Further dyno evidence of no under rating

[jc05e46m3]

Originally Posted by FogCityM3

I feel like I'm picking up my kids at school, with all the "Butt-Hurt" commentary that seems to show up here in disproportionate amount, even relative to the school. Nowhere did Swamp even bring in the S65 into the picture or claim that the old car was as fast as the new one. The whole thesis is that the shape of the curve, with flat constant, full power over a very wide RPM range is responsible for how quick the M4 is. I personally think its extremely interesting and I have learned many things from swamp and others in these analyses, they touch on something that I love and gets my neurons firing outside my area of expertise.

I also think the Insoric is extremely interesting tool, especially the "coast down" feature which will help estimate driveline losses. I would love to use this tool, the BT tool and actual moment calculations (which come directly from the ECU and is what the ECU believes to be the power given all the real time inputs), plus a similar Dash Dyno tool that I use, where you enter (and can calculate) gear ratios, temperature, frontal area, pressure, etc and it plugs into the ECU to get speed, acceleration to obtain a rolling road whp calculation. Between all of these we can triangulate a true dyno plot with full heat exchange and very accurate estimates of driveline losses over an rpm range (that go far beyond the traditional "divide by .85" to convert from driveline to bhp). I've said this many times in M3 post, the wheel dynos leave a lot to be desired with huge variation, no matter many are in whatever database is out there.

You don't like the "butt hurt" verbiage? Get over it. Free speech is a hell of a thing.

Anyway, your rant was misguided and you quite possibly missed my points. What I was saying was a couple things.

1. When trying to be this "exact" and "accurate" poster, one should not contradict one's own argument (Or post something that misguides thoughts)

2. I have never talked trap speeds because there are so many factors that all play into that. I was simply talking about dynos. Whether it's a "happy dyno" or a "heartbreaker dyno" or any other excuses that people use to determine whatever they want to believe, is irrelevant. I used the E9x/F8x example just because that is what has been thrown around the most.

To elaborate (and this is excruciatingly simple and repeated so often)
If I bring my E92 to shop X and dyno at 360hp
then F82 to shop X the same day/similar time and conditions and dyno at 400
Great, 100hp delta

If I bring it to Shop Y and dyno e92 at 352
and then f82 at 396

Then at shop Z e92 reaches 365
and f82 reaches 410

The variation may inflate/deflate numbers, but educated assumptions can be derived from comparisons of a known data point. I digress...

3. With them being rated at 414/425 respectively, IMHO, it's a safe bet that the car is underrated. And with that being said, who cares? Deltas are what's important.
Do I think it makes 500bhp? Negative. I think its closer to 440-450 so I understand Swamp's frustration with some members when they claim these glorious things. I do, however, agree about the area under the curve (but that has also always been obvious... ) That is why the M4 is so much quicker, sure.

Anyway, thanks.

[GrussGott]

Originally Posted by FogCityM3

I would love to use this tool...where you enter (and can calculate) gear ratios, temperature, frontal area, pressure, etc and it plugs into the ECU to get speed, acceleration to obtain a rolling road whp calculation.

Why do you care to be so accurate (assuming you can be)?

[swamp2]

Originally Posted by FogCityM3

I feel like I'm picking up my kids at school, with all the "Butt-Hurt" commentary that seems to show up here in disproportionate amount, even relative to the school.
...
I also think the Insoric is extremely interesting tool, especially the "coast down" feature which will help estimate driveline losses. I would love to use this tool, the BT tool and actual moment calculations (which come directly from the ECU and is what the ECU believes to be the power given all the real time inputs), plus a similar Dash Dyno tool that I use, where you enter (and can calculate) gear ratios, temperature, frontal area, pressure, etc and it plugs into the ECU to get speed, acceleration to obtain a rolling road whp calculation.
...

Thanks for an appropriate and intelligent response. Inappropriate, childish and hyperbole will unfortunately remain the norm in most of these threads. Who really cares if the car only makes 10 hp more than the outgoing model. It's performance trouncing is what matters...

[swamp2]

Originally Posted by jc05e46m3

You don't like the "butt hurt" verbiage? Get over it. Free speech is a hell of a thing.

Anyway, your rant was misguided and you quite possibly missed my points. What I was saying was a couple things.

One can have plenty of completely free speech and also a modicum of decorum as well...

Originally Posted by jc05e46m3

1. When trying to be this "exact" and "accurate" poster, one should not contradict one's own argument (Or post something that misguides thoughts)

I think I've addressed this 2% is within experimental error...

Originally Posted by jc05e46m3

2. I have never talked trap speeds because there are so many factors that all play into that. I was simply talking about dynos. Whether it's a "happy dyno" or a "heartbreaker dyno" or any other excuses that people use to determine whatever they want to believe, is irrelevant. I used the E9x/F8x example just because that is what has been thrown around the most.

