31 replies to this topic

[scooperman]

I searched this forum, and the web, and found little information on wet sump gearbox efficiency. If you can point me to some reference material, I would appreciate the help. I am interested in optimising a production-based gearbox, not a bespoke dry sump race car gearbox.

I did find a few online anecdotal references comparing dyno measurements, where the engine was dynoed and then the vehicle was dynoed after the engine was installed, so from this one can estimate losses in the gearbox and final drive. Straight cut gears are often quoted as 98% to 99% efficient, but then it becomes hard to find any data explaining the rest of the gear train losses. Inside the gearbox, there must be some considerable drag on the gears from the lubricant pool, I have seen it suggested that at low rpm this drag is proportional to velocity cubed, then as rpm increases the gear carves a groove in the surrounding lubricant until air is entrained, after which the windage drag should reduce, eventually becoming proportional to velocity. So if this is true, I question why we do not see gear scrapers in wet sump gearcases, similar to crank scrapers in wet sump engines. Certainly, the shift forks would act like scrapers, but there are few shift forks compared to the number of gears. Thanks for any insight.
John R.

[J. Edlund]

There are three types of losses in a gearbox:

Constant losses (pre-loaded bearings for instance)
Speed losses (windage, bearings for instance)
Load losses (mainly losses from gear to gear)

When we talk about the efficiency of gears it's typically a load loss we're talking about, and that efficiency is normally given per gear set. If in a gearbox the power is directed through three sets of gear with an efficiency of 99% the overall efficiency is about 97%. Increase the losses on average two percentage points more and you got an overall loss of 5% which is about what you see in a F1 transmission which use stright cut and polished gears. Obviously though, the overall efficiency will depend on load and speed so for the complete understandning of a gearbox one should do measurements and make a load and speed based map of the efficiency. There are dyno equipment to handle this sort of measurements. Often measurements are taken from a complete drivetrain where brakes and load cells are attached in place of wheels and one additional load cell is placed between the engine (or electric motor, as these dynos often use motors for simplicity) and gearbox.

If the gearbox also contain an oil pump, it will also contribute to the total losses from the gearbox.

[cheapracer]

There are most certainly dry sumped gearboxes in racing with spray jets direct to the gears brought about by needing a colling system firstly then just redirecting the return as the spray jets for direct lube and cooling.

[Joe Bosworth]

Like many queries on this forum we find a supposedly good thought or query lobbed in without stating where you are starting from or what the objective is. Most important questions to be answered if you want good answers.

If you are interested in improving on road/track performahis post abovence please keep in mind that performance costs money. The objective should be to utilise the most cost and time effective means first before going to the less cost/time effective.

As can be drawn from Edlund´s notes above, there is certainly to be less than 2% and most probably less than 1% to be gained from no end of gearbox development work. The reality is that it has almost entirely been done before you by people with a lot more smarts and equipment at there hands than you are likely to find on a forum. (I draw this from an old classmate who worked for Ford in transmission development.)

These levels of improvement can result in but small fractions of a second improvement in a typical lap time. I am certain that you can find similar or greater time improvements at less cost and infinitely less time than chasing the direction that you are thinking.

Unless of course you are already far down the paths of closing out every other performance option. In which case money is no problem so go out and buy the latest known state of art transmission from the recognised buildfers.

Or unless you are just after the intellectual exercise.

So, please provide more and better particulars.

Rrgards

[gruntguru]

I did find a few online anecdotal references comparing dyno measurements, where the engine was dynoed and then the vehicle was dynoed after the engine was installed, so from this one can estimate losses in the gearbox and final drive.

Chassis dyno power figures are significantly lower than for engine dynos and the difference is typically quoted as "driveline losses" when in fact the biggest factors in accounting for the difference are actually
1. Tyre slip and rolling resistance losses
2. Windage losses at the wheels and tyres

As detailed in above posts, gearbox and final drive losses are very low by comparison.

