The absurdity of fixed percentage drivetrain loss

TDIMeister

Phd of TDIClub Enthusiast, Moderator at Large
Joined
May 1, 1999
Location
Canada
TDI
TDI
The whole concept of a fixed percentage drivetrain loss in evaluating chassis dyno tests is one of the most absurd things i've ever heard of in my experience as a powertrain engineer, and is the motivation behind this post.

First, a primer: energy (or in the case relevant to this discussion -- power) cannot be created or destroyed. Simple enough? Where then, does that power that is transmitted from the crank, to the wheels, and ultimately to the road, go? Most of it goes to friction and therefore heat.

Second, an illustration: For argument's sake, take a stock 90HP TDI, and we'll arbitrarily say that it has a 15% driveline loss. That means that the engine would be developing about 104HP (90/1.15) at the crank. The loss through the drivetrain was 14HP (104-90). Now, you do a bunch of engine mods without touching the drivetrain, and you now measure, say, 135HP at the wheels. Adding the customary 15% to refer back to the crank, you get 155HP, but the loss through the drivetrain is now 20HP, a difference of 6HP, WHEN NOTHING HAS BEEN TOUCHED THERE!

Do you now see the absurdity of this concept?

Firstly, may I submit that 2WD vehicles with manual transmissions have very good mechanical efficiencies, as evidenced by the fact that 2 quarts of non-pressurized, non-circulating oil is sufficient to keep the entire transmission cool and lubricated. In fact, to attach a number to it, manual transmissions are usually over 90% efficient, and many over 95%. That implies a loss through the transmission of between 5.3-11%. Even the best automatic transmissions with lock-up TCs achieve between 80-85% efficiencies.

Secondly, may I submit that contrary to popular (mis)conception, flywheel weights, rim weights/diameters and tire type (should) have very little contribution to the HP numbers on a rolling road dyno. Heavy flywheels and rims act as inertial dampers but do not destroy or create energy, nor transform it to heat, as would have to happen to if it is to result in a greater or lesser HP value on the dyno. Tires will shed energy in the form of heat by the simple contact with the ground and also though the flexing of the treads and sidewalls, but this amount is negligable in the scheme of things that it is generally ignored unless you are an engineer for an OEM, race car team or tire manufacturer. More on inertia in a moment.

Thirdly, I hope the above underscores that an accurate measurement of drivetrain loss cannot be overgeneralized. For one, it is not constant across the entire measurement range within a given run. In fact, friction increases roughly linearly with speed. In automotive engineering speak, this is quantified by a parameter called the FMEP (friction mean effective pressure), and although it's is not called that, it is manifested in many engine graphs you may read without even realising it. Frictional losses are different at 2000RPM to 4000 RPM, etc., etc. You cannot, therefore, equate the drivetrain loss of a car whose engine is turning at 8000RPM at the maximum rated power to one turning at 4000RPM, because on the basis of the RPM alone, frictional losses at 8000RPM are roughly double that at 4000RPM.

That said, yes, it's true: gear selection when performing a rolling road dyno DOES have an impact on HP, but it is not usually borne in dyno results, because the difference is small and within the inevitable variation from test-to-test and also measurement error.

Further, engine/driveline design considerations mean that there is a wide variance in frictional losses between different cars; the comparison of mechanical efficiencies between manual- and automatic transmissions have already been discussed above. Cars with AWD, automatic trannies, and large-displacement/many-cylinder engines will tend to have higher frictional losses than small-displacement, 2WD, manuals.

Finally, The importance of "motored" or coast-down tests in a dyno evaluation is important and needs to be stresssed, because that is what accounts for your true frictional losses and balances the inertial "ledger sheet" of the different driveline components, including the wheels and tires. The energy that is absorbed in the form of inertia in the flywheel/wheels/tires, etc. is accounted for ("given back," to oversimplify) in this coast-down, and when doing a street (i.e. butt) dyno, also accounts for the very important aerodynamic drag.
 

Davin

Top Post Dawg
Joined
Jan 4, 2001
Location
L.A.
TDI
2001 Golf GLS 5spd blk/blk
Good post 'meister... but everyone here knows that engineers don't know anything. /images/graemlins/wink.gif
 

20IndigoBlue02

Top Post Dawg
Joined
Aug 22, 2001
Location
Was North NJ, now SoCal
TDI
2002 Golf TDI-- deceased
Keep in mind, while absurd, the fixed percentage is only an rough estimation, when using a chassis dyno

If you want the actual engine power/torque, you don't use a chassis dyno, you use an engine dyno (where it pumps water or some other incompressible medium).

