2014 pace vs. 2004 pace, where, how are they better?

[Ogami musashi]

It was during one of those exchanges that he told me those 5g's were overestimated and that anything higher than 3,5G (sustained) was unlikely. Years later, again IIRC, i mailed a pirelli F1 engineer which told me about the same thing.

RB6 last corner, 4+ pure lateral sustained troughout with 1 hand on the f duct.

https://www.youtube.com/watch?v=CiNIDaC ... a76rouC_SA

If you read again my post, you'll see the point i convey is that displayed G-meters exaggerates G forces.

[Blanchimont]

I just had quick at the qualifying lap of Hamilton at Sochi. At the end of turn 3, he does 286km/h. With an estimated corner radius of 180m taken from Google Earth this gives a lateral acceleration of:

(286/3,6 m/s)^2 / 180 m = 35,06 m/s² = 3,57 G

Which other corners can be analysed in this way?

[siskue2005]

I just had quick at the qualifying lap of Hamilton at Sochi. At the end of turn 3, he does 286km/h. With an estimated corner radius of 180m taken from Google Earth this gives a lateral acceleration of:

(286/3,6 m/s)^2 / 180 m = 35,06 m/s² = 3,57 G

Which other corners can be analysed in this way?

Maybe pouhon?

[Juzh]

It was during one of those exchanges that he told me those 5g's were overestimated and that anything higher than 3,5G (sustained) was unlikely. Years later, again IIRC, i mailed a pirelli F1 engineer which told me about the same thing.

RB6 last corner, 4+ pure lateral sustained troughout with 1 hand on the f duct.

https://www.youtube.com/watch?v=CiNIDaC ... a76rouC_SA

If you read again my post, you'll see the point i convey is that displayed G-meters exaggerates G forces.

I did and I'm not sold. I don't see a reason why G-meters would be so inaccurate.

[Ogami musashi]

I did and I'm not sold. I don't see a reason why G-meters would be so inaccurate.

Opinion from Engineers that directly work(ed) in F1 is valuable, as well as discussion we had here about that very topic, plus my own professional experience with data measurements lead me to think that this may well be the case.
things like, localisation of the sensors, noise, sources of variability, number and sampling frequencies, clarity of display make for a lot of measurements errors.

Now you just have to bring in arguments to show i'm wrong

[Juzh]

And you think G-meters in F1 are installed by amateurs? Sorry, I don't buy this.

[mrluke]

It was during one of those exchanges that he told me those 5g's were overestimated and that anything higher than 3,5G (sustained) was unlikely. Years later, again IIRC, i mailed a pirelli F1 engineer which told me about the same thing.

RB6 last corner, 4+ pure lateral sustained troughout with 1 hand on the f duct.

https://www.youtube.com/watch?v=CiNIDaC ... a76rouC_SA

From google earth I get a radius of ~105m for the last turn cant see the speed for the RB6 but the RB9 exits at 245-250km/h

(245/3.6 m/s)^2 / 105 m = 44.11 m/s² = 4.5 G

Using the myth busters scale I would go with "plausible"

[Tommy Cookers]

....... I don't see a reason why G-meters would be so inaccurate.

what you call a G-meter is an accelerometer that is necessarily sensitive to accelerations up to maybe 100 Hz
(and vertical plane vibration anyway can and will give a response on our so-called lateral G meter)

we need to separate our 4g at below 1 Hz from relatively large accelerations at a mix of higher frequencies aka vibration

arbitrary sampling as used for instant TV replay will give large aliasing errors ie poor separation
this seems clearly visible in the footage Artur supplied - so the display of g swings in a way quite poorly related to the car behaviour
also the TV will use sample-and-hold as part of the compromised package, this makes aliasing error worse

ideal processing would show a smoother cornering g better related to the car behaviour
ideal processing wouldn't be the same for different cars or different tracks, and it won't be quick

processing could even show eg lateral road surface related vibrations at say 10 Hz unpolluted by the cornering g at 1 Hz
either of these different purposes from the same data
by matching the processing to the purpose, with some experimentation
there will be anti-aliasing elements selected according to the purpose, they won't stay the same from one test to another

and which part of the need for correction for body roll etc don't people get ?

