Power loss over age.

[Pumaracing]

[quote=Mr Whippy]
As much as I generally agree with Pumaracing's figures, they are only a guide on a standard car. Take a 200bhp Golf V GTi with 197 bhp and you get ~ 30bhp losses. Tune it to be 400bhp, while leaving all else the same, and it doesn't now have 50bhp loses...
Davequote]

Yes it will which is the whole point of my transmission loss equations. Most of them are proportional to bhp fed into the system.

[Mr Whippy]

Pumaracing said:

Mr Whippy said:

As much as I generally agree with Pumaracing's figures, they are only a guide on a standard car. Take a 200bhp Golf V GTi with 197 bhp and you get ~ 30bhp losses. Tune it to be 400bhp, while leaving all else the same, and it doesn't now have 50bhp loses...
Davequote]

Yes it will which is the whole point of my transmission loss equations. Most of them are proportional to bhp fed into the system.

I don't get why a bearing will generate more friction spinning at the same speed... surely we can express a drivetrain as a series of masses, inertias and rotational velocities, each with a coefficient of friction (simplified I know)

All it should care about is speed surely? friction force = coeff * velocity (simplified)

I'm not sure how losses can be apportioned to the engine speed either, the drivetrain speed surely is what is important?

I guess it makes some sense that we could generalise to % based on rpm though, if it is true that there is a constant friction * torque loading included in the sum of total friction force (ie, independent of speed), because low gears means high torque, so higher torque = higher losses, and then in higher gears lower torque, so lower losses, proportional to the gears, and so rpm...

Hmmm...

So what is the best way to express the friction loss in say, a wheel bearing?

friction force (Nm) = (torque load * coeff_1) + (rpm * coeff_2)

I just don't get why something generates more friction because more torque is passing through it. Ie, a propshaft support bearing only see's velocity and the weight of the propshaft pushing down... it simply doesn't even care if it is idling or accelerating hard at all does it? How does it 'see' the load and make itself hotter?


Hmmm

Dave

Edited by Mr Whippy on Friday 24th October 19:10

[Mr Whippy]

Pumaracing said:

I'll need your gear ratios and FD ratio plus a power curve of any sort just to give me rpm at which things happen. I'll then run it through my performance simulator.

Dave Baker

3rd gear run... truncated after 3000rpm a bit because I forgot to adjust a limiter, and haven't done any power runs since (this was when I did my 1/4m too, so maybe a tad slow)



Ratios are, errmm, about, not 100% sure, but they get the right mph/1000 in a simulator...

3.455
1.870
1.148
0.822
0.66

FD 3.784


I'll have to throw the numbers into a sim/game thing for cars I play about with. It's pretty good, but I could put my wheels calculated curve in there, remove driveline friction, and see... I did do a test with a 150bhp curve, and simulated columb friction, and plotted the acceleration curves over each other (g-tech and game) and they were give or take perfect... downside is that I guesstimated columb friction... actually KNOWING what losses are, like you say, is bloody hard work

Thanks

Dave

Edited by Mr Whippy on Friday 24th October 19:22

[Huff]

I just don't get why something generates more friction because more torque is passing through it. Ie, a propshaft support bearing only see's velocity and the weight of the propshaft pushing down... it simply doesn't even care if it is idling or accelerating hard at all does it? How does it 'see' the load and make itself hotter?

Gears are rotating levers; the loads at the meshing teeth are reacted against the shaft bearings. So frictional losses due to both components (lsiding engagement at the tooth, and at the bearings) will be proportional to the torque transmitted. Since at a given rpm power is directly proportional to torque - if you increase the power transmitted you increase losses pro rata.

[Pumaracing]

I get an exact match to your 1/4 mile performance at every measurement point with a wheel bhp of 123.8 BUT ideally NOT starting off in 1st gear. The computer decided it was better to start off in 2nd which also saves one gearchange. All 1st gear does is spin the tyres pointlessly and you have to change into second before you even hit the 60 ft mark which mucks up that data point too. If I force it to start in 1st it changes the time at each measurement point by maybe 0.1 seconds. Nothing huge but the fit is slightly better starting in 2nd.

Flywheel figure is 148.7 bhp.

Top speed is rev limited to about 111 mph. It actually has the power to do 133 mph but the gearing is too low for that. Optimum gearchange points are

1st to 2nd 4200 rpm
2nd to 3rd 4000
3rd to 4th 3650
4th to 5th 3470

The major unknown variable I have is flywheel weight. I've gone with an average for a 4 pot petrol engine of about 8 kg but if I put a really heavy diesel one in there you'd need about 5 more bhp than shown above to haul it down the strip at the same speed.