Edited by gruntguru, 29 May 2009 - 22:43.

[McGuire]

Since a crankshaft is oiled internally, scrapers make sense for oil management. However, in a splash-oiled gearbox a scraper would be counterproductive.

In NASCAR they have done considerable work on gearbox losses for the Daytona/Talladega restrictor plate applications, where every tiny bit helps. Precision assembly tolerances, low-viscosity lubricants, microfinishing, narrowed gears. IRL speedway boxes are also fairly well tricked-out in this area. But for road racing or other general applications the potential gains are not worth the cost and effort, just as Joe said. Not even close.

Practical approaches: have a look at the oil level, which is often determined by the filler plug location -- which is not optimized for low operating losses or even for any specific installation -- all applications may get the same housing. I have been down this road a few times, not looking for operating losses but to reduce foaming etc. But obviously the correct lubricant level will also reduce windage. And there have been advances in high-pressure lubes in recent times. You can certainly do better than the old 90EP in traditional OE gearboxes.

An old lesson repeatedly relearned: engine/bell housing/gearbox alignment. Amazing how often it can be off (or maybe not so amazing, considering the components all come from different sources). Or fine when assembled but wonky when screwed down in the chassis. Time consuming to measure but rather simple to check by taking your time with disassembly/assembly and paying careful attention. The parts will tell you how well they like each other if you watch and listen to them. Misalignment and flex are also the frequent cause of "mysterious" clutch and gearbox failures, excessive wear, shifting problems, etc. For example, the potential service life of a pilot bearing or clutch sleeve is effectively infinite. They should outlive the car. If they show assymmetrical wear pattern at all something is not straight.

In the conventional RWD package the gearset with the big drag is the ring and pinion. Hypoid gears run extreme angles by design, and everyone's favorite, the 9-inch Ford, runs an especially high hypoid angle. A Spicer for instance will produce a documented, measurable improvement in running losses. It's not big but it can be significant depending on the application. The Cup cars would surely be running something other than the Ford if NASCAR allowed it.

[phantom II]

Did you see Matchet's interview with Red Bull's Geoff Willis on Speed? They showed the insides of the 7 speed gearbox. 1st and second were not straight cut and were about an inch thick, while the rest were cross cut. All gears had different widths while 6 and 7 were the narrowest. Reverse and 6 and 7 had holes cross drilled into them for weight reduction. The length and diameter of the shafts effects efficiency because of bending.

I searched this forum, and the web, and found little information on wet sump gearbox efficiency.
Straight cut gears are often quoted as 98% to 99% efficient, but then it becomes hard to find any data explaining the rest of the gear train losses. Thanks for any insight.
John R.

[Greg Locock]

Chassis dyno power figures are significantly lower than for engine dynos and the difference is typically quoted as "driveline losses" when in fact the biggest factors in accounting for the difference are actually
1. Tyre slip and rolling resistance losses
2. Windage losses at the wheels and tyres

As detailed in above posts, gearbox and final drive losses are very low by comparison.

Cite? Final drive losses are rather large, that's why your diff gets so hot.

[gruntguru]

Cite? Final drive losses are rather large, that's why your diff gets so hot.

A 2kW electric heater gets pretty hot too.

Chassis dyno power results are typically 80% (ie 20% loss) of engine dyno results (depends a lot on gear selection - taller gear->higher wheel speed->bigger losses).
Gear losses on the other hand are typically 1% max for each spur gear pair and 5% max for hypoid.
FWD cars typically have spur gear final drives.

http://www.hbm.com/f...1-page15-20.pdf
p20 fig 10 hypoid gearset. 97+% typical


http://www.ohiolink....m=osu1128372109
pp 188-207 parallel axis gears. 99% typical
pp 208-219 Cross axis gears 97% typical

Greg. I'm not a gearbox specialist either but the references were pretty easy to find. No point citing the source in my original post. It was based mostly on gut feeling from numerous dyno experiences over the years.