[Artur Craft]

I've said from G meter and/or articles. For this specific one (the 6g's) it was from an article dated 2004 (that i tried to recover but couldn't) about 2004 cars being too fast, and i recall they said the cars were doing 6g's at some parts with drivers quotes.
However people tend to exaggerate numbers and we had a discussion here (i think the title was about Kart G forces) that telemetry as presented in media is not an accurate one.

I think we are pretty safe to assume that the 6Gs were not sustained, but they probably read peaks of it in the telemetry. The peaks are related to the things Tommy Cookers referred to and the actual lat-acel was less than that.

Although modifications made in 2003 mean taking it at full throttle is not quite the challenge it once was, jinking left in seventh gear at speeds in excess of 310 km/h remains a supreme test of both car and driver, with a lateral cornering force of up to 6G.

http://www.formula1.com/news/features/2 ... 11438.html

"in the ultra-fast 130R turn you get lateral acceleration forces of up to 6g" Mario Theissen, BMW Motorsport Director:

http://www.pressebox.de/presscorner/fir ... 7/iframe/1

"We have measured lateral acceleration of up to 6g there" Mario Theissen, BMW Motorsport Director

http://www.autosport.com/news/report.php/id/54851

Not that much, we discussed about the potential of a racing tyre, he basically told me that 2.5 mu coefficient would be possible but that then the reactions of the car would be sluggish. It was during one of those exchanges that he told me those 5g's were overestimated and that anything higher than 3,5G (sustained) was unlikely. Years later, again IIRC, i mailed a pirelli F1 engineer which told me about the same thing.

Even with narrow Formula Student tyres, you can reach 2.5 mu.
http://www.fsae.com/forums/showthread.p ... #post48223
If you read all of the thread, you'll see that our fellow member Jersey Tom was a part of that tire consortium

In the thread, they discuss why the actual measured lat-acel is much lower(around 1.6G) than what is measured from the sandpaper coated belt tester in CALSPAN(the 2.5 mu).

I think the Michelin engineer told you about real 2.5 mu, measured on actual tarmac, though. F1 tires are wider and far more developed so I guess they were reaching real values of 2.5 during the tire war.

Dragster tires can easily surpass that. They are very wide(450mm) and soft(lasts only 2 miles), though.
http://en.wikipedia.org/wiki/Top_Fuel

At maximum throttle and RPM, the exhaust gases escaping from a dragster's open headers produce about 900–1100 pounds (~4,5 kilonewtons) of downforce. The massive airfoil over and behind the rear wheels produces much more, peaking at around 12,000 lbf (53 kN) when the car reaches a speed of about 330 mph (530 km/h)

It takes the top fuel half a second to go from from 0 to 74mph, which translates to an average acceleration of 6,5Gs. The max downforce during that period happens, obviously, at top speed(74mph). The car have 1600lbs of aero downforce plus it's 2300lbs weight loading the tyres at 74mph, which gives a real friction coefficient of 3.8 but because there is less downforce at lower speeds, average mu is even higher than this during the 0,5s time, to allow it to accelerates 6,5Gs on average.

That's the thing. Peak friction coefficient occurs at some 500-600kg(per individual tyre) for some tyres while at as little as 200kg for others. It's entirely possible that with more weight on low speed corners, the extra load will increase mu and thus cornering speed.

Could you please provide me some articles on that? i didn't know.

I have quite some data in my old HD, which unfortunately I can't have access right now. But, just as an example:


GT tire has 1.12 mu at 900lbs, 1.08 at 1350lbs and 0.97 at 1800lbs. So, the typical load sensitivity behabiour of mu dropping with increase in load.

With a sample of one of the many NASCAR's tyres(Short Oval Stock Tire), 1.5 mu at 800lbs, 1.52 mu at 1100lbs, 1.37 at 1500lbs and more than 1.25 at 2000lbs(more slip angle needed with this load). You can see that mu was still increasing untill the 1000lbs region, where it reached it's peak, to finally drop again.

Street Corvette tire, 1.03 mu at 405lbs, 1.01 at 920lbs, 0.97 at 1450lbs and 0.95 at 1956lbs. Again typical behaviour.
ChampCar Rear Tire 360mm tread width(Road Course), 1.83 mu at 1000lbs, 1.64 at 1400lbs and 1.43 at 1800lbs. Same as above


F1 Front tire 220mm tread width, 1.6 mu at 400lbs, 1.69 at 600lbs, 1.53 at 800lbs and 1.38 at 1000lbs

Here you see that the GoodYear F1 front tire(this is from the 90s) had bigger mu at 600lbs(270kg) than at 400lbs(180kg). A 600kg F1 car with 45% weight on front tires will have 135kg on each front tire, stand still. While a 690kg F1 car with the same weight distribution will have 155kg on each.