This is your 1/4 mile run vs my computer's simulation with 123.8 wheel bhp

.......Yours..........Simulation
60ft.. 2.83...........2.83
330ft. 7.05...........7.00
1/8mi. 10.52 @ 71.9...10.50 @ 72.0
1000ft 13.51..........13.52
1/4m.. 16.04 @ 89.0...16.08 @ 89.0

I found it a very straightforward run to simulate. Once the power curve was correct all the times and speeds matched perfectly without any juggling so I can be certain the computer's power prediction is very accurate.

Edited by Pumaracing on Saturday 25th October 00:01

Edited by Pumaracing on Saturday 25th October 10:34

[Mr Whippy]

Pumaracing said:

I get an exact match to your 1/4 mile performance at every measurement point with a wheel bhp of 123.8 BUT ideally NOT starting off in 1st gear. The computer decided it was better to start off in 2nd which also saves one gearchange. All 1st gear does is spin the tyres pointlessly and you have to change into second before you even hit the 60 ft mark which mucks up that data point too. If I force it to start in 1st it changes the time at each measurement point by maybe 0.1 seconds. Nothing huge but the fit is slightly better starting in 2nd.

Flywheel figure is 148.7 bhp.

Top speed is rev limited to about 111 mph. It actually has the power to do 133 mph but the gearing is too low for that. Optimum gearchange points are

1st to 2nd 4200 rpm
2nd to 3rd 4000
3rd to 4th 3650
4th to 5th 3470

The major unknown variable I have is flywheel weight. I've gone with an average for a 4 pot petrol engine of about 8 kg but if I put a really heavy diesel one in there you'd need about 5 more bhp than shown above to haul it down the strip at the same speed.

This is your 1/4 mile run vs my computer's simulation with 123.8 wheel bhp

.......Yours..........Simulation
60ft.. 2.83...........2.83
330ft. 7.05...........7.00
1/8mi. 10.52 @ 71.9...10.50 @ 72.0
1000ft 13.51..........13.52
1/4m.. 16.04 @ 89.0...16.08 @ 89.0

I found it a very straightforward run to simulate. Once the power curve was correct all the times and speeds matched perfectly without any juggling so I can be certain the computer's power prediction is very accurate.

Ah I might have to go weigh a flywheel for one now (have a spare engine sat there, so could just unbolt it)

So in the case of the 2nd gear launch, I guess I am using lots of slip and so quite a bit of flywheel power as it were (ie, energy stored at higher rpm in the flywheel)...
The only downside for me with a 2nd gear launch is my new racing clutch, it's quite snappy, and agressive, so slip is generally a bit bad...
That said, the turbo lag in 1st is quite bad, gearbox ratio's designed for 90bhp are not doing it much good maybe

That figure looks very nice and close to what I sort of expected... I had been correcting my figure based on 120bhp wheels and using about 23bhp losses to get to ~ 143-144bhp flywheel...


Thank you very much for doing that.

I'm a firm believer in maths and real life testing giving as realistic results as any dyno... and while I sat up last night doing the figures for lots of cars, your +10bhp then divide by 0.9 seemed to work for all the dyno's I've seen on 306's (and the corrected figures given on Dyno Dynamics) and generally accepted wheels/flywheel figures for other cars...

My only thought now is that since the provided curve peaks at 3000rpm and then drops off, is that representative of the shape at the engine? Or will the fact the drivetrain and engine rpm speed is increasing (and so friction is increasing), and aero friction is increasing, will that drop-off be more of a plateau?


I'm unsure still why more power = more losses in my head, but I think I'm looking at it a little wrong, so I'll have to go through the thought-process again and see where I'm forgetting something... just to prove it right in my head

Thanks once again!

Dave

Edited by Mr Whippy on Saturday 25th October 11:03

[Pumaracing]

Mr Whippy said:

% loss is all arse anyway...

Want to retract that now ?

[Mr Whippy]

Pumaracing said:

Mr Whippy said:

% loss is all arse anyway...

Want to retract that now ?

What I might do is do some coastdown tests in and out of gear, and with the clutch up and down, in different gears, and then try isolate out my forces more, and see what I think.


It's not that I don't believe you, the figures work amazingly well for the peak calculation, I'm just sceptical about how simple 'rules of thumb' can be mis-understood by people, myself included, and make incorrect assumptions... ie, the two figures you gave me look right (wheels and flywheel), so does that mean the bhp loss curve vs engine rpm is linear? If so, is the figure you gave me at about 3100rpm... ie, at 4000rpm, the losses would be 4000/3100 * higher?

At 2000rpm, would the effective bhp losses be 2000/3000 * higher?