How convenient is an engine dyno? Great when you have the resources, but to us, and the smaller tuning shops, not that convenient.

Given that we are always looking for the best deal possible, who would pay a few hundred dollars (or more) to take the engine, exhaust, & ECU out, attach it to an engine dyno, then reinstall everything?

Given the limitations of the real world....we'll have to survive on estimations.
 

peter pyce

Veteran Member
Joined
Nov 6, 2001
Dave, this is a typical great post from you, but I wonder why this info somehow never came out while a lot of dyno numbers had been discussed in the past? A lot of people use very different numbers, and somehow no one ever said a thing about being wrong.....

The question here would be - what do we do then when we go to the dyno? How can we get the most accurate numbers? How do we make sure that we all compare our cars equally? Could you please make a post (a guideline) of how a dyno should be specifically performed for a TDI and what are the different variables we have to be careful with. If such post had been made long time ago, please do point us to it. We all have to synchronize on this one, because a lot of people use the results as final decision maker for next step mods and the different percentage of mistakes we all make with our numbers may mislead the rest and create a base for misconnects of what works and what does not. Thank you very much!
 

TDIMeister

Phd of TDIClub Enthusiast, Moderator at Large
Joined
May 1, 1999
Location
Canada
TDI
TDI
[ QUOTE ]
Keep in mind, while absurd, the fixed percentage is only an rough estimation, when using a chassis dyno

...

Given the limitations of the real world....we'll have to survive on estimations.

[/ QUOTE ]

I agree, and my intent was not to say that it is a wrong practise out of hand, but, if it's to be used, consideration must be put into whether a value of, say, 15% is valid, or just a "figure that everyone else uses therefore so will I." Also, it is important that given that we drive roughly similar cars, a figure -- whatever it is arrived to be -- is used consistently so that people get apples-to-apples comparisons; as best as can be had given the rough estimation nature of this calculation.


[ QUOTE ]
Could you please make a post (a guideline) of how a dyno should be specifically performed for a TDI and what are the different variables we have to be careful with.

[/ QUOTE ]

I'll defer that to others who have more hands-on experience on this sort of thing than I. In the meantime, I'm digging dyno plots to see if I find the data I'm looking for. Stay tuned...
 

VelvetFoot

Top Post Dawg
Joined
May 17, 2001
Location
Sand Lake, NY
TDI
NB, 2000, Yellow
You'll be acquiring these numbers at the wheels, right, just like everybody else and that's what you will compare. Discussion of dyno characteristics (and calibrating?), methods, etc that seems more germane for comparison to others.
 

VWRacer1

Veteran Member
Joined
Mar 17, 2004
Location
SF Bay Area
[ QUOTE ]
For argument's sake, take a stock 90HP TDI, and we'll arbitrarily say that it has a 15% driveline loss. That means that the engine would be developing about 104HP (90/1.15) at the crank. The loss through the drivetrain was 14HP (104-90). Now, you do a bunch of engine mods without touching the drivetrain, and you now measure, say, 135HP at the wheels. Adding the customary 15% to refer back to the crank, you get 155HP, but the loss through the drivetrain is now 20HP, a difference of 6HP, WHEN NOTHING HAS BEEN TOUCHED THERE!

[/ QUOTE ]
I'm not an engineer (my degree is in math), but it appears to me that something has changed there which may account for the difference. The input load on the gearbox has gone up ~50%. Assuming no tire spin or clutch slip, does not this increased load add friction over and above speed-related friction? Perhaps enough to account for the 6 HP in your example?
 

RiceEater

Veteran Member
Joined
Aug 13, 2001
Location
96595
TDI
gray 2k2 Jetta GLS
I guess we should change Davin's post to mathematicians, chemists, scientists and engineers.

When I drive my tdi for extended periods I sometimes get out and rest my palm on the tire treads, sometimes even the transmission. There's maybe a tenth or two hp in BTUs transferred there (think LaPlace equation here). The correction factors that these snake oil peddlers and wrench turners are using is HUGE. I would guess our tdi 5 speed in fifth is like 99.5% efficient, probably 95% efficient in 1st, 2nd and reverse, figure a little less for the tires.
 