Trap speeds are a better indication of power than chassis dynos, especially when you have a cross section of multiple results. Obviously you can have a "happy" 1/4 run or a "heartbreaker" one as well but times are more likely to be affected than traps...

Originally Posted by jc05e46m3

To elaborate (and this is excruciatingly simple and repeated so often)
If I bring my E92 to shop X and dyno at 360hp
then F82 to shop X the same day/similar time and conditions and dyno at 400
Great, 100hp delta

If I bring it to Shop Y and dyno e92 at 352
and then f82 at 396

Then at shop Z e92 reaches 365
and f82 reaches 410

The variation may inflate/deflate numbers, but educated assumptions can be derived from comparisons of a known data point. I digress...

Delta's should in theory be better predicted by a chassis dyno, but when deltas differ by as much as they have in this case (10 - 150 hp) how do you pick the
"correct" delta?

Originally Posted by jc05e46m3

3. With them being rated at 414/425 respectively, IMHO, it's a safe bet that the car is underrated. And with that being said, who cares? Deltas are what's important.
Do I think it makes 500bhp? Negative. I think its closer to 440-450 so I understand Swamp's frustration with some members when they claim these glorious things. I do, however, agree about the area under the curve (but that has also always been obvious... ) That is why the M4 is so much quicker, sure.

Anyway, thanks.

We mostly agree here, although I think a reasonable limit of under rating is ~5% or 20 hp. This covers the gap well between the upper trap speeds and reasonable simulation results.

[swamp2]

We're largely not getting on the same wavelength here at all...

Originally Posted by Stevens21234

I've examined your threads with software and disagree on the approach. At the end of the day, you're trying to replicate the dyno curves by using sensitivity analysis of various software analysis parameters for that specific curve. You try to limit uncertainty and randomness by trying to derive the curves for different tests at the same company. But again, you're assuming that specific dyno is correct. I could use the same software, curve fit and come up with favorable performance measurement parameters for a variety of different dynos (witha 13% spread) as you've analyzed.

Your characterization here is entirely incorrect, especially the bold part. I have no interest in "replicating" a dyno. Factory engine dyno (NOT CHASSIS inertial/wheel hp) data are one of the primary a-priori inputs into these types of tools. The statement itself is also not at all completely clear at that...

My spreadsheet I've constructed 100% on my own as well as both CarTest and Quarter Pro commercial software tools are first principles, physics based, performance simulation. The basically just include F=ma, T=Iα, P=Fv. Yes some input such as inertia, losses and friction are difficult to determine and are test derived, but all simulation regardless of how rigorous, is nearly always informed by test data for multiple parameters.

Originally Posted by Stevens21234

Analysis software would introduce results in a vacuum. furthermore, you could even develop a probabilistically based approach for nonlinear inelastic dynamic finite element analysis and come up with median values that would be more reliable than focusing on singular dynos. Errors would be limited if we could use the cad (to transform into FEA) or finite element model files from the manufacturers themselves (which could be mapped into one consistent program). Doing so would allow for replicable analysis with simulated vacuum conditions limiting variance in extrinsic testing condition properties. Doing so would allow you to consistently use the same modeling approach with again nonlinear and inelastic material deformation to capture small deformations (due to transmission movement and other car component movement) and thermal properties (something typically missing in rigid kinematic modeling) further limiting analysis spread error.

I too have access to a host of these tools as I helped google develop some of their technology behind their self driving car while I was a student at that institute of technology in massachusetts.

I still strongly disagree, there is little room for or need for FEA or any spatial discretization technique in straight line vehicle maximum performance simulation. Let's start from absolute basics. Here is what matters and what is needed to accurately model it in order of increasing sophistication:

1. Peak power. Nothing more, since cars power typically varies by more than their weight (certainly in a given class/size of vehicle, power is absolutely number one
2. Power (peak) to weight ratio. Thing single best simple predictive tool. Although significant variation exists. This is the most simple/telling metric. See more here.
3. Add the full power curve to the above. This is especially important if you want to compare curves with radically different engine character (i.e. different shaped power curves). Comparison the S65 to S55 engines this makes 25 hp of difference! More on this here.
4. An estimate of time averaged power to the wheels including gearing and weight. Both aerodynamics and tire rolling losses are the last essential components and these come from quite accurate closed form formulas, which I have discussed here on the forum in other placed. This tecnique solves for x,v,a all explicitly as a function of time across gears.
5. One can further refine the above with acceleration based weight transfer, and various dynamic tire friction models. These advanced features are in all of the tools I have mentioned. Mine and commerical tools.