Ie, I'm currently using this...

http://www.forums.gtechpro.com/viewtopic.php?f=44&...

With my G-tech CSV file for the run (PM me and I can send it over)

It has inputs for corrections for aero, drivetrain etc, and then plots a power curve.

Which drivetrain loss correction to use though, is important. It has two, one where you input a % factor change (in my case to match your data, 17%), and then a second system where you define a bhp at a given rpm, at two points in the rev range, to get a more accurate change.

The % system works well, but it seems to exaggerate my peak torque to over 280lbft (eek), while also keeping that odd shape to the curve (drops off after 3200rpm or so, surely in real life aero losses would flatten that a little?!)
The second system seems to make a more realistic torque estimation while using your losses at peak power, while also getting the same peak power...


From your calculations does the power plateau, or falloff?

17% loss factor on torque all over the rpm range:


0bhp at 0rpm, 31bhp loss at 4000rpm, (effectively 24bhp loss at 3050rpm (ish) as per your figures):


Dave

[Mr Whippy]

Huff said:

I just don't get why something generates more friction because more torque is passing through it. Ie, a propshaft support bearing only see's velocity and the weight of the propshaft pushing down... it simply doesn't even care if it is idling or accelerating hard at all does it? How does it 'see' the load and make itself hotter?

Gears are rotating levers; the loads at the meshing teeth are reacted against the shaft bearings. So frictional losses due to both components (lsiding engagement at the tooth, and at the bearings) will be proportional to the torque transmitted. Since at a given rpm power is directly proportional to torque - if you increase the power transmitted you increase losses pro rata.

Ah that makes alot of sense for the cog wheel friction...

So then, do elements such as say a propshaft support bearing, or wheel bearing, generate any more friction, or is it reasonable to simply look at those with roughly friction ~ veloctity, irrespective of torque transmitted?

I wonder what the split is...

Like I said in a previous post, I'll have to go for a drive and use my G-tech to record the coast down phases in several gears, clutched, in/out of gear, etc, and then I can hopefully get an idea of the components and what kind of power they are using to decellerate the car

Sounds excessive, but it's really interesting, and means I can get a much better picture of what my various settings (ecu) are achieving


Thanks

Dave

[Pumaracing]

You're doing a masterful job of outthinking yourself in just about every aspect of this. What accelerates the vehicle is the power at the wheels. Neither the car nor my program care what the flywheel power curve and the transmission losses are and there is no way to derive these from testing the performance.

All I did was take the power curve you first gave me and scaled it up and down evenly everywhere until the 1/4 mile times matched. That's as much as my or anyone else's program can do. You can play with flywheel power curves and linear or exponential losses until you're blue in the face but you won't actually achieve anything because they'll all translate back to the same wheel curve. If you want to know the flywheel curve put it on an engine dyno.

[Mr Whippy]

Pumaracing said:

You're doing a masterful job of outthinking yourself in just about every aspect of this. What accelerates the vehicle is the power at the wheels. Neither the car nor my program care what the flywheel power curve and the transmission losses are and there is no way to derive these from testing the performance.

All I did was take the power curve you first gave me and scaled it up and down evenly everywhere until the 1/4 mile times matched. That's as much as my or anyone else's program can do. You can play with flywheel power curves and linear or exponential losses until you're blue in the face but you won't actually achieve anything because they'll all translate back to the same wheel curve. If you want to know the flywheel curve put it on an engine dyno.

I know that there is no way to derive the flywheel figure really effectively, what I am trying to determine now is how the curve I provided can be interpreted with respect to the test factors.

Ie, as the rpm's increased, power required to overcome aerodynamic losses increased exponentially, the power absorbed by the tyres increased, and I'm basically wondering if my power delivery actually peaks at 3100rpm really, or whether it does actually plateua or drop-off less significantly from that rpm to 3750rpm or so, as the different projection suggests.


I'm merely trying to interpret what my test data is saying, in a meaningful way.

Thank you for the help so far, it really is much appreciated... I am ultimately less interested in the outright figures than the shape of the curve (thought an outright guesstimate is often handy), I am simply wondering if I am actually losing THAT much power revving beyond 3100rpm or not.

I know the curve I sent you was straight from the G-tech, showing net wheel power, but without correcting for aerodynamic drag and other forces, is it actually not what the 'real' curve might look like? The two projections seem to make a large difference to that, so I'm just curious what might be a reasonable/sensible assumption to make from that about my revving habits


Thanks

Dave

[Pumaracing]

I'm getting rather lost I'm afraid in what you're saying or trying to achieve here. To answer the points as I think I understand them.

