Significant Recant - The New M3/4s are not Under Rated (long, technical)

[paddy335]

Originally Posted by swamp2

Just a fun hobby... Car enthusiasts can in general greatly benefit from a bit more science/engineering.

Mate, I am still in awe of you predicting the F8x performance numbers so accurately back in, what was it, 2011? Always pay a lot of attention to your analysis, even though that tome above got pretty heavy going

[CanAutM3]

Originally Posted by swamp2

Despite some of my criticisms of Quarter Pro (in general massively fewer inputs and outputs) the loss by gear is a very nice feature and your suggestion here is probably perfectly valid. I've followed some of the other discussions here on less gear engagements required for certain gears. However, are you certain that a 1:1 ratio absolutely means there should be 1/2 the losses in that gear (for this or in general for all MT transmissions exhibiting a 1:1 ratio). Is the same effect going to be present also in all DCTs with a 1:1? Now lastly the results for all of the metrics output from Quarter Pro don't depend on 5th gear, so in that respect this detail isn't at all relevant.

I am just nitpicking for the sake of improving the model. I agree that whatever is done to fifth gear will have zero impact on the early stage of the acceleration run (0-60mph, 0-100mph, ¼ mile, etc). Only high speed acceleration would be impacted.

Regarding transmission types, mostly front engine rear wheel drive cars, where the transmission input shaft is on the same axis as the output shaft, can get that benefit on the 1:1 ratio. I have personally not seen it on a FWD transaxle and I am not sure it could be done on a mid/rear engine car. It is also possible that some FR cars don't have a direct drive, but that would be a rather inefficient design as it introduces an additional set of gears.

I have also seen sketches of a 7 speed DCT with a 5th gear direct drive, so it is possible. I cannot confirm how the ones in the ///M models are configured though. Looking at cutaways or cross sections and counting the number of gear sets would tell us (8 gear sets equates to one direct drive gear, 9 gear sets equates to no direct drive).

The losses in the direct drive gear is probably less than half of the other gears in a MT. Since the power for other gears needs to go through two gear sets (from the input shaft to the the layshaft and then from the layshaft to the output shaft), the only losses left in the direct drive are the parasitic losses of oil churn, the gears spinning with no power transmitted and the losses in the bearings. On the DCT, the oil pump still needs to be driven even in the direct drive gear, so IMO there might not be as much reduction in losses (less than half) in the direct drive of a DCT.

[CanAutM3]

Originally Posted by Boss330

Was this dyno'd in the same gear?

That is a very good point that I wanted to bring up. The F8X dyno was likely done in the more efficient 1:1 fifth gear while the the E9X was most likely done in a less efficient gear. That can explain a ~2% difference. Further, because of inertial impacts, the E9X would probably show lower power in 7th gear (assuming the dyno could take the speed) even if there are less drivetrain losses.

[CanAutM3]

Originally Posted by swamp2

Unfortunately the M3 dyno database doesn't fully back up this idea. It shows a 318-361 hp variation for totally stock M3s. That is a 12% total spread from (min+max)/2. It is at the low end of the spectrum as compared to the other machines but is also about equal to the total loss of the vehicle. Not at all useful for an absolute value nor determining and under/over rating.

Pure speculation on my part, but my guess is that most dynapack operators don't run the dyno in steady state, thus negating most of the benefit that the dyno provides. Depending on the selected acceleration rate, the output results will vary. Further the test environment is very important, poor heat management will skew the results. Heat management is probably a reason why they are not operated in steady state to begin with, the test environments probably do not have the heat extraction capabilities for the extended time needed for steady state runs.

It does not take away CatPat8000's point that Dynapacks can be significantly more accurate, they just need to be operated properly.

[CanAutM3]

Another great thread. I really appreciate those discussions. They force me to think, learn from others and expand my knowledge. Thank you Swamp .

I agree with Swamp that the S55 and other BMW engine are most likely not "underrated", I seriously doubt that would be accepted by the regulatory agencies. I also agree that the "power plateau" provided by the FI engines are a significant contributor to the improved acceleration performance.

It however does not explain what we see on dynos. I think that there is more to it.