When they corner at low speed, they got a bit of downforce and load transfer as well, this will put a bit more load into the outer tires, and the heavier car will put it closer to the load where the peak coefficient of friction occurs.

Obviously, with the Champ Car rear tire, we see a different behaviour, but that's also because the minimum load tested is already very high(1000lbs or 450kg, the max value tested with the F1 tire) while the F1 tire also have measures with 400,600 and 800lbs. I think the behaviour is the same at lighter loads and from some 400 to 700lbs, the friction coefficient likely increases for the Champ Car tire too.

It's also curious to see the effect of wider tires into cornering force. Some people said that contact patch area depends only on load and inflation pressure(due to P=F/A) but here we see that a wider tire gives far more grip even with more inflation pressure(24psi on the ChampCar and 21psi on the F1). At the same load(1000lbs), the F1 should have had more contact area as it has less pressure and that alone would point towards more grip. Add to that, the F1 tire has a slightly softer tread compound(durometer scale in the last page). So, even with supposed less contact area and a harder tread, the ChampCar 360mm rear tire have a mu of 1.83 versus F1's front 220mm 1.38

[mrluke]

It's also curious to see the effect of wider tires into cornering force. Some people said that contact patch area depends only on load and inflation pressure(due to P=F/A) but here we see that a wider tire gives far more grip even with more inflation pressure(24psi on the ChampCar and 21psi on the F1). At the same load(1000lbs), the F1 should have had more contact area as it has less pressure and that alone would point towards more grip. Add to that, the F1 tire has a slightly softer tread compound(durometer scale in the last page). So, even with supposed less contact area and a harder tread, the ChampCar 360mm rear tire have a mu of 1.83 versus F1's front 220mm 1.38

P=F/A is only true for soap bubbles, it ignores all of the structure of the tyre.

[Artur Craft]

I just had quick at the qualifying lap of Hamilton at Sochi. At the end of turn 3, he does 286km/h. With an estimated corner radius of 180m taken from Google Earth this gives a lateral acceleration of:

(286/3,6 m/s)^2 / 180 m = 35,06 m/s² = 3,57 G

Which other corners can be analysed in this way?

I also did this during that weekend as T3 is an easy corner to do that on Google Earth. My estimation was towards that 180m too(185m, actually), and that was in good agreement with FOM's telemetry showing around 260kmh of entry speed and 280kmh of exit and a 3-3.2G range showed by G-meter(in one of Friday Practices)

In my humble opinion, the G-meter values are not far off from real lat-acel values when you are carefull to filter bumps, off camber non flat surfaces, drivers riding the kerb, and look only for sustained, smooth values instead of the G-meter flicks showing peaks with all sorts of vibrations at high frequencies and etc ..... For instance, 275kmh with 3.2G gives a 182m radius


Tommy Cookers , thanks for that explanation.

mrluke, I know. I brought this because there was a major discussion once:
http://www.performancesimulations.com/w ... -pressure/

[hollus]

RB6 last corner, 4+ pure lateral sustained troughout with 1 hand on the f duct.

https://www.youtube.com/watch?v=CiNIDaC ... a76rouC_SA

Juzh, if you watch again carefully, you'll see that those 4+g are not sustained, the g-meter's number tends to stick with whichever number was the maximum (above 3g) for the last few tenths of a second and that's why you get a total of 4 readings in the whole corner, 4 max peaks (peaks including noise) is all that you see and all the 3.x values are now output to the screen. This stick at the max value behavior is over the whole video. The graphic RPM meter did that for years.

Back in 2010 when the Bulls first managed T8 at Istambul flat, I estimated 3.74G average (which probably means 4.little peak):
http://www.f1technical.net/forum/viewto ... at#p182679
Very interesting to see Reca explain it with the lack of power, and not a mythical amount of downforce, allowing them to go flat there.