The flywheel bhp will always peak at higher rpm than the wheel bhp because tyre losses rise exponentially. If say a rolling road derived curve showed the same wheel bhp at 3000 and 3500 rpm you can be sure that the losses are higher at 3500 rpm and therefore so is the flywheel bhp. None of this matters to the vehicle. All it sees is wheel bhp and I've already given you your optimum gearchange rpm for every gear change. If you rev beyond these figures you'll go slower.

As for the wheel curve the G-Tech derived, if it didn't take any account of aero drag or rolling resistance it obviously won't be completely accurate. However aero drag is not a big factor until high speed and in 3rd gear which only goes to about 70 mph the error won't be large. The curve will clearly be more accurate at low speed and progressively less accurate, and in fact always show too low a bhp, at higher speeds. So quite possibly the power actually peaks a bit higher than this showed.

My own program can derive a completely accurate power curve from a set of acceleration figures taken in a single gear because it accounts for all drag and inertia factors but it can't do this automatically. I have to keep altering the power curve until the performance matches and that takes quite a while. I'm surprised the G-Tech can't accept drag data but then I've never used it.

[Mr Whippy]

Ah some of that makes sense now.

Like I said though, it's all about interpreting the data to get a realistic picture, not the picture you want... my problem has been not wanting a specific picture to sell myself anything, it's been wanting what is simply the picture as it is... and ultimately the wheels figure peak and the rough shape is about the best I need to know in that case as you have said

I think I may well go for a dyno, simply for the wheels figure curve, then I can cross-reference it to my G-tech plots and get an idea of the difference in the shape if there is one


It's just for the tuning aspect that I ideally wanted to get an idea, because what I ask of the engine is what the engine does, and knowing better what it is actually doing is good in that respect. Ie, 75mg of fuel per stroke = X Nm of torque at a given rpm is handy info for further tuning etc...


Thanks once more for the advice and help!

Dave

Edited by Mr Whippy on Sunday 26th October 16:39

[Pumaracing]

I suppose it's something I ought to have got round to sooner but I've spent today writing a program to calculate wheel torque and bhp from an acceleration curve in a given gear. Rather like the G-Tech does but my program takes account of all the following factors.

Tyre rolling resistance
Aerodynamic drag
The kinetic energy absorbed in accelerating wheel and tyre mass radially as well as in a straight line.
The kinetic energy absorbed in accelerating the flywheel, crank and other internal engine components.

It takes as its input the incremental time to cover each 1 mph so for example the input for 16 would be the time in seconds to accelerate from 15 mph to 16 mph. It obviously requires the gearing and tyre sizes to be known with certainty as well as the frontal area and drag coefficient. The wheel and tyre masses as well as the internal engine component masses I estimate from previous simulations.

It outputs wheel bhp and torque versus rpm at each 1 mph increment.

I've run some test data through it to verify it against my main simulation program but if you have a data stream input in the above format I'll see what it says about your actual wheel bhp and torque.

[Mr Whippy]

Pumaracing said:

I suppose it's something I ought to have got round to sooner but I've spent today writing a program to calculate wheel torque and bhp from an acceleration curve in a given gear. Rather like the G-Tech does but my program takes account of all the following factors.

Tyre rolling resistance
Aerodynamic drag
The kinetic energy absorbed in accelerating wheel and tyre mass radially as well as in a straight line.
The kinetic energy absorbed in accelerating the flywheel, crank and other internal engine components.

It takes as its input the incremental time to cover each 1 mph so for example the input for 16 would be the time in seconds to accelerate from 15 mph to 16 mph. It obviously requires the gearing and tyre sizes to be known with certainty as well as the frontal area and drag coefficient. The wheel and tyre masses as well as the internal engine component masses I estimate from previous simulations.

It outputs wheel bhp and torque versus rpm at each 1 mph increment.

I've run some test data through it to verify it against my main simulation program but if you have a data stream input in the above format I'll see what it says about your actual wheel bhp and torque.

2am

Still, that sounds really fun.

My wheel weights are give or take 16kg per corner including the tyre (8.4kg wheel, about 8kg tyre)...

I can get the flywheel weight in the next few weeks, next time I visit the spare engine

I will also send over a link to a G-tech file that has been exported to the CSV format (excel will import), that gives g's, extrapolated speed from that, time, rpm etc.


So I'm guessing you are a programmer extrodinaire as well? That sounds pretty damn impressive, and with rpm should calculate out power quite nicely too

I'll also make sure to use one of my latest developed map files for a new run, because I've made a few improvements since, and I think it's got a little bit more power beyond 3100rpm now (fingers crossed)


Thanks once again

Dave