I will be the first one to say that dynos in general do not provide accurate or repeatable numbers. However, I believe that we will see a general trend where the S55 will produce more than the advertised 11hp difference over the S65 and that does tell us something. As I have stated in other threads, I believe this has to do with the testing standards and methodology combined with the new technologies introduced in the powertrains.

First of, the S55 powertrain has very likely less overall inertia that the S65 powertrain. Smaller crankshaft, two less camshafts, one less timing chain, CF driveshaft, lighter wheels and less mechanically driven accessories (more on this below). Those will all contribute to higher readings on chassis dynos.

An other example is the ever increasing use of electrically driven accessories over mechanically driven accessories. When establishing the "official" power ratings, IIRC, all accessories need to be driven and loaded (alternator, AC compressor, power steering pump, water pump, etc) thus lowering the rated power output.

On an engine like the S55, most of those accessories are electrically driven. A bigger alternator, the AC compressor and the oil pump are probably the only mechanically driven accessories left. During an acceleration run, the ECU disconnects some of the accessories reducing the parasitic losses. Further, since there are less mechanically driven accessories on the S55, it also reduces the overall inertia of the engine. This will show as more power to the wheels on a dyno operated in transient mode.

The above is only speculation on my part trying to understand what is being observed. It probably does not explain the whole difference but IMO, at least covers part of it.

[flinchy]

Originally Posted by catpat8000

As an observer following this thread, I have to say the english in your post is very difficult to understand. In addition, where do your trap speed numbers come from? I am not aware of any F80 trapping over 120 mph.

With respect to your dyno comments, I think one of the axioms in swamp's write-up, which is true, is that chassis dynos have so many variables, many of which operators don't control well, that they are simply not accurate in absolute terms and can be very inaccurate even in relative (run to run) terms.

So I would assert that most dyno runs are not useful real world data points.
Swamp's post is simply trying to show how the F80 doesn't need to be rated more than 425 hp to crush the E92 and perform at the level it does.

Bms one is

Ed: only loa bearing dynos (mustang, dynamics) have controllable variables and calibrations

Inertial dynos (dynojet) don't.

Inertial dynos are effected more by wheel weights and gearing, yes.. But as long as the operator is honest, picks the right gear etc. the numbers are legit.

[flinchy]

Originally Posted by swamp2

I am recanting a fairly strongly held prior position. It is also a very widely held position from many on this form. From my perspective that requires considerable evidence.

What part of massive random and deterministic errors present in all chassis/intertia dynos don't you seem to accept? Look at my the link with my analysis of E9X M3 dyno variation. Hmmm, perhaps you believe that S65 engines vary car to car by 15% or so? Sorry I simply don't accept that dynos are generally "real world".

In contrast I have reported real world performance result and compared those against simulation. Perhaps you also misunderstand the obvious yet somewhat subtlety laden idea that there is no such thing a THE number for any performance metric for any given car. The "database" of E9X M3 real world results I posted clearly demonstrates this.

As far as one car being a baseline for another car, even on the same dyno "proves" absolutely nothing.

There is now consistency among three things; 1. Factory quoted power figures, 2. Two totally different simulation programs that have demonstrated success predicting E9X M3 performance quite precisely (not to mention hundreds of other cars in the standard CarTest library...) and 3. Observed performance results. The "odd man out" here are the dynos. They are a bit like religion, which one is right, surely not all of them can be.

Lastly the Quarter Pro results are for the 6MT E90 M3, not the DCT. That number very well might be below the eventual average or best number we will have from a large array of results. That will only mean that the aero or parasitic drive train losses are also a small bit off of the real world. This minor difference in no way is some "nail in the coffin" on my method nor conclusion.

I'm not talking about comparing numbers from different dynos.

I'm talking a base tune s65 or n54 etc on bms's dyno (for example), similar conditions

To a base tune s55

The s55 made more peak power untuned on the same dyno in similar conditions, than an s65. That's all i'm saying there.

Remember, we're talking dynojet... Dynojet DON'T vary (much), they're a fixed load that's the beauty of them

The numbers may not be 100% accurate depending who you talk to... But as a comparison tool, as long as atmospheric conditions are the same, it's the only dyno's you CAN compare.. Ish.

[Boss330]

Originally Posted by flinchy

I'm not talking about comparing numbers from different dynos.