Edited by gruntguru, 30 May 2009 - 08:22.

[Greg Locock]

Ah, so you didn't know. Diffs are often only 85% efficient, hence they can absorb more power than anything else in the flywheel to chassis dyno chain.

Edited by Greg Locock, 30 May 2009 - 10:47.

[gruntguru]

Ah, so you didn't know. Diffs are often only 85% efficient, hence they can absorb more power than anything else in the flywheel to chassis dyno chain.

I find that hard to believe. We are talking about full load aren't we? (as in full load dyno testing). Does a 195 kW FG Falcon lose 29 kW to the diff oil at max power? 5 minutes at full power on the dyno and you would need to add a heat exchanger.

If a typical RWD diff is 85% efficient, what is the efficiency for the final drive in a FWD transaxle? Are FWD cars significantly more fuel efficient than RWD's?

[McGuire]

On a chassis dyno with tire rollers, wheelspin is not an operating loss so much as a poorly controlled variable. You can tug down the straps and get little to no wheelspin, but then you have distorted the tire carcass, inceasing rolling resistance. Light on the straps and you reduce rolling friction but increase wheelspin. It is difficult to obtain the same strap tension every time even if you measure it, while the actual traction level is determined by power, tire size and type, etc and so forth.

This (and safety) is the advantage of the hub-mount chassis dyno. And then you have the inertia chassis dyno, which upon further examinatin is not really a dyno at all, but more of a goofy accelerometer thingie. Instead of placing an actual and known load on the engine, we make the wheels accelerate a great big drum and note the velocity rate change, calculating power and torque thusly. Huh.

[J. Edlund]

I find that hard to believe. We are talking about full load aren't we? (as in full load dyno testing). Does a 195 kW FG Falcon lose 29 kW to the diff oil at max power? 5 minutes at full power on the dyno and you would need to add a heat exchanger.

If a typical RWD diff is 85% efficient, what is the efficiency for the final drive in a FWD transaxle? Are FWD cars significantly more fuel efficient than RWD's?

Usually the total powerloss between engine and wheel hubs is smaller than 15%, even with four wheel drive. Nissan engineers did for instance quote a drivetrain loss of slightly above 10% for it's four wheel driven GT-R.

Rototest Research Institute found after measurements of more than 400 cars that the average power and torque discrepancy was 9% and 7% respectivly. That is, the power and torque measured on the wheel hubs is 9% and 7% lower than the figure quoted for the engine by the manufacturer.

[gruntguru]

And then you have the inertia chassis dyno, which upon further examinatin is not really a dyno at all, but more of a goofy accelerometer thingie. Instead of placing an actual and known load on the engine, we make the wheels accelerate a great big drum and note the velocity rate change, calculating power and torque thusly. Huh.

Inertia dynos have 2 big advantages (and no I am not a fan)
1. They can produce accelerating torque and power curves (although not much else) without the control-system sophistication necessary for a real dyno to do so.
2. They are really cheap (to make)

[gruntguru]

Usually the total powerloss between engine and wheel hubs is smaller than 15%, even with four wheel drive. Nissan engineers did for instance quote a drivetrain loss of slightly above 10% for it's four wheel driven GT-R.

Rototest Research Institute found after measurements of more than 400 cars that the average power and torque discrepancy was 9% and 7% respectivly. That is, the power and torque measured on the wheel hubs is 9% and 7% lower than the figure quoted for the engine by the manufacturer.

So with a typical chassis dyno power reading 15-25% lower than the engine - tyre slip (not wheelspin), windage and brake drag etc account for 6-16% of the losses.

The figures of 9% for power and 7% for torque are telling. The only reason they can be different is the different speeds at which peak power and torque occur. If the losses were a fixed percentage the power and torque numbers would suffer equally. Obviously a large proportion of the losses are speed related eg viscous drag and windage.

Edited by gruntguru, 31 May 2009 - 00:03.

