent if you remember that our powertrain has transmissions with multiple gears.

Supposing that we had a five speed transmission, identical for both engines, we've already established that if t1 is less than t2, Engine X will get from A to B quicker than Engine Y. Then Engine X's transmission will shift to the next gear and do that transition from A to B again. Meanwhile Engine Y will still be trying to make it from A to B the first time. Engine X has already put down P through first gear and will continue to do it through second, and so on. Engine Y will do the same. (There will be objections to the subsequent engine speed range between shifts, these are dependant on the gear ratios, but that A' for the next shift should fall somewhere between A and B. B shouldn't be changing.)

If we were to limit the maximum amount of time that each of our engines were allowed to produce power and go between gears to a finite value 'T' less than the time it takes for Engine X to hit the revlimit in top gear, then we should be able to follow that in that time T, Engine X has put down P more times than Engine Y.

And this is what I believe they were trying to say. There's far too many variables out there in competition that you would likely never get two different engine configurations that follow this model, but it makes sense.

I'd believe it if the high-end racing research and development teams were taking such 'responses' into account for the end-result of optimal performance.

Forgive me if your brain hurts... I don't even know if this answers the question if there even was one.

-Hash

sackdz

In my mind, they are one in the same. I'll try to tie in with your theme and use the same variables.

Given identical car/transmission, and 2 motors X and Y, 'transient response' or the elapsed time from one rpm to another is dictated solely by the torque curve, more specifically the area under it. To say that motor X achieves t1 in less time than motor Y achieves t2 is to say that given an identical load, motor X has more area under its torque curve. Analogically speaking, a stone dropped from 100 feet on the earth strikes groud before a stone dropped from 100 feet on the moon 100% of the time. The same principles apply for power and displacement as they do for acceleration and displacement.

The relationship between instantaneous rpm rate of increase F'(R) and time is directly proportional to the instantaneous torque output of the motor. The only way for rpm to increase is for the torque to be produced and transfered to the ground and increase the rotational velocity of the driveline. They go hand in hand. A motor making less torque over a given rpm range will take longer to accelerate (rpm), period.

Assume that motor X produces a constant 100 ft lbs. of torque between 3000 and 4000 rpm, and 80 ft.lbs. between 4000 and 5000 rpm. Motor Y produces 80 ft.lbs. 3k-4k, and 100 ft.lbs. 4k-5k. These numbers may not be realistic but nonetheless between 3k and 5k rpm,
f(Tx) dR = f(Ty) dR where f(T) is torque T as a function of rpm R for motors x and y. Substitue t (little t, time) for rpm R and you have the exact same relationsionship:
f(Tx) dt = f(Ty) dt Equal area under the curve = equal 'transient response'

That is how I see it. Keep in mind this is assuming all factors equal save the motor itself (weight, traction, gearing, etc.)

[e] Integral symbol didn't show up right. I improvised.




[Modified by sackdz, 8:54 AM 1/28/2003]

ZygSpeed

Good stuff as usual Scott, I'll have to read the rest of the posts after I get to work today.

hehe ... leave it to sackdz to put it in "engineering terms"!


[Modified by zygspeed, 8:33 AM 1/28/2003]

Floyd

Hmm, I can tell I won't be able to digest this one over my bowl of cereal this morning. I'll check it out when I get home.

RR98ITR

My brain only hurts because you wasted alot of words saying very little and added nothing but your share of confusion to the discussion. I have some familiarity with mathematics and I see that you've been thru Integral Calculus - considering your approach to responding then it would have been nice if sloppiness in units hadn't muddied up your understanding of the phenomena under discussion (unless you want to share why you think revolutions per second squared is of interest).

Sack is right - it's the same thing.

Scott, who says it isn't hard to understand.....it's easy.....making a nice torque curve with bigger numbers across the spread and no weak spots - now that's a little harder....


[Modified by RR98ITR, 7:04 AM 1/28/2003]

b19coupe

Ditto. I am glad there are guys out there thinking about this stuff. I just want to step on the throttle and go real fast.