I'm talking a base tune s65 or n54 etc on bms's dyno (for example), similar conditions

To a base tune s55

The s55 made more peak power untuned on the same dyno in similar conditions, than an s65. That's all i'm saying there.

Remember, we're talking dynojet... Dynojet DON'T vary (much), they're a fixed load that's the beauty of them

The numbers may not be 100% accurate depending who you talk to... But as a comparison tool, as long as atmospheric conditions are the same, it's the only dyno's you CAN compare.. Ish.

Didn't swamp do a statistical analysys that showed a spread of 13-17% on dynos, including the Dynojet?

[CanAutM3]

Originally Posted by Boss330

Didn't swamp do a statistical analysys that showed a spread of 13-17% on dynos, including the Dynojet?

I think that flichy's point is that comparing two cars on the same physical dyno with the same operator, there will be less variation than the 13-17% total spread seen on the dyno database for Dynojets.

My guess is that the error is still high, but with enough statistical data points, I believe that, on average, we will see F8X put more power down at wheel than the 11hp rated difference with the E9X M3.

[ASAP]

I think it's very simple... two cars were dyno'd on the same dyno. S65 vs S55.. one showed considerably higher power and tq... trap speeds have backed this up so far. Everything else is mumbo jumbo that means nothing in the real world and to most people.

Also... not to be a pain... but I would change those 60 ft numbers immediately to prevent myself from looking like a fool lol. 1.6s and 1.7s on rwd cars on street tires are laughable at best.

[Black Gold]

Originally Posted by CanAutM3

It however does not explain what we see on dynos. I think that there is more to it.

I will be the first one to say that dynos in general do not provide accurate or repeatable numbers. However, I believe that we will see a general trend where the S55 will produce more than the advertised 11hp difference over the S65 and that does tell us something. As I have stated in other threads, I believe this has to do with the testing standards and methodology combined with the new technologies introduced in the powertrains.

First of, the S55 powertrain has very likely less overall inertia that the S65 powertrain. Smaller crankshaft, two less camshafts, one less timing chain, CF driveshaft, lighter wheels and less mechanically driven accessories (more on this below). Those will all contribute to higher readings on chassis dynos.

An other example is the ever increasing use of electrically driven accessories over mechanically driven accessories. When establishing the "official" power ratings, IIRC, all accessories need to be driven and loaded (alternator, AC compressor, power steering pump, water pump, etc) thus lowering the rated power output.

On an engine like the S55, most of those accessories are electrically driven. A bigger alternator, the AC compressor and the oil pump are probably the only mechanically driven accessories left. During an acceleration run, the ECU disconnects some of the accessories reducing the parasitic losses. Further, since there are less mechanically driven accessories on the S55, it also reduces the overall inertia of the engine. This will show as more power to the wheels on a dyno operated in transient mode.

The above is only speculation on my part trying to understand what is being observed. It probably does not explain the whole difference but IMO, at least covers part of it.

ok, but how do you explain the same phenomenon in

the merc m157 (new 5.5l turbo v8)
the audi s4 3.0 SC
the 335i n54

all of which also dyno'd at much higher than advertised numbers using the standard 15-20% DT loss.

none of those had a CF driveshaft or the like

I don't know the answer, im asking

[swamp2]

Originally Posted by paddy335

Mate, I am still in awe of you predicting the F8x performance numbers so accurately back in, what was it, 2011? Always pay a lot of attention to your analysis, even though that tome above got pretty heavy going

Thanks, yes, long/heavy. Again when recanting a strongly held prior position significant evidence is needed.

[swamp2]

Originally Posted by CanAutM3

Another great thread. I really appreciate those discussions. They force me to think, learn from others and expand my knowledge. Thank you Swamp .

I agree with Swamp that the S55 and other BMW engine are most likely not "underrated", I seriously doubt that would be accepted by the regulatory agencies. I also agree that the "power plateau" provided by the FI engines are a significant contributor to the improved acceleration performance.

It however does not explain what we see on dynos. I think that there is more to it.

I will be the first one to say that dynos in general do not provide accurate or repeatable numbers. However, I believe that we will see a general trend where the S55 will produce more than the advertised 11hp difference over the S65 and that does tell us something. As I have stated in other threads, I believe this has to do with the testing standards and methodology combined with the new technologies introduced in the powertrains.

First of, the S55 powertrain has very likely less overall inertia that the S65 powertrain. Smaller crankshaft, two less camshafts, one less timing chain, CF driveshaft, lighter wheels and less mechanically driven accessories (more on this below). Those will all contribute to higher readings on chassis dynos.

An other example is the ever increasing use of electrically driven accessories over mechanically driven accessories. When establishing the "official" power ratings, IIRC, all accessories need to be driven and loaded (alternator, AC compressor, power steering pump, water pump, etc) thus lowering the rated power output.

On an engine like the S55, most of those accessories are electrically driven. A bigger alternator, the AC compressor and the oil pump are probably the only mechanically driven accessories left. During an acceleration run, the ECU disconnects some of the accessories reducing the parasitic losses. Further, since there are less mechanically driven accessories on the S55, it also reduces the overall inertia of the engine. This will show as more power to the wheels on a dyno operated in transient mode.

The above is only speculation on my part trying to understand what is being observed. It probably does not explain the whole difference but IMO, at least covers part of it.

Good post. I think you have covered some of the factors here for sure. Undoubtedly. But as you mention is it not the whole difference. Perhaps the most significant single factor may not even be mentioned.

[swamp2]

Originally Posted by flinchy

Ed: only loa bearing dynos (mustang, dynamics) have controllable variables and calibrations

Well if they were truly "controllable" we would not see 15% or so variation among operators.

Originally Posted by flinchy

Inertial dynos are effected more by wheel weights and gearing, yes.. But as long as the operator is honest, picks the right gear etc. the numbers are legit.

I disagree. What does "legit" mean, precisely? Precise, accurate, for one operator for all, etc.

Originally Posted by flinchy

The s55 made more peak power untuned on the same dyno in similar conditions, than an s65. That's all i'm saying there.

Yes, no disgareement here. They car has more stated hp but makes way more hp than stated compared to the S65, this is the mystery. It is abundantly clear that the car does not make 425 rwhp. I think I proved that well beyond doubt. We are not going to see S55 Ms trapping at 121-122 (fully stock including tires).

Originally Posted by flinchy

Remember, we're talking dynojet... Dynojet DON'T vary (much), they're a fixed load that's the beauty of them

Dynotjet, S65, stock vehicles, 13% variation: Link here.

Originally Posted by flinchy

The numbers may not be 100% accurate depending who you talk to... But as a comparison tool, as long as atmospheric conditions are the same, it's the only dyno's you CAN compare.. Ish.

Not according to the above...

[swamp2]

Originally Posted by ASAP

I think it's very simple... two cars were dyno'd on the same dyno. S65 vs S55.. one showed considerably higher power and tq... trap speeds have backed this up so far. Everything else is mumbo jumbo that means nothing in the real world and to most people.

Again, 425 hp to the wheel (EAS dyno) should make the car 1/4 in 11.5-11.6 @ 121-122. It simply is not making anywhere close to this power. So no trap speeds do not in anyway justify this EAS dyno.

Honestly, if you think simulation is "mumbo jumbo" you should just stop driving your car (and throw away your computer and mobile phone, and never ride in an airplane, etc., etc.). Do you have any idea how much simulation is used to design modern cars? I work in this field and what is simulated spans every major subsystem of the vehicle and includes huge efforts by the OEM and all tiers of suppliers. It is a major predictive enabler that makes modern cars as good (and continually improving) as they are. These techniques increase understanding, speed development, save cost, reduce and in some cases eliminate test. They are almost always compared to test though to insure correspondence, strengths and weaknesses. The simulation techniques used in CarTest and Quarter Pro are simply much more simple version of those type of simulations.

Originally Posted by ASAP

Also... not to be a pain... but I would change those 60 ft numbers immediately to prevent myself from looking like a fool lol. 1.6s and 1.7s on rwd cars on street tires are laughable at best.

This has been discussed. Traction models are just that, models and they are important in such simulations. Most of them are underestimating 60 foot times. Such isolated nitpicking of results is very healthy for good discussion. However, at the same time, you can not deny the predictive capability and accuracy of the plethora of other numbers resulting from simulation.

The traction models clearly allow wheelspin at launch, during shifts, account for weight transfer and decreased tractive force during wheelspin. The fundamentals are all there, sure they aren't prefect and seem to represent more like a drag tire on a drag course.

Care to comment on the E9X M3 simulation results vs. test. Pick any number you like other than the 60' time. Ever hear of not throwing out the baby with the bathwater. It applies here. I'd rather have poor 60' numbers and wonderful everything else than some bogus test telling us that the car makes 425 rwhp...

Of course, as you should know, I can't and won't just "change the numbers". They are what they are as the result of the inputs, physics and algorithms used by the tools. Anything else would be massively dishonesty.

[ASAP]

Originally Posted by swamp2

Again, 425 hp to the wheel (EAS dyno) should make the car 1/4 in 11.5-11.6 @ 121-122. It simply is not making anywhere close to this power. So no trap speeds do not in anyway justify this EAS dyno.

Honestly, if you think simulation is "mumbo jumbo" you should just stop driving your car (and throw away your computer and mobile phone, and never ride in an airplane, etc., etc.). Do you have any idea how much simulation is used to design modern cars? I work in this field and what is simulated spans every major subsystem of the vehicle and includes huge efforts by the OEM and all tiers of suppliers. It is a major predictive enabler that makes modern cars as good (and continually improving) as they are. These techniques increase understanding, speed development, save cost, reduce and in some cases eliminate test. They are almost always compared to test though to insure correspondence, strengths and weaknesses. The simulation techniques used in CarTest and Quarter Pro are simply much more simple version of those type of simulations.

I do think it is mumbo jumbo for the average reader / user. As you have just mentioned and since I work in an engineering company, I understand the value of models... I also know full and well that models are just that. Real world often brings different realities.

I also do not believe the M3 is making 425 whp... I feel that it is closer to 405-410 WHP, which would then make the traps appropriate. I am basing thins on terry's and EAS' use of that dyno which have both been shown to dyno about 15-20 WHP high when compared to other DJs. At that point, everything starts to make sense. None of this changes, the fact that on the same dyno... no matter what mumbo jumbo you write; the F80 makes significantly more power than than the E90... it's as simple as that.


This has been discussed. Traction models are just that, models and they are important in such simulations. Most of them are underestimating 60 foot times. Such isolated nitpicking of results is very healthy for good discussion. However, at the same time, you can not deny the predictive capability and accuracy of the plethora of other numbers resulting from simulation.

The traction models clearly allow wheelspin at launch, during shifts, account for weight transfer and decreased tractive force during wheelspin. The fundamentals are all there, sure they aren't prefect and seem to represent more like a drag tire on a drag course.

Care to comment on the E9X M3 simulation results vs. test. Pick any number you like other than the 60' time. Ever hear of not throwing out the baby with the bathwater. It applies here. I'd rather have poor 60' numbers and wonderful everything else than some bogus test telling us that the car makes 425 rwhp...

Of course, as you should know, I can't and won't just "change the numbers". They are what they are as the result of the inputs, physics and algorithms used by the tools. Anything else would be massively dishonesty.

Differing opinions here... I've said all I'm gonna say with my above statement. There is no more to talk about here nor to argue.

[CanAutM3]

Originally Posted by swamp2

Good post. I think you have covered some of the factors here for sure. Undoubtedly. But as you mention is it not the whole difference. Perhaps the most significant single factor may not even be mentioned.

Agreed, there is more to it and I have not yet put the finger on it. I am eager to find out.

Regarding the EAS dyno output of the M4, I believe it showed 412~414hp corrected to SAE standards, not 425hp. Not a huge difference, but one nonetheless.

[swamp2]

Originally Posted by ASAP

I also do not believe the M3 is making 425 whp... I feel that it is closer to 405-410 WHP, which would then make the traps appropriate. I am basing thins on terry's and EAS' use of that dyno which have both been shown to dyno about 15-20 WHP high when compared to other DJs. At that point, everything starts to make sense. None of this changes, the fact that on the same dyno... no matter what mumbo jumbo you write; the F80 makes significantly more power than than the E90... it's as simple as that.

Differing opinions here... I've said all I'm gonna say with my above statement. There is no more to talk about here nor to argue.

Seems obvious you are discounting the idea of the time averaged power. This is the key contributor to making the new M3/4 fast, not its peak crank power being significantly over the factory claimed number. And again here for both peak power and the resulting performance I mean those in comparison to the E9X M3.

Also 405-410 whp does not match current traps. 11.9@117 is 425 hp * (1 - 15% loss) or around 360 whp. I could run a simulation with 410 whp but since you think all of these results are useless I suppose I won't bother. Also as posted below, the 121 trap was for 417 whp, not 425. This car performs like a 360 whp car with 360 whp available from 5500 rpm to redline (along with massively more hp than the S65 below 5500 all the way down basically to idle - as per published BWM documents). You just can't compare cars with similar peak power when their power curves look as different as the S55 and S65, completely nonsensical comparison. You know this - modern turbos are a "different beast".

Nothing wrong with some healthy debate. I'd like to think of this as debate rather than argument...

[swamp2]

Originally Posted by CanAutM3

Agreed, there is more to it and I have not yet put the finger on it. I am eager to find out.

Regarding the EAS dyno output of the M4, I believe it showed 412~414hp corrected to SAE standards, not 425hp. Not a huge difference, but one nonetheless.

For the simulation I did in CarTest using a digitized EAS dyno I ended up with a peak rwhp of 417. It also did not have a true 417 hp flat from 5500 rpm to redline. This is quite a bit closer to the numbers you mention above. That still resulted in the staggering performance numbers I mentioned in the OP and again just a few posts above.

Good luck on the ongoing dyno investigation. If you can "solve" this, it will be a huge success.

[ASAP]

Originally Posted by swamp2

Seems obvious you are discounting the idea of the time averaged power.This is the key contributor to making the new M3/4 fast, not its peak crank power being significantly over the factory claimed number. And again here for both peak power and the resulting performance I mean those in comparison to the E9X M3.

I have a feeling like this would have more to do with times covered over a distance more so than peak trap speeds where peak power is more contributing.


Also 405-410 whp does not match current traps. 11.9@117 is 425 hp * (1 - 15% loss) or around 360 whp. I could run a simulation with 410 whp but since you think all of these results are useless I suppose I won't bother. Also as posted below, the 121 trap was for 417 whp, not 425. This car performs like a 360 whp car with 360 whp available from 5500 rpm to redline (along with massively more hp than the S65 below 5500 all the way down basically to idle - as per published BWM documents). You just can't compare cars with similar peak power when their power curves look as different as the S55 and S65, completely nonsensical comparison. You know this - modern turbos are a "different beast".

I am confused here a bit... I don't know of a single car with this weight and power level that could trap 117 @ 360 WHP. I can guarantee I could do not 117 right now on my pump gas tune in a lighter 135i making about 360-370 WHP. I simply think that you may be underrating power and overrating performance.

Nothing wrong with some healthy debate. I'd like to think of this as debate rather than argument...

.

[CanAutM3]

Originally Posted by KennyPowers

ok, but how do you explain the same phenomenon in

the merc m157 (new 5.5l turbo v8)
the audi s4 3.0 SC
the 335i n54

all of which also dyno'd at much higher than advertised numbers using the standard 15-20% DT loss.

none of those had a CF driveshaft or the like

I don't know the answer, im asking

Agreed there is still more to it. My point is that it is not possible to use that blanket 15-20% "drivetrain loss" number that is so often used. As I posted more than once, I suspect that there is also something with the testing standard/methodology where FI engines react differently when tested on a bench dyno at steady state compared to accelerating on a chassis dyno. I have not yet found/read a good technical explanation for this though.

Note that the CF driveshaft is only one small part of the total equation in the total powertrain intertia.

On the engines you referenced, do they also leverage electrically driven accessories?

[Myrder]

Originally Posted by swamp2

Again, 425 hp to the wheel (EAS dyno) should make the car 1/4 in 11.5-11.6 @ 121-122. It simply is not making anywhere close to this power. So no trap speeds do not in anyway justify this EAS dyno.

The car will trap 120-121 when the weather cools some more and the cars are fully broken in. Just IMO. EAS didn't do a 1/4 mile yet and Terry has only done VBOX times in the heat. I would wait to see some actual 1/4 mile times in nice weather.