[Greg Locock]

Well, given the complexity of the system I don't know why anyone would think that a single number would suffice. Your efficiencies are for gears with the correct geometry, the design of differentials is heavily compromised by the desire for low weight - as such the gears only mesh correctly at one torque for a given oil temperature. As such the efficiency vs torque curve starts off appalling, because of friction, then rises to a maximum, and then falls away. The location of the peak efficiency torque is up to the designer, it is an interesting tradeoff.

Perhaps you can see why your example is at once correct and misleading.

I haven't read

http://www.sae.org/t...rs/2009-01-2064
http://www.sae.org/t...rs/2002-01-1046
http://www.sae.org/t...l/papers/981190
http://www.sae.org/t...l/papers/800438

but they look relevant.

I think this is the one I was quoting

http://www.sae.org/t...rs/2002-01-2821

[gruntguru]

Well, given the complexity of the system I don't know why anyone would think that a single number would suffice. Your efficiencies are for gears with the correct geometry, the design of differentials is heavily compromised by the desire for low weight - as such the gears only mesh correctly at one torque for a given oil temperature. As such the efficiency vs torque curve starts off appalling, because of friction, then rises to a maximum, and then falls away. The location of the peak efficiency torque is up to the designer, it is an interesting tradeoff.

Perhaps you can see why your example is at once correct and misleading.

I think this is the one I was quoting

http://www.sae.org/t...rs/2002-01-2821

I don't have access to the full text from home but this paper seems focused on fuel economy and of course light throttle cruise areas that are most relevant to economy and which will see a large influence from seal friction, cold oil viscosity etc.

Dyno testing (and particularly comparison of engine/chassis figures) in the tuning and racing industries is focused on WOT numbers and all of my comments (and most of my experience) relate to that region.

Edited by gruntguru, 31 May 2009 - 00:32.

[McGuire]

Rototest Research Institute found after measurements of more than 400 cars that the average power and torque discrepancy was 9% and 7% respectivly. That is, the power and torque measured on the wheel hubs is 9% and 7% lower than the figure quoted for the engine by the manufacturer.

The Rototest paper does not support a conclusion that drivetrain losses are typically 9 percent power/7 percent torque. Here is why:

1. The study compares hub hp as measured on Rototest dynos installed throughout the field vs. flywheel hp as rated by the manufacturers of the various vehicles. So note, in none of these vehicles was the flywheel hp ever measured. This is important as we know that measured hp (flywheel or hub) may vary among individual examples of the same make and model due to manufacturing variances, break-in, state of tune, etc. Variations of at least several percent are commonplace. There are other variances, including those among mfg'ers; under-rating, over-rating, etc.

2. And so sure enough, when we examine the Rototest data we find it distributed in a classic bell curve, with some vehicles measuing more than 7/9 percent difference (as much as 18/19 percent) and even more interesting, a significant number of vehicles that produced more hp at the hubs than their mfg'er-rated flywheel hp.

Since we know that no vehicle can produce MORE output at the hubs that it does at the flywheel -- that is absurd, obviously -- we have a problem. This is a statistical study of statistical interest (about mfg'er hp ratings, perhaps) but it does not represent an actual comparison of flywheel hp vs. hub hp in any particular vehicles that ever existed. So from this data we cannot assert that typical drivetrain losses are 7/9 percent. This data doesn't really indicate that and we do not know that.

... as a practical matter, we know there will be a relatively wide range in drivetrain losses among vehicles. For example, a RWD luxury sedan with a torque converter and hypoid rear axle will exhibit significantly greater losses than a FWD compact with a manual transmission. To represent drivetrain losses we are pretty much back to out oft-quoted rule of thumb: in the neighborhood of the ballpark of 7 to 18 percent-ish. To get closer than that we need to test the specific vehicle.

The Rototest paper:
http://www.rri.se/index.php?DN=30

[cheapracer]

.

