Measuring Engine Power

[murph7355]

We would bother but we might focus on trying to minimise losses if they were conclusively of this order?

 

What I meant was, loads of different cars' figures with est. power and est. losses. So we could see whether 15%-20% is sensible or not.

 

Robert's isn't, yours is. I'd say Robert's looks a bit odd from the gut feel approach.

 

[Old Septimus]

Haven't yet got to the library to borrow the book but ( at the risk of stating what may be obvious to other people)is the main point that the "losses" seem to occur mainly in the tyres? There must be mechanical losses in the drivetrain because the oil in the diff and gearbox gets hot and the (?first) law of thermodynamics implies that this is work. So is it the hysteresis in the tyres which causes these losses when running at speed on rollers set close together? That loss must depend on all sorts of factors, like pressures, widths , load etc. This must make the whole thing inexact as an absolute measure but presumably it remains a useful comparative measure if all these factors are kept constant for different roller sessions.

[Roland]

Simon,

Thank you. I did not realise the car was worked on at that time. (only measured).

 

Thx, Roland

[R2D2]

The discussion about dynamometers always seems to generate quite a lot of argument and disagreement.

 

Obviously there is quite a lot of difference between chassis dyno's and engine dyno's and some of the details are quite important.

 

Car engine manufacturers obviously spend millions of pounds on their engine test cells and the results they obtain must be both accurate and repeatable because of the effect on sales, legislation etc., etc.

 

The after market has a completely different set of values.

 

The typical chassis dyno that I have seen not only has all of the problems of tyre contact etc. but usually has a fairly crappy torque measuring system.

 

The power absorber is usually a trunnion mounted truck retarder and reaction torque is measured using a strain gauge mounted on a short reaction beam.

 

The L : D ratio of the beam probably results in a great deal of non-linearity and I doubt if the strain gauge system has good enough temperature and modulus compensation. The condition of the trunnion mounts would also have a significant effect on repeatability.

 

I have never seen how easy is to calibrate this type of dyno and I think that it is usually relies on a shunt calibration method that is completely unreliable.

 

I believe that chassis dyno's should only be used for "same day" comparative measurement. I don't think there is any basis for comparing results from one machine to another.

 

It would be very interesting to take a "golden car" of known performance around a number of different chassis dyno's and look at the differences and then check to see the effect of different operators had on the same machine.

 

In general I guess we would all be attached to the chassis dyno that gave us the highest power reading.

 

Engine dyno's are quite different and there are a number of different ways of absorbing power. Water brakes and eddy current brakes are probably still the most common. (The more advanced engine dyno's now have inverter driven AC machines that can both absorb and motor but they are vastly expensive)

 

Most of these dyno's are still mounted in trunnions and measure reaction torque but they commonly use a fairly high quality load cell and can be calibrated quite easily.

 

I would believe that engine dyno results are much more likely to be comparable between machines and much more reliable providing that the dyno has been reasonably well maintained and calibrated.

 

 

Companies like Froude and Schenck often quote accuracies of 0.25% for engine dyno's but I would tend to think 0.5% is more likely even for a very well maintained system.

 

The power figures often quoted from engine dyno's are not always the complete picture as they often don't include the water pump, alternator, and other engine driven ancillaries.

 

Transmission power losses are also interesting and I think that must be non-linear as there must be some proportion of the loss that is fixed depending on the size of transmission, gear tooth profile, finish etc. and some proportion of the loss that is a function of the power being transmitted.

 

 

Edited by - chris flavell on 26 Apr 2002 13:59:52

[ashaughnessy]

There's another thread in ChitChat about the Swindon 235 Vauxhall engine. I have heard it said on this very site that Swindon rated this engine at 235BHP when measured on their engine dyno. However, people say that it doesn't make nearly that much in it's Caterham installation. I have heard this is because of the exhaust pipe, possibly among other things. The Caterham installation uses a less efficient exhaust design that loses a lot of power. So I've heard, on old threads on blatchat.

 

The moral being that performance measured on an engine dyno can also be variable due to things like the exhaust system used. So when the engine is installed in the car, the engine doesn't produce the same power.

 

Ram air systems are also presumably tricky, because there is no ram effect when you're testing.

[Tony Martyr]

Nice summary from Chris. Just one comment about quoted accuracies of dynamometers.

Schenck and Froude and AVL quote dyno unit accuracy as plus or minus 0.25 percent of Full Scale Deflection, not a percentage of any measurement. This is important if you are testing a small engine output on a large capacity dyno.

