2014 engine power from onboard videos

[Tommy Cookers]

efficiency based on crankshaft energy and MGU-H/K energy (minus braking recovery) might be fair ?
on road or other typical fuel of course
would they then still be claiming 40% ?

the Wright engine bsfc of 0.325 lb/hp-hr, if corrected for lower mass-specific energy of 115/145 Avgas might still be a landmark
this is fair because it could have run this way on road fuel
but this was at 30000', which gave more recovery due to the low ambient pressure
at sea level it made 0.38 lb/hp-hr on Avgas, say about 0.37 lb corrected for road fuel

[idfx]

Some information may be useful:
http://www.mclaren.com/formula1/live/
source: http://www.mclaren.com
I never tested it.
Another way to have telemetry: access the ATLAS (Advanced Telemetry Linked Acquisition System).

[Cold Fussion]

A 40% efficient ICE would be the in the region of 690hp, when adding the the 160hp from the mgu-k you'll have 850 hp. There is no way they are running 850 hp.

Would 40% efficiency of the total power unit, ICE and ERS, be seen as a huge milestone and technological leap?

Compared to a normal petrol road car, 40% BTE is still a fairly significant step forward.Whether or not the mgu-h tech is applicable to a road car is another question though. The high percentage of WOT time, coupled with the fuel flow limit are the main drivers of the mgu-h, so it's probably more a step forward purely because of what the rules require, rather than any real leap in technology.

[chip engineer]

A 40% efficient ICE would be the in the region of 690hp, when adding the the 160hp from the mgu-k you'll have 850 hp. There is no way they are running 850 hp.

Would 40% efficiency of the total power unit, ICE and ERS, be seen as a huge milestone and technological leap?

Compared to a normal petrol road car, 40% BTE is still a fairly significant step forward.Whether or not the mgu-h tech is applicable to a road car is another question though. The high percentage of WOT time, coupled with the fuel flow limit are the main drivers of the mgu-h, so it's probably more a step forward purely because of what the rules require, rather than any real leap in technology.

Please explain why you think WOT and fuel flow limit is needed for MGU-H technology to be useful.
It seems to me that if cost issues can be overcome, MGU-H technology should be very applicable to road cars for fuel efficiency and eliminating turbo lag. Even if near WOT is necessary for best efficiency, hybrids can operate the ICE that way.

[johnsonwax]

Please explain why you think WOT and fuel flow limit is needed for MGU-H technology to be useful.

To be useful? No. But engineers work designs from their constraints (even artificial ones), and if fuel flow is a constraint then they're going to look for solutions that can overcome the limits imposed by that constraint - how can they get more than 100kg/h of power in order to overtake, etc. Taking the constraint off doesn't necessarily make the technology useless, but without the constraint the technology likely would never have been developed to begin with.

[Tommy Cookers]

....... Please explain why you think WOT and fuel flow limit is needed for MGU-H technology to be useful.
It seems to me that if cost issues can be overcome, MGU-H technology should be very applicable to road cars for fuel efficiency and eliminating turbo lag. Even if near WOT is necessary for best efficiency, hybrids can operate the ICE that way.

at takeoff power (supercharged to 2.2 bar abs) the Wright Turbo-Compound added by recovery 18% to crankshaft power
at slow cruise (supercharged to 1.17 bar abs, lean mix, and coarse pitch, WOT or almost) it added 6% of a lower crankshaft power
so any significant throttling would further reduce the recovery ?

and much of the value of recovery is in downsizing the ICE beyond the downsizing from the turbocharging
which means at low rpm more boost will be needed in the compounded engine than in the turbo only engine
the compounded engine will have a greater region of rpm where full boost cannot be used without retarding the spark (inefficient)
because the full boost will need to be higher (because the engine is smaller)

hybrids and others eg Fiat vary their valve motion to reduce cylinder filling (to reduce power/torque but keeping WOT)
the reduced filling gives much less pressure to the exhaust pulses, so there is less % energy available to be recovered

VVT and MGU-H compounding are rather in conflict, that's why the rules don't allow VVT
it costs almost nothing to motor the MGU-H, and ingeniously this motoring makes VVT unnecessary

[Brian Coat]

What a cracking thread.

Just thinking ...

With enough data it should be possible to extract power unit inertia effects and the different drag components e.g. by examining acceleration in different gears and at different road speeds, track sections, gradients, ...

However, transcribing such data from TV coverage is laborious and only yields a limited number of data points.

Since the gearing is known (see other posts), could we get more acceleration data by using on-board *audio* data?

The extraction of rpm data could be automated, allowing much more data to be available.

This might allow us to the statistical confidence to calculate more about the power unit performance.

Before anyone says "go on then" I must add I am not a student or unemployed!!

[Brian Coat]

I had a go at extracting rpm from Youtube audio as a sort of 'proof of concept'.

As you'd expect, it works.

The firing order peak is very pronounced on the fft plot so it would be relatively straightforward to track this peak and thereby extract rpm, rate of change of rpm and gear number and (using known gearing), vehicle speed.

I don't know how to post images on here so you'll have to believe me when I say 1330 Hz is the biggest peak on the graph.

This sort of method might allow large amounts of data to be analysed.

Using the method demonstrated by the OP, it might be possible to see (for example) how P_acc varies by gear (P/T inertia effect) and therefore adjust power estimates for this effect.