Since we know that no vehicle can produce MORE output at the hubs that it does at the flywheel -- that is absurd, obviously --

Quite obviously you have never been to an Australian speedshop

[Tony Matthews]

Quite obviously you have never been to an Australian speedshop

...or heard of go-faster stripes...

[gruntguru]

1. The study compares hub hp as measured on Rototest dynos installed throughout the field vs. flywheel hp as rated by the manufacturers of the various vehicles. So note, in none of these vehicles was the flywheel hp ever measured. This is important as we know that measured hp (flywheel or hub) may vary among individual examples of the same make and model due to manufacturing variances, break-in, state of tune, etc. Variations of at least several percent are commonplace. There are other variances, including those among mfg'ers; under-rating, over-rating, etc.

Since we know that no vehicle can produce MORE output at the hubs that it does at the flywheel -- that is absurd, obviously -- we have a problem.

Perhaps it also indiates that Rototest have a fudge factor built in, to make the numbers a little closer to flywheel numbers? (A La dynojet)

In the majority of cases, the actual flywheel power will be less than the manufacturer's spec' which makes the Rototest results sound even more optimistic.

[OfficeLinebacker]

Did you see Matchet's interview with Red Bull's Geoff Willis on Speed? They showed the insides of the 7 speed gearbox. 1st and second were not straight cut and were about an inch thick, while the rest were cross cut. All gears had different widths while 6 and 7 were the narrowest. Reverse and 6 and 7 had holes cross drilled into them for weight reduction. The length and diameter of the shafts effects efficiency because of bending.

Yeah that was cool. Only I thought only the reverse had been hollowed out for weight reduction.

[Greg Locock]

FYI

eyeballed efficiency for a final drive with a rated output of 2000 Nm (ie first gear max torque, plus a safety factor)

2000 rpm (pinion speed) - there is a speed dependency but it is not very significant

Output torque Nm eff %
110 75
170 83
230 87
340 92
450 95
570 97
from there on up 97-98%

So I'd agree that so long as you are in a reasonably high gear you'd be OK, but if you were in 4th and looking much below 50% peak engine torque things would be a bit hairy.

For extra points find similar data for the gearbox. and the tires on rollers.

[gruntguru]

FYI

eyeballed efficiency for a final drive with a rated output of 2000 Nm (ie first gear max torque, plus a safety factor)

2000 rpm (pinion speed) - there is a speed dependency but it is not very significant

Output torque Nm eff %
110 75
170 83
230 87
340 92
450 95
570 97
from there on up 97-98%

So I'd agree that so long as you are in a reasonably high gear you'd be OK, but if you were in 4th and looking much below 50% peak engine torque things would be a bit hairy.

For extra points find similar data for the gearbox. and the tires on rollers.

Something strange in those figues Greg. The losses (absolute not %) decrease as the torque increases. From the top, I get a torque loss (c.f. 100% efficiency) of

36.7
34.8
34.4
29.6
23.7
17.6
????

Edited by gruntguru, 01 June 2009 - 07:03.

[McGuire]

So with a typical chassis dyno power reading 15-25% lower than the engine - tyre slip (not wheelspin), windage and brake drag etc account for 6-16% of the losses.

The figures of 9% for power and 7% for torque are telling. The only reason they can be different is the different speeds at which peak power and torque occur. If the losses were a fixed percentage the power and torque numbers would suffer equally. Obviously a large proportion of the losses are speed related eg viscous drag and windage.

In the case of the Rototest dyno I think we can discount wheelspin/slip/whatever and windage: it's hub mount.

I can think of another reason for the discrepancy in torque vs. hp losses: the dyno is off. Perhaps it produces proportionally different values than the engine dynos that produced the mfg'er ratings.

5 minutes at full power on the dyno and you would need to add a heat exchanger.

It wouldn't be a bad idea. Five minutes at full gorilla on a chassis dyno is a long time. Hypoid axles generally require a cooler for prolonged high speed use... once again we refer to the tiresome NASCAR example. An increasing number of production cars run axle coolers as well.