Somebody asked if anyone has a 'shedload' of data to allow correlation between engine and chassis dynos. The answer is Yes and we can correlate quite well the exhaust emission drive cycles run on a chassis dyno with those of VCOT tests (Vehicle Calibration On Testbed) run on the most modern and expensive engine test cells. All this data confirms that guestimating engine power from a poorly calibratable rolling road used by the after market is self deluding.

Tony Martyr (now General Manager Scheck Test Automation, ex-Ricardo and ex-Technical Director Froude Consine)

[Roger King]

I would second the contributions from Chris Flavell and Tony Martyr, because they come from people with vastly more knowledge on the subject than I have.

 

I would never regard the figures that we obtain on the rolling road as being in any way 'absolute'. The rolling road is a useful tool for setting an engine up in the chassis and if you are familiar with the particular road and type of vehicle being tested it will give you a rough comparison - no more.

 

I know of at least one rolling road in my vicinity that will show about another 30bhp compared to the one I use if testing a 150bhp engine. Once, in the past, this rolling road showed more at the wheels than we had seen at the flywheel on the engine dyno!

 

If you stop for a cup of tea on the rollers and then do another power run on the same settings, you will often see an improvement of 5bhp. This is simply due to temperatures cooling off. Yes, you can put in correction factors for this but in most cases it isn't practical to do this for the exact moment of taking the readings. In other words, rolling roads are not really repeatable with any reliability, so don't get too excited by a gain of 5bhp.

 

Good engine dynos are a different kettle of fish entirely and are essential if you are doing development work and looking for incremental improvements from small changes.

 

On the subject of exhaust systems being different in the car, we did some development at Brian Hart Ltd in the 1970s on the works Escort rally exhausts. On the dyno we used a 4 into 1 race system for convenience, but when we fitted the system from the rally car we lost about 40bhp. Two days of work, which was basically suck it and see, saw the loss drop to around 5bhp. The moral of this is obvious.

[ian crocker]

Tony and Chris, what is your opinion of the Dynojet rolling road ? This uses a huge roller so there is only one contact patch with minimal distortion. In addition it uses the moment of inertia and rate of acceleration of the huge roller to calculate the power. I understand there is only 1 of these (maybe 2) in the UK but they seem to be rated highly in the US.

 

Ian

[Dazza]

Many thanks to Andy for kicking off the replies and all the boys for their sage comments. I will be keen to locate and read up on the previous threads on this subject.

My interest in this subject is because I am lucky enough to own a 1994 JPE (the only one that made its way here to Australia), and I have just had it properly tuned on a Dyno Dynamics Dynamometer for the first time. A rolling road type where the back wheels are spinning on drum rollers. All your comments have helped me put this interesting process into some perspective.

Now we all know that the JPE has the 250bhp Swindon version of the VX engine, and the original owners paid quite a premium for the privilege. However, the power figure quoted on this rolling road dyno (after tune up) was 120kW (sorry, we are metric down here, but I think that makes it around 160bhp at the back wheels). Obviously this is a long way from the 250bhp. However, the important thing is that this tune up gave me a before and after difference of 30kW! The vehicle was so fast (compared to my entire automotive experience) that I did not realise how out of tune it was. The main point is that this tune up has made a huge improvement over what was already a mind-boggling driving exerience. The best A$300 I have yet spent on the Caterham.

 

 

[Tony Martyr]

Ian, Chassis Dyno roll diameter is just one of many variables in the measurement of power and the simulation of on road conditions. It is an important one since a 2 metre roll gets nearer to a real life tyre loading than a 400mm or smaller roll.

You have to be very careful of tyres on a chassis dyno since they can get very hot and suffer side wall damage from side sway restraints and other causes. A large roll allows more heat to be dissipated and a more realistic tyre energy loss.

My advice to anybody having a car tested on any rolling road would be to ask to see the latest calibration certificate. So many units have never been checked since installation.

[edmandsd]

I don't think we should simply disregard wheel horsepower figures as these are values that a lot of rolling roads provide (And a lot of people use rolling roads to tune their cars as opposed to an engine dyno) - I believe the real confusion comes when you convert these measured wheel values to flywheel values, and I for one simply don't believe the losses that some people quote. Whatever anyone says and regardless of the tyre width/compound and number of people sitting on the back, a Caterham with 160 bhp at the wheels quite simply hasn't got 250 bhp at the flywheel !