NRU73

Veteran Member
Joined
Mar 8, 2004
Location
N.E.
TDI
02 GOLF TDI Silver, 5 Speed
Meister, Great post! Though correct me if i'm wrong but a 90hp TDI is supposed to be the hp at the flywheel and it may produce a little bit more HP than expected.
 

20IndigoBlue02

Top Post Dawg
Joined
Aug 22, 2001
Location
Was North NJ, now SoCal
TDI
2002 Golf TDI-- deceased
Maybe for us (and other VAG products), maybe using the data logging feature of the VAG-COM could be a standard. Of course, one of the variables is weight.

If you want to do a find out the actual drivetrain loss for a particular car, the only way to do it is performing an engine dyno, then chassis dyno (same shop)

Of course, the more cars tested...the better.

But since we're limited...say a 5 car sample size, then compare it to a Weibull distribution. If it's good for Weibull, you take the mean and characteristic life (63.2%).

Yes, when you do Weibull, it is typically used for time to failure
 

GoFaster

Moderator at Large
Joined
Jun 16, 1999
Location
Brampton, Ontario, Canada
TDI
2006 Jetta TDI
Before this thread gets over-complexified, I'm gonna make one very simple statement:

When installed in a vehicle, engine horsepower does not matter to the end user. Wheel horsepower matters because that's what is propelling the vehicle.

It simply doesn't matter to the end user how much driveline loss there is. It's irrelevant. In the world of tuning motorcycles, which is the one I'm more familiar with, nobody pays attention to engine horsepower. Everything in the aftermarket is gauged by wheel horsepower as determined by a suitable chassis dynamometer.

If you make mods to the engine that result in 5 more wheel horsepower, that's great. If you make mods to the drivetrain that result in 5 more wheel horsepower (good luck, but the way!! very difficult), that's equally great. IT DOESN'T MATTER.

Sometimes it's nice to know how much power is being lost at different places, in order to pursue areas of improvement, and this is where it's kinda nice to dyno the engine out of the vehicle and then dyno the whole vehicle. But invariably, the engine installation at the engine dyno differs from the installation in the vehicle, and that muddles the situation. Result, you know something but not everything, and this approach isn't really practical for anyone short of professional race teams with lots of $ and time on hand.

Dave is correct that driveline loss is not really a "percentage". There is a component that's proportional to the torque transmitted (friction between the gears and in the bearings - but the bearings are all rolling-element and gear-sets are typically 98% - 99% efficient except maybe for hypoid rear-ends that are a bit less). If there are two gear-sets to go through and no 90-degree gearsets, which is typical of transverse front-drive gearboxes, then the mechanical efficiency of the gears not counting other types of losses can be in the 97% range.

There is a torque component of the losses that depends on the rotation speed (typically, churning and pumping losses of the lubricant - these losses increase with rotation speed but aren't affected by how much torque is being transmitted), and there is a torque component of the losses that is almost constant regardless of rotation speed or torque transmitted (typically, sliding friction in seals). And automatic transmissions have complex patterns of losses (typically much greater than manuals) due to the torque converter and the pump that operates all the hydraulics.

There are things that can be done about some of these losses (different lubricant viscosity, for example) but much of it is set by the design - nothing can be done about it other than different lubricant viscosity or oil level, and that's limited by what is specified by the manufacturer. Bottom line: Transmission losses are there, they vary and are not a set percentage, and little can be done about them, so the most suitable course of action is to simply not pay attention to it and only pay attention to wheel horsepower.

The bit about wheel and rim weight and other inertial loads deserves comment: how much difference this makes depends on what type of dynamometer is used. If a brake dyno is used, in which speed can be held constant while the torque is absorbed by a (typically) water brake, then Dave is correct, nothing you do about the inertia will make any difference at all. But if the most common aftermarket-type dyno is used (the Dynojet inertial dyno, or similar) then inertia of those parts CAN and WILL make a significant difference.

Finally, the most complex element of all: Like it or not, there are variations from one dyno installation to the next. Some of it is variance due to the dyno itself, but Dynojet claims that it's small. Cycle Canada did some testing a while back which found that perhaps the variation is greater than Dynojet admits to. For sure, there are variations due to the installation, and there are variations due to the way the testing is done, and there are variations due to temperature and barometric pressure.

If you are doing serious tuning, then comparison tests should always be done on the same dynamometer by the same operator using the same test procedure, and preferably under ambient conditions that are as close as possible. Comparisons between different dyno installations aren't really valid.
 