The main uncertainties in the last approach are roughly in order:

1. Parasitic drivetrain losses (more on estimating those here and here)
2. Drivetrain inertia. I have a big post upcoming on this. It isn't rocket science but is largely overlooked and given CAD drawings (no simulation) can be computed extremely accurately.
3. Road tire dynamic friction. Mainly affects time to speed not trap speed and things like 60 foot times. Models are largely capable of strong predictive capability for both rolling, slight slip and large slip (full spin). Of course, matching what is happening when on any given launch is harder and this is in part most of the reason for large variations in 0-60 and 60' times reported from test.

What role do finite elements have here? Little to none. It is complete overkill. And, just like in these simple tire models detailed road/tire FEA models are laden with unknown or purely test derived constants.

If you are so confident in the value of FEA for accurate a-priori type results here knock yourself out. Really put up or shut up. Let's see them. Either a component or system level model. It is most likely I would tear them apart from an "a-priori" perspective. The above approach I use, with very little "curve fitting" has shown to be highly accurate over a large range of standard performance metrics. Now one place a more advanced form of modeling may have something to contribute is a better understanding of why turbo charged vehicles are much more likely to show a significant under rating on a (cheap, demonstrably inaccurate) inertial dyno. However, FEA would again be overkill. A rigid body kinematic model would likely be more than sufficient. Start there, if you truly solve that puzzle you'd be a hero to me and many others here and in other places...

It's much easier to be an armchair critic than producing something of value.

[FogCityM3]

More intellectual curiosity than anything. For example, have been logging some E85 mixes and the bhp gain according to the BT tool is quite substantial, ie almost too good to believe. That seems interesting to me, as are the concepts of how to derive estimates of measured power accurately, various shortcomings, how thing work, etc. I tend to be analytical by nature (in a different line of work though) so these types of things interest me quite a bit and I always like learning new things. So I see accuracy/validation using different approaches as a challenge that's worthwhile to learn, personally speaking.

Originally Posted by GrussGott

Why do you care to be so accurate (assuming you can be)?

[GrussGott]

Originally Posted by FogCityM3

More intellectual curiosity than anything. For example, have been logging some E85 mixes and the bhp gain according to the BT tool is quite substantial, ie almost too good to believe. That seems interesting to me, as are the concepts of how to derive estimates of measured power accurately, various shortcomings, how thing work, etc. I tend to be analytical by nature (in a different line of work though) so these types of things interest me quite a bit and I always like learning new things. So I see accuracy/validation using different approaches as a challenge that's worthwhile to learn, personally speaking.

The e85 thing is pretty intriguing, hopefully you'll share the outcome? I'm also curious about octane and useful max for this car stock. For the e9x i believe anything over 93 was useless with 91 being the sweet spot in terms of $ for daily performance

[ASAP]

Are we talking about E85 on an S65?

[jc05e46m3]

Originally Posted by swamp2

One can have plenty of completely free speech and also a modicum of decorum as well...



I think I've addressed this 2% is within experimental error...



Trap speeds are a better indication of power than chassis dynos, especially when you have a cross section of multiple results. Obviously you can have a "happy" 1/4 run or a "heartbreaker" one as well but times are more likely to be affected than traps...



Delta's should in theory be better predicted by a chassis dyno, but when deltas differ by as much as they have in this case (10 - 150 hp) how do you pick the
"correct" delta?



We mostly agree here, although I think a reasonable limit of under rating is ~5% or 20 hp. This covers the gap well between the upper trap speeds and reasonable simulation results.

I don't think you "pick" the right one, but rather just use gains (or differences in individual cars on the same dyno) as a reference. You're absolutely right though, these don't give an accurate representation of true power.

Also, I think we are definitely in a "mostly agree" situation. I think the car makes 440-450 bhp and from seeing your other posts, seems like you're guessing right around the middle of that.

Thanks for doing the legwork to try to figure it all out though. We definitely appreciate the good input on the forums and enthusiasts who truly care.

Originally Posted by ASAP

Are we talking about E85 on an S65?

Sounds like it. There has been a lot of talk about this on the E9x forums lately!

[ASAP]

Originally Posted by jc05e46m3

Sounds like it. There has been a lot of talk about this on the E9x forums lately!

I just wandered over there and can't find anymore info, is there a link?

Who upgraded the pumps and injectors? and what mixes?

[CSanto]

I dont get this thread. Look at the pocono shoot out vids. I'm not saying its underrated, but the package (weight, TQ, HP, Powerband) is recipe for a seriously quick car.

[jc05e46m3]

Originally Posted by ASAP

I just wandered over there and can't find anymore info, is there a link?