 

Some of these drag strip desktop dynos are an alternative method of determining power production and probably not far off the mark either.

 

Re the 235bhp Swindon vx - If an engine has different dimensions to it's inlet/exhaust manifolds it quite simply isn't the same engine and shouldn't be referred to as such (These parts are not ancillary to an engine but integral, and certainly major contributors in determining output). What intigues me is why would Caterham commission Swindon to build a 235bhp engine for their cars that is obviously intended for a different installation ?

 

 

[oilyhands]

Dave,

 

RR at the rollers figures can be used for comparative purposes when tuning the same vehicle with the same tyres at the same pressure and roller speed. Otherwise they are missleading and unreliable even from the same RR, thats why the coastdown test is used to go someway to estimating the losses based on the rolling resustance of the drivetrain/tyres.

 

To win an 'at the wheels' power contest, run narrow tyres inflated to 60PSI, and run in the lowest possible gear that will give you traction, this will reduce the roller losses dramatically and give you a higher 'at the wheels' figure, but does it make the engine more powerful?.

 

I have run my car on one particular set of rollers many, many times, at various roller speeds with varying weight over the rear axle, the at the wheels figures varied tremendously from run to run when they were taken in different gears, only when the measured losses were added back to give a flywheel figure did the consistency appear. Comparing two particular runs, one in 5th and the other in 3rd, the difference in the at the wheels figures was 22BHP (45 losses vs 33), *but* the difference between the power curves when the coastdown test losses were added back was less than 2BHP.

 

Oily

[edmandsd]

Fair enough Oily although I'm only losing circa 40hp with 260hp at the wheels on Sanspeeds rollers. This was in 4th gear (would have been in 5th but I haven't got one) at 130mph (9,000 rpm) with 7 inch wide ACB10's and 6 people on the car to stop wheelspin.

HTR's 200 wheel hp Pintos are only showing 30hp power losses on the same rollers which reflects consistent losses as a percentage i.e. circa 15%

 

 

 

 

Edited by - edmandsd on 29 Apr 2002 11:45:31

[Peter Carmichael]

Please Dave, can you see sense and drop this 15% nonsense. It just doesn't happen like that. The losses are largely related to roller speed. Lots of people have told you this. Choice of gear can throw the losses out by more than you will ever gain or lose in a tuning session.

[edmandsd]

All i'm saying Peter is that with all HT Racing rear wheel drive cars rolled up in 4th gear on the same rolling road, with average size road/race tyres and people on the back to hold the cars down - They're seeing losses that equate to an average of 15% of the wheel power figures recorded. These are the results of rolling up 100's of cars over the years on this particular rolling road be it accurate or inaccurate.

[EFA]

Dave,

 

CAterham did not commission Swindon to build the 235, they adapted Swindons 235 spec for fit the Seven. Hence the changes and subsequenmt power loss.

 

Same Dazza for your JPE. Sadly its pretty well known that the JPE didn't give much if any over 220bhp.

 

My 235 gave 207bhp after tuning , 196bhp out of the box. Dissapointing.

 

 

Fat Arn

Visit the K2 RUM siteid=red>

See the Lotus Seven Club 4 Counties Area Website hereid=green>

 

[edmandsd]

I never knew that Arnie - only 220bhp for a JPE ! Surely the record breaking 0-100-0 car was breathed upon to give those results ?

[Peter Carmichael]

A little knowledge...is a misleading thing?

 

What relevance is an average if the deviation is high?

 

You are misleading people with a bit of flawed logic:

 

"I measure things with my 12 inch ruler and on average the things I measure are 9 inches long. When things are much smaller I use my 6 inch ruler. When things are much longer I use a tape measure. Everything I measure with my 12 inch ruler is therefore 9 inches long (on average)."

 

Edited by - Peter Carmichael on 29 Apr 2002 13:51:44

[EFA]

Dave,

 

Not that I know of. Fixing the problem would require some pretty visible changes.

 

Fat Arn

Visit the K2 RUM siteid=red>

See the Lotus Seven Club 4 Counties Area Website hereid=green>

 

[edmandsd]

My point is simply that a car with 200bhp at the wheels on Sanspeeds rolling road in 4th gear will show between 230bhp and 240bhp at the flywheel depending on the rolling circumference, type of tyres and number of people on the back. From my experience losses quoted over and above this are simply unjustifiable.

[neilsjuke]

My Webb rollers have got 5hp and if you let the clutch out to quick you can get spin on grass