Michael Delaney

perhaps a better variable to measure is thrust (in units of lb.).

thrust is the sum of all the forces acting on an object that results in that object propelling itself forward.

it basically measures the product of mass times acceleration but takes into account negative resistance forces. simple.

it's used in road racing motorcycle performance evaluation and integrates the effects of having a steep wide torque curve and properly sized tires/wheels and gear ratios.

I believe that Hytech, Endyn, and Mr. McFarland are not the only ones using the term "transient response" or "torque transients". Whether they are using it correctly or incorrectly is the result of their inability to demonstrate a statement or definition with empiric data...and therein lies their "fault" if you want to call it that. I believe in Formula One and CART circles, the teams measure partial throttle torque increases and WOT torque increases on a dyno. This is not the same as your neighbourhood Dynojet however. They are linked to a computer which contains data collected at tracks and they can simulate the loading of uphills, straightaways, corners on every track they run. In essence they can do testing without taking the car to the track. They say they measure engine transient response and driveability in this manner.

If you use the search function on your browser and enter the word transient response and Toyota, I believe there used to be a brochure by Toyota gmbh (Eurpean arm which runs their F1 complex) describing this dyno. In suspension geometry, there is no debate as to whether transient response exists or not. The fact that the concept of recovery time from a load is applied to the behaviour to a motor is not difficult to understand, if you look at it that way.

As I said, the failure of the sophistry RR98ITR states is in the lack of data demonstrating the idea not whether it exists. And to me, using the road racing secrecy excuse is not an excuse at all. There must be data from obsolete race cars showing transient data.

I myself have tried to demonstrate this idea using data over at our website , team-integra.net in the Performance Articles, dyno section. It is mostly motorcycle data taken at a road racing circuit in the UK from Performance Bikes magazine. There is also an acceleration graph compared to the respective motorcycles' traditional dynojet dynos. Surprisingly the motorcycle not having the highest peak torque but rather the steepest rise in torque (with the proper gearing) is the highest accelerating motorcycle (same driver, tires, and a FD disadvantage...weight difference minimal).

http://www.team-integra.net/sections...?ArticleID=286

It was my attempt at getting a hold of data to validate the idea. It's not my data (ie. I did not manipulate it nor can I manipulate it) and these people are not linked to the sophists whatsoever (Idoubt the racing motorcycle industry in the UK even knows about the sophists). Whether it "proves" the existence of the idea is to be judged by the anti-sophists.

cheers





[Modified by Michael Delaney, 10:25 AM 1/28/2003]

RR98ITR

Tuan,

We've got torque and we've got rpm - we really don't need any more units.

Big racing teams like you've mentioned have programs for each track because they want to select the most optimal power curve for each track. It's that simple. Where are they spending the most time each lap, what part of the powerband is most critical, how much can they bias it without running into counterbalancing decreases elsewhere, etc.

Transients, as I'm criticizing it, is a bogus marketing shpeel.

Transients, as you cite it's use by bigtime race teams, is real and easy to understand.

And, as with so many scientific investigations, the smaller the difference you're trying to measure the more you will be frustrated by your apparatus. For what it's worth though small but expensive horsepower increases are a very poor value for self financed club racers. A far better value is realized by increasing the tire budget. It takes alot of power to go just a little bit faster, but come off the corner a little faster and you're doing good.

Scott, who having said that still want's more power, more tire, and more money to pay for it with.....that's racing.....

Mike D

Excellent thread. Havent seen one like this is quite some time.

Anyway, here is how i see it. "transient response" is simply a technique to make something non-magical, sound magical. The time needed for an engine to travel from x RPM to y RPM is directly related to the amount of torque available between x and y RPMs. Not doubting Mr Endyne's, Mr Hytech's, or anyone who has proposed this theorys knowledge, just strikes me as another way to explain a engine's torque curve. Simple physics really.