One thing that makes the 9-inch Ford axle attractive for racing is its well-supported pinion shaft... which was in turn required by its extreme hypoid offset, 2.250 inches on a 9-inch ring gear. (The large ring gear diameter being the other attrraction, of course.) But there is a measurable loss versus other hypoid diffs, two to four percent. The largest Spicer-style diff you can cram under a passenger car (Dana 60) runs a 9.75 inch ring gear but with only 1.125 inches of hypoid offset -- it's essentially a truck piece.

Another trait of hypoid axles: a critical speed for gear whine... two actually, one on drive and another on coast, with the speed dependent on the mesh patterns. With a hypoid gearset, tooth mesh is critical in two planes, backlash and pinion depth; the other critical factors are pinion and carrier bearing preload. So you will find considerable variations in efficiency and losses in installation due to assembly tolerances, which also vary with speed. There is thrust in three planes and the pinion gear is basically trying to distort the ring gear into a wave washer shape, while the ring gear is trying to spit the pinion out of the case. In practice a hypoid gearset will generally display three (depending on ratio) tight spots in one rotation of the ring gear. Hyperboloid gear mesh is a sort of unnatural act.

[Joe Bosworth]

This whole branch into chassis dynos is very much a red herring to the main point of the thread which is transmission/gearing efficiency. For the very reasons that have been started to be alluded to it is completely fruitless to try to measure gearing efficiency through a chassis dyno.

I have only ever seen one set up for t4sting transmission efficiency and that was a variable speed electric motor with suitable power input instrumentation. That ran the transmission with outputs measured by a conventional brake type dyno.

The whole thing was calibrated with a spool piece in place of the transmission so that the input data and output data could be agreed for 100% efficiency. When it came to test a transmission the spool piece was replaced with the transmission at question.

They found that such a set up was the only way to get efficiency readings that meant anything for development and testing purposes where one was trying to see if tiny discrteet changes made any difference or not.

For instance, changing oil levels or lubricant types making changes in the order of 0.1% at a time could then be measured to see if one was starting to go in the correct direction. The matter of teeth form is a whole major deal in itself.

The things to be played with are voluminous but I believe that there is not much more to be tried that wasn´t tried years ago. The real problem is in accessing the old test data. No one talks much about the development directions that were never successful. So one needs to go down the failed paths time after time with each new generation believing they are in discovery mode.

Changes that intuitevly would be thought to be better sometimes aren´t and fine precision is needed for rational development. The kind of precision that is many orders of precision beyond the variabilities of a chassis dyno.

It may be that the vagaries and usage of chassis dynos are worthy of there own thread.

Regards

[Joe Bosworth]

Even writing the above reminds me of the costs of stting up a good test rig.

And time then there is the matter of time burnt playing with things like splash vs spray vs mist lubrication.

I go back to where I was with my first post on this subject. Don´t even start down this path of transmission development until every other performance option has been investigated and then only start on such development if you are either a real boffin or have access paying those who are.

Regards again

[gruntguru]

So with a typical chassis dyno power reading 15-25% lower than the engine - tyre slip (not wheelspin), windage and brake drag etc account for 6-16% of the losses.

The figures of 9% for power and 7% for torque are telling. The only reason they can be different is the different speeds at which peak power and torque occur. If the losses were a fixed percentage the power and torque numbers would suffer equally. Obviously a large proportion of the losses are speed related eg viscous drag and windage.

In the case of the Rototest dyno I think we can discount wheelspin/slip/whatever and windage: it's hub mount.

Er that was my point actually - 15-25% minus 9% = 6-16%

I can think of another reason for the discrepancy in torque vs. hp losses: the dyno is off. Perhaps it produces proportionally different values than the engine dynos that produced the mfg'er ratings.

Perhaps it also indiates that Rototest have a fudge factor built in, to make the numbers a little closer to flywheel numbers? (A La dynojet)

In the majority of cases, the actual flywheel power will be less than the manufacturer's spec' which makes the Rototest results sound even more optimistic.