GeWilli

Top Post Dawg
Joined
Aug 6, 1999
Location
lost to new england
TDI
none in the fleet (99.5 Golf RIP, 96 B4V sold)
[ QUOTE ]
Comparisons between different dyno installations aren't really valid.

[/ QUOTE ]

Nice thread guys!

---

So Mr. engineer sirs:

Can you address the correction factors in relation to this? Or is that something completely separate?

And if I'm reading this correctly: A "Butt-Dyno" properly done on a flat level surface with an accurate coast down, sounds much more accurate for measuring real/true HP than these dynos... no?
 

Rammstein

Top Post Dawg
Joined
Sep 3, 2002
Location
Québec city baby, Canada
TDI
Golf 92 D
[ QUOTE ]
Second, an illustration: For argument's sake, take a stock 90HP TDI, and we'll arbitrarily say that it has a 15% driveline loss. That means that the engine would be developing about 104HP (90/1.15) at the crank. The loss through the drivetrain was 14HP (104-90).

[/ QUOTE ]

So, as I've been saying all along, German cars horse power and torque ratings are measured at the wheel and not at the crank like Asian (Corean and Japanese) or American cars.

Am I right or am I right? /images/graemlins/smile.gif

Therefore, I don't see any purpose knowing what's the crank value of hp and lb-ft of my TDI. /images/graemlins/cool.gif

I hope I don't sound too arrogant, it was not my intention. /images/graemlins/blush.gif

Very informative discussion btw guys. /images/graemlins/wink.gif
 

TDIMeister

Phd of TDIClub Enthusiast, Moderator at Large
Joined
May 1, 1999
Location
Canada
TDI
TDI
No, that's not what I'm saying at all. Horsepower measurements by ALL car manufacturers follow standardized test methods according to either the SAE, DIN or JIS (Japanese), depending on the country. These standards define test parameters, ambient conditions, etc., for engine dynamometers, NOT chassis.

However, car manufacturer's will tend to design the engines a bit on the generous side with respect to published horsepower ratings. The primary reason is that if a company claims 90HP, they better make damn sure there isn't even the possibility of one less or they will have some very irate customers (recent example is the Mazda RX-8). Manufacturing tolerances and other variabilities (fuel, ambient conditions, engine condition, state of tune, etc.) mean that, in a large sample population, there is a normal distribution in the power output of the engines, and OEMs would want to quote the lower spec limit rather than the mean. How this lower spec limit is defined by the OEM and the inherent variability of engine design, manufacturing and the control systems combined determines how power ratings quoted by the OEMs differ from what is actually measured in an independent dyno.

There are other reasons why this may be done as well (regulations, taxation and insurance ratings); it is documented that some cars are intentionally rated extremely conservatively for these reasons (e.g., unwritten 280HP limit for Japanese market cars among the Japanese automakers).

Summary: VW quotes 90HP in literature, but it's intended to be at the crank, even though rolling-road tests here have shown that completely stock A4s can make AT LEAST 90HP at the wheels. VW is just being generous /images/graemlins/wink.gif -- actually they're just hedging that the variabilities mentioned above won't result in an uproar if some engines are measured to be below the claimed output. Also, it has already been mentioned that 3rd party dyno tests are not done in particularly well-controlled conditions from a scientific standpoint (too many variables unaccounted for). GoFaster pointed out above that there can be significant variation from one unit to another of the same type, and even greater variation among the different brands out there (e.g. Dynojet, Mustang, etc.).

For A3/B4 TDIs that have dynoed far less WHP... hmmm I wonder why: How many miles are on those engines? /images/graemlins/rolleyes.gif
 

sciroccohp

Veteran Member
Joined
Oct 9, 2003
Location
New Haven CT
TDI
Jetta GLS TDI, black with grey leather
I have a dyno (brake dyno)though I have never put my TDI on it, I have done other TDI's. I have to disagree on the wheels and tires, I do not know the reason but when I put my MK1 scirocco race car (1.6 with about 100hp)on with my rain tires I get a couple more horse then with my race tires plus I get a differnce of 3-4 horsepower between 15psi and 25psi I don't know why like I said. The type of dyno makes a huge differnce on comparing them. A dynojet will usually give you about 15% more power for a less powerful car and the difference gets less the more the powerful the car is. I agree that it is only important if you really care about crank power but who does? I spend hours on the dyno for 2-3 whp. I have no idea what I have at the crank and don't really car.
 
Top