Again, excellent thread

Mike

hitechex

Okay Scott you bring some good points. But there is more to it. I will try and help you throught what we have found out. Lets take the Formula Ford engine. One, becasue I know a lot about it. We used to dyno the engines at a steady state rpm. Computers weren't in vogue yet so it was all done by hand and eye. We would start at 4,000 rpm and jump every 500 rpm's to 7,000. Then when the computer came around we did the magic sweep runs. The dyno would run from 4,000 to 7,000 and data would be samples over the run and averaged out into the print out. But why would the some engines perform better on the track than others? That was a very confusing. It is more apperant in road racing where you are continually on and off the throttle, part throttle, and braking and shifting. So we went back to the dyno to try and figure out what is going on. We built a water accelerometer attachment with the help of Larry to try and see if we could find anything there. What we found was that a certian rpm's the motor would stumble. So we went back to the dyno and ran spot checks at every 50 rpm's and what we found was that some engines had at least one spot in the usable rpm band that it would not run clean. But with the other sweep tests, the motor would run right past the problem spot. We finally traced it back to the cylinder head and how the head was ported was the difference. We can't fix all of the motors we find that has this problem but we can tell you that the ones we can fix do run better in the track.

You might be able to see if yout ITR has the same problems as our Fords do. I know my new RSX has a flat spot in the rpm band at about 4,800 rpm. If I run it up and hold it there at 4,800 it doesn't run very well. But if I accelerate up through the gears. It runs right past the 4,800 probelm like it wasn't there.

And as for the dyno it is only one samll part of the whole picture. You can not dynamically load the car like it is on the track. Aerodynamics and pitch and yaw(angle of attack) create forces that hold the car from accelerating. And if at certian tracks where the load happens to coincide with a part of the rpm band that has a slight probelm in it. You car will not accelerate as well if the problem wasn't there.

There are no easy and fast answers to those questions. If we all could explain it, then we all could build the best racing enignes out there. And we know that is not the case.

We also know that motors with carburetor's have a fair amount of reversion going on over the carb entrance. And that if we put a set of anti-reversion chambers on them. The reversion is reduced. And we pick up better throttle response in the lower rpm's, where the overlap in the camshaft has its worst effect.

Only continued test and R&D will eventually get us answers to those hard questions. We learn more everyday and get a little bit closer to the answering them.

B18CXr

My $150 HP Racing header works pretty good.

Got boost?

ciRcuitJerk

ya me too im confused...im gonna read this tonight ...very carefully

ENDYN

Unfortunately there are no units of measure for quantifying an engine's response time and transient capabilities, so individuals who've spent the time and money on dyno hardware and software for studying their programs are producing data that's only relavent to testing on their equipment.
The best anyone can do at this point is to "race" an engine combination on their dyno and "race" the engine after modifications to determine whether the elapsed times are quicker and there'll be an improvement in the real world.
I believe that there are a couple operations that are able to perform these tests on chassis-type dynos manufactured by Superflow, but most real development programs are still conducted using an engine dyno, as their sensitivity is much greater.

Big Phat R

This myopic view is suitable for people who go to the track very often or actually compete - but is inappropriate for people who mainly drive on the street (i.e. basically 85% of H-Ters).

But it is one of the most valid points in this rather long-winded discussion that hasn't really said much so far.

Coles Notes: More power = faster

ZygSpeed

Cliff Notes : motors running carbs may have dead spots. (no kidding)

Not many of us here in the ITR forum are running carbs.

Furthermore, more of us are switching to stand-alone EFI systems for their tunability, and most vehicle produced now a days come with some form of fuel injection!

Are you saying there would be no change in "transient response" in a fuel injected motor, and how does this relate to your header designs?

Ed -- an inquiring mind and former patent examiner in engine and vehicle control systems.

Benny Hinn

OK....

1GreyTeg

Thanks for this one.

It's good to see intelligent posting like this again that is above my head just enough that I have to go scrambling around and read up even more on some random topic to amass even more information so I can begin to totally understand it all and form my own opinion.


for making me think on my day off!


Nothing much to add but, that I'm along for the ride... Thanks again,,,


A.

Michael Delaney

so you say po-tay-toe and I say po-ta-ta {turn off Harry Connick Jr voice}

how about (in non-marketing terms) :

a broader peak torque and a steeper slope on the rising part of the torque curve coupled with the correct gear ratios and tire diameter combination that allows the post-upshift rpm to fall in that torque range will make a car accelerate faster through each gear.

is that more palatable?