Edited by gruntguru, 02 June 2009 - 11:58.

[McGuire]

A fudge factor would be easy to spot when calibrating with the free weight.

Also, it would be pretty stupid of Rototest to incorporate such a fudge factor and then do the white paper study.

[gruntguru]

A fudge factor would be easy to spot when calibrating with the free weight.

Also, it would be pretty stupid of Rototest to incorporate such a fudge factor and then do the white paper study.

Agreed.

[J. Edlund]

The Rototest paper does not support a conclusion that drivetrain losses are typically 9 percent power/7 percent torque. Here is why:

1. The study compares hub hp as measured on Rototest dynos installed throughout the field vs. flywheel hp as rated by the manufacturers of the various vehicles. So note, in none of these vehicles was the flywheel hp ever measured. This is important as we know that measured hp (flywheel or hub) may vary among individual examples of the same make and model due to manufacturing variances, break-in, state of tune, etc. Variations of at least several percent are commonplace. There are other variances, including those among mfg'ers; under-rating, over-rating, etc.

2. And so sure enough, when we examine the Rototest data we find it distributed in a classic bell curve, with some vehicles measuing more than 7/9 percent difference (as much as 18/19 percent) and even more interesting, a significant number of vehicles that produced more hp at the hubs than their mfg'er-rated flywheel hp.

Since we know that no vehicle can produce MORE output at the hubs that it does at the flywheel -- that is absurd, obviously -- we have a problem. This is a statistical study of statistical interest (about mfg'er hp ratings, perhaps) but it does not represent an actual comparison of flywheel hp vs. hub hp in any particular vehicles that ever existed. So from this data we cannot assert that typical drivetrain losses are 7/9 percent. This data doesn't really indicate that and we do not know that.

... as a practical matter, we know there will be a relatively wide range in drivetrain losses among vehicles. For example, a RWD luxury sedan with a torque converter and hypoid rear axle will exhibit significantly greater losses than a FWD compact with a manual transmission. To represent drivetrain losses we are pretty much back to out oft-quoted rule of thumb: in the neighborhood of the ballpark of 7 to 18 percent-ish. To get closer than that we need to test the specific vehicle.

The Rototest paper:
http://www.rri.se/index.php?DN=30

Treated as individuals some of the tested cars produce more power at the flywheel than the manufacturers specify, but some also produce less. Treated as a population instead it's however reasonable to assume that the rated power output is correct within a few percent (while rated power of some individuals probably can be up to 20% off). Then we should get some idea of the average powerloss in a transmission. But this is if course the average loss in a population of cars, most of which probably have manual tranmissions and FWD.

Perhaps it also indiates that Rototest have a fudge factor built in, to make the numbers a little closer to flywheel numbers? (A La dynojet)

In the majority of cases, the actual flywheel power will be less than the manufacturer's spec' which makes the Rototest results sound even more optimistic.

No built in fudge factor. The readings are very accurate given that the equipment have been calibrated correctly. The measured value should be correct within +-1%. The Rototest equipment also uses load cells between the input shafts and the hydraulic brake, so the measurement is not affected by the condition of the hydraulic system unlike some dyno equipment which measure torque indirect through pressure.

Some car manufacturers have however complained about inadequate cooling for the engines during a dyno run, which can hurt the performance of specifically turbocharged engines, but that was probably only a problem early on.

I don't think most manufacturers rate their engines higher than the actual power anymore, possebly even underrate them. To be forced to take back a car because it doesn't have the advertised power output ain't fun. Some manufacturers also advertise the same power rating in for instance USA as they do in several countries in Europe even though the US uses imperial horsepowers while Europe mostly uses metric horsepowers.

[gruntguru]

But this is if course the average loss in a population of cars, most of which probably have manual tranmissions and FWD.

Testing was done in Europe then . . .

Edited by gruntguru, 04 June 2009 - 03:23.