I wonder if any teams are playing with the audio data?


*I used the idle and max rpm sections on the well known Cosworth CA 20K dyno run video. The 4th order peak aligned exactly with the dyno display (I should hope so!).

[Brian Coat]

Thanks to Aleks for showing me how to post the images

This is the video and the FFT of the section where it is at full chat

http://m.youtube.com/watch?v=BPdm51QwZEw&feature=kp

http://imgur.com/GBf9XZM

The cursor is on the firing order peak @ 1330 Hz.

20000 rpm is 1333 so it's just about spot on.

A tracking FFT software could produce a 'tachograpgh' of any in car video.

The 660ish peak is interesting : is the mic picking up bank firing frequency because it is to one side of the cell or is this something else?

Now we just need a student who's bored enough with "Countdown" to code it up ...

[Brian Coat]

Sorry ...

I now realise that the audio rpm thing was covered in this forum in 2006!

No wonder I am talking to myself on this.

Some of you F1 Technical diehards have seen all this before.

http://www.f1technical.net/forum/viewtopic.php?t=2218

What has changed since then of course is the fuel flow rate limit and better in car telemetry on tv, which together allow the interesting analysis such as that by the OP.

[hollus]

You are not talking to yourself. We are listening. And thanks for posting, even if yes, it is second time around. Very few might have read that 2006 thread, so your input is helpful and might inspire someone to do just that. It is just that in the technical corners of this forum, if people have nothing to add, they will just be silently satisfied with what you said. And since it was well said...
I am looking forward to someone really picking up the ball in this matter.

[Brian Coat]

Well I could not resist downloading that sonic imager thingy they mentioned in 2006 and then analysing some 2014
data. It is freeware by QMC, London (told you there'd be students required!).

Heres the kind of thing you get instantly (I've got it in full-screen mode here so you cant see the numbers etc.):

http://imgur.com/bftehwB

This is F1 2014 Monaco GP - Hamilton vs Rosberg Onboard

The cursor is at 27.5s where he gets on the throttle in 2nd gear. You can see the rate of rpm rise inflection very clearly.

Its great fun, you can easily see all the rpms, rate of change of rpm, in all the gears (hence easy to estimate P_acc)

You can even slow down the playback on the s/w so you can hear exactly where each gearchange/lift/whatever.

You think a 2014 car sounds grotty - try it at 10% playback!

A floating cursor gives you all the exact frequencies, times etc.

I bet you can save the tracked orders if you know what you are doing.

[Brian Coat]

I have now figured out how to get this s/w to isolate the dominant order and export the graph.

I've found some freeware auto graph digitizing s/w and am trying to extract an rpm data set.

I fear that getting a decent acceleration profile will be tricky - double differentiation of logged data often = big scatter.

I'll see if I can get it into Excel and play with the data.

Compared to other data (say an Adcole-digitized cam profile) the acceleration does not change quickly so maybe we'll be able to smooth it without corrupting the results.

[andylaurence]

I have now figured out how to get this s/w to isolate the dominant order and export the graph.

I've found some freeware auto graph digitizing s/w and am trying to extract an rpm data set.

I fear that getting a decent acceleration profile will be tricky - double differentiation of logged data often = big scatter.

I'll see if I can get it into Excel and play with the data.

Compared to other data (say an Adcole-digitized cam profile) the acceleration does not change quickly so maybe we'll be able to smooth it without corrupting the results.

Brilliant stuff! A walk-through of your process to turn the audio into usable data would be a great resource.

[Brian Coat]

OK - so what I've learned is ...

1) There are various web sites where you can turn Youtube into MP3 audio file.

(Obvious disclaimer: do not do this with *music* without checking for risk of digital rights infringement).

So you can easily get an MP3 audio file from dyno or in-car (on track NOK due to Doppler effect).

2) Dr. Matthias Mauch and Dixon at QMC, London (Centre for Digital Music) have developed A FUNDAMENTAL FREQUENCY ESTIMATOR USING PROBABILISTIC THRESHOLD DISTRIBUTION

It is called Probabalistic YIN or PYIN and there is a GNUware called TONY

Seems like clever stuff ...

http://matthiasmauch.de/_pdf/mauch_pyin_2014.pdf

You get TONY here:

https://code.soundsoftware.ac.uk/projects/tony/files

You just load in the MP3 and ... well play about a bit and you'll see.

I'm not certain its exactly what we require but it looks promising.

---------------------------------------------------------------------------------------------------------------------------------------------

The fundamental frequency it estimates is not always the engine order you expect and it can leave the odd gap (e.g. where the sound is weak) or throw in the odd octave "jump" but its pretty good.

I tested it on the Cosworth 20,000 rpm dyno engine video and it worked.

http://imgur.com/32pW2Fv

It will export the frequency/time data (called "pitch track data" - musical innit?) to a .csv file which is readable by Excel.

The data is real and therefore lumpy but this can be smoothed using the normal techniques, I guess.

The Hamilton at Monza soundtrack also works but I'm struggling to get the data to line up with the on-screen display data for rpm. Some of this could just be a time-shift due to the algorithm or lag in the telemetry display - but I have not really looked into it.

The jumps on this one are not Octave jumps by the algorithm they are gearshifts ...

http://imgur.com/bl33xHT