On a dyno you load the engine in a steady manner from 4000-redline...it does not reflect sudden changes in load and therefore the behaviour of the torque curve under those conditions. How can we demonstate this?

If our budget as club racers cannot show these changes, does it mean that it does not exist? Is it really merely a marketing ploy if the "big" teams test to achieve the best torque curve (whatever best means) for the track. Drag racers have no concept of sacrificing higher power for better driveability to achieve a lower lap time...it's obvious by the cliff notes replies you see here.

Jack Black

Scott, I must say I'm a little disappointed. On the one hand, you advise against trying to over-simplify the realm of chassis tuning, but on the other hand, you're willing to sum up something as complex as an internal combustion engine with a simple torque/RPM curve.

If "transient response" is defined simply as the time it takes an engine to accelerate a given load from RPM A to RPM B, it can be very easily demonstrated that changes to the engine can change the transient response without changing the measured torque/RPM curve.

Take a given engine and measure its torque/RPM curve on an engine dynamometer. As previously noted, this is done by measuring the torque load required to hold the engine at a given RPM. This is done at certain points across an RPM range and then the final curve is fit through those points. That gives you the torque/RPM curve which tells you the whole story of the engine, right?

Now take the exact same engine and change to a lighter flywheel. The torque/RPM as measured on an engine dyno will be exactly the same as before, but you can be pretty sure that the "transient response" as defined above will now be better.

The chassis dynos that are usually used by little guys like us measure this kind of transient response to a certain extent, but they're still only tracing a single path through a multi-dimensional space. If you do your runs using a higher gear, the acceleration will be lower and the results closer to the steady-state curve one would measure on an engine dyno. If you do your runs in a lower gear, the acceleration will be higher and the transient effects more noticeable.

From what I've seen, though, most people tune their engines using just one relatively high gear. Thus the transient effects, if any, are both minimized and left unmeasured. By "unmeasured", I mean that they're in there, but you have no way of knowing how much they contribute to the outcome.

Now I'm not going to suggest that I'm any kind of expert on headers (what little knowledge of the subject I do possess is limited to a grasp of basic physics and a few articles aimed at lay people), but I do see the possibility of transient effects factoring into header design.

As I understand it, much of header design centers around manipulation of pressure waves, for example, the high pressure pulse that is created when the exhaust valve opens. When designing a header, the frequency and velocity of these waves are combined with the length, size, and shape of the tubes to maximize scavenging and flow across a range of RPM.

At a particular steady-state RPM, the pressure waves will be coming at a certain frequency, and will combine together to produce a certain result. On the other hand, if the engine is accelerating, the waves will be coming at different frequencies, with earlier ones coming farther apart and later ones coming closer together. This will result in an instantaneous combination that could be very different from the steady-state combination at that same RPM.

All that being said, you are quite right in noting that "the smaller the difference you're trying to measure the more you will be frustrated by your apparatus." I wouldn't be at all surprised to learn that the real-world effect of the transient effects in a header are negligible at a club racing level.

Big Phat R

Low blow.

Since when are "lap times" a part of this discussion? I thought engine power and transient response were being discusses ad nauseaum. Engine power is a small part of the equation that equates to a lap time.

Surely you should know this.

Michael Delaney

that's the whole point of the issue dude...that rule is not necessarily true if you do a run up the highway through the twisties to Whistler and time it. More power at the wrong part of the rpm range and the width of that power is as important as the level you achieve itself. A drag racing mentality does not apply to street driving if you do more than go in a straight line from stoplight to stoplight.

We are trying to discuss whether one word: "transients" describes driveability and acceleration response to changing loads. Maybe a better term is throttle response? I don't know. F1 and CART teams use the term "engine transients" so it isn't just a marketing term by the 3 groups Scott always wants to challenge.

if you want to name that entity something else other than "transients" so that you can dissociate yourself from the sophists and feel better then, be my guest.