[Juzh]

Back in 2010 when the Bulls first managed T8 at Istambul flat, I estimated 3.74G average (which probably means 4.little peak):
http://www.f1technical.net/forum/viewto ... at#p182679
Very interesting to see Reca explain it with the lack of power, and not a mythical amount of downforce, allowing them to go flat there.

ok, ok. i believe you. lack of power would obviously made it much easier to go flat. maybe you can calculate t3 and t9 campsa at barcelona. both flat.

[Miguel]

The problem with Montmelo is that both T3 and T9 are very strongly uphill, which I'd estimate at 6-8%. Mind you, haven't been there since 2004. Nevertheless, any strong gradient will limit available grip, and alter available G's. Another problem is that, except in very long corners like Sochi's T3, fitting a cirle is pretty far off ideal. A much better approximation is fitting a clothoid (linear increase/decrease of curvature). It may happen that, even in cars with downforce, the peak G will be achieved at the apex.

Copse would be a good candidate for high-G, but you'd need a decent model racing line there.

Also, 6G cornering is so far above any other claim that I'd seriously need to see numbers before taking it as gospel.

[Reca]

Ok... sit down and take some coffee as this is long...

For a start, a big thank to all for your nice words, I’m more than happy to share things that I have and that possibly can help the discussion.

Couple of people asked about method and accuracy, if you don’t care about this boring stuff, you can jump straight to next paragraph.

The method of audio analysis I’ve explained a few times already, there’s also a thread I made years ago you might want to search (title was something like “car speed from engine sound”), but Blanchimont made a basically correct summary, only thing to be adjusted is that from the frequency you get directly rpm (which obviously is proportional to speed, minus wheel slippage etc).
So, in reference to Ogami’s remark, it’s not a model, it’s more like a measure.
Obviously it’s affected by an error but I never bothered doing a proper estimate of it, simply because I don’t have access to an easily manageable complete sets of real data to do an accurate comparison (and anyway it’s all a work done more for personal use).
In the years though I made several, “more qualitative”, comparisons using the on screen data when available and my results are always a very good match for these. Very good match means, for example, that when, after having analyzed “blindly” the audio of a qualifying lap, I compare “my” derived speed data with the reference points (apex, end of straights etc) of FOM graphics, the difference is within 1km/h, more than good enough for me. (BTW, in favorable light conditions, like at night, the speed data shown on some drivers’ dash display are visible, and a few times I verified these match exactly the FOM graphics data, so I consider FOM graphics a reliable reference).
For older data I can’t do similar verifications, but the reference speed data available for speed traps, plus the comparison of total distance travelled, sector differences and similar, allow to keep a reasonable confidence in the results.
As for how I estimate lateral acceleration, that too I explained couple of times already when I posted similar data, repetita iuvant though (and there goes the once per year usage of Latin to justify the 5 years wast... err... happily spent enhancing my education level by studying it...), I made a program which calculates a plausible racing line, by minimizing an opportune fitness function of lateral acceleration, velocity and other parameters, using as limits the track borders from satellite image.
Again, that’s obviously not 100% accurate reproduction of the racing line followed, but it’s close enough that small divergences from the real one would hopefully only cause limited change of estimated lateral acceleration. Good enough for my needs.


End of general stuff, back to specific thread related things.

I’ve to make a small correction/addition to my previous post in relation to 2006 car not being capable of matching 2005 car at Turkey’s T8.
In doing that reasoning I just took Artur Craft mention of the 2005 car approaching 5g level and expecting 5.5 for 2006, without watching the video, so I only compared these claims with the 2006 data I had already analyzed years ago and found lower lateral acceleration.
Now I did a sound analysis of that video too and found out that the 2005 speed at the first apex of T8 was actually slower than 2006’s (and consequently lower is peak lateral acceleration):


The reasoning that for cornering of 2006 car vs 2005 the lower power was an important factor is still valid mind you, the only difference is that contrarily to what I previously said, the power available in 2006 was still enough to beat 2005 in T8, at least for the first apex.
The importance of power you can see by how much quicker the black line climbs thru the corner and basically keeps increasing at only slightly reduced rate even at the second apex (around 2850m or so), while the red line increases more slowly and stops growing for a while at the apex itself.
In term of theoretically available lateral force, going by what seen at first apex, the 2006 car should have enough to take also the second part at higher speed than 2005’s, but power doesn’t allow to, so the 2005 car can match it there.

I’m sorry for the small inconvenience that caused a not completely accurate comparison the first time, I should