Who upgraded the pumps and injectors? and what mixes?

I'll try to find a link. AFAIK, no one had upgraded pumps or injectors. It was very minimal mixtures like E20/E30.

[Boss330]

Originally Posted by CSanto

I dont get this thread. Look at the pocono shoot out vids. I'm not saying its underrated, but the package (weight, TQ, HP, Powerband) is recipe for a seriously quick car.

And, once again, no one is disputing that it's a seriously quick car that trounces the E9x M3.

But, looking at those videos from Pocono, would you draw a conclusion as to which dyno result, or hp rating, that is the correct one based on those videos?

That is what this thread is about. Why do we see such a spread on dynos (not only for the F8x, but for every car) and is the S55 really under rated by more than 5%? The postulation is; NO. The trap speeds etc is what the F8x should do, with it's broad power plateau, and within a 5% tolerance on factory claimed HP.

[ASAP]

Originally Posted by jc05e46m3

I'll try to find a link. AFAIK, no one had upgraded pumps or injectors. It was very minimal mixtures like E20/E30.

Yes that was what i was thinking. The adjustments in timing and cooling benefits that can be had w e85 are huge.

[swamp2]

Originally Posted by jc05e46m3

Also, I think we are definitely in a "mostly agree" situation. I think the car makes 440-450 bhp and from seeing your other posts, seems like you're guessing right around the middle of that.

Thanks for doing the legwork to try to figure it all out though. We definitely appreciate the good input on the forums and enthusiasts who truly care.

To be clear I think that 5% is an upper limit. I also believe that the cars are 425 crank hp on the money along with say a 1% manufacturing variation.

Your welcome.

[MFL]

................
JB4 + E30 Blend..

[jc05e46m3]

Originally Posted by MFL

................
JB4 + E30 Blend..

That was E50**....

[swamp2]

Let's not get too far off topic here. This is not about tuned M3/4s at all. Stock on regular pump gas only.

[MFL]

Originally Posted by jc05e46m3

That was E40....

40% of E85 (85% Ethanol 15% Petrol) in a 15.85 Gallon tank is almost exactly E30 (30% ethanol, 70% petrol)....

These cars are drastically under rated and this post is pointless.

Think I'm wrong? Go drive an M4, then go drive an E92 M3. For having such a close power rating, these M4's sure do drive a lot faster for some mysterious reason that can be in no way/shape/form be attributed to the verifiable numbers that they produce at the wheels at several independent dynos.

Nope. Lies. Shenanigans. Witchcraft.

[swamp2]

Originally Posted by MFL

These cars are drastically under rated and this post is pointless.

There is significant evidence to the contrary, trap speeds, dyno results (here) and simulation. The only evidence of this is dynos which are completely all over the map from no under rating to about 75 hp of under rating. Have a read of the (my) other thread on this topic where there is pretty much a consensus of a 5% maximum under rating, which according to EU law is not really technically an under rating...

Originally Posted by MFL

Think I'm wrong? Go drive an M4, then go drive an E92 M3. For having such a close power rating, these M4's sure do drive a lot faster for some mysterious reason that can be in no way/shape/form be attributed to the verifiable numbers that they produce at the wheels at several independent dynos.

Yes, very much so.

Another fallacy purporting that personal experience is more valuable than science... We all know what the performance numbers are and that is what matters. In fact driving the car, especially without the proper contextual understanding of the importance of radically different shaped power curves is in fact likely more confounding to the issue than enlightening.

Again, what is "THE" conclusion from dynos? The wide range of result on both the F8X M3/4 and E9X M3 shows empirically that dynos can not be trusted for anything better than about 15% absolute accuracy.

[MFL]

Originally Posted by swamp2

There is significant evidence to the contrary, trap speeds, dyno results (here) and simulation. The only evidence of this is dynos which are completely all over the map from no under rating to about 75 hp of under rating. Have a read of the (my) other thread on this topic where there is pretty much a consensus of a 5% maximum under rating, which according to EU law is not really technically an under rating...



Yes, very much so.

Another fallacy purporting that personal experience is more valuable than science... We all know what the performance numbers are and that is what matters. In fact driving the car, especially without the proper contextual understanding of the importance of radically different shaped power curves is in fact likely more confounding to the issue than enlightening.

Again, what is "THE" conclusion from dynos? The wide range of result on both the F8X M3/4 and E9X M3 shows empirically that dynos can not be trusted for anything better than about 15% absolute accuracy.

Have you driven one yet? (I Have).

[jc05e46m3]

@MFL... You ARE wrong. Here are the different variations with his mixtures. E30 and e50:



and by the way. I HAVE driven them both. Owned an E92 and take delivery of an F82 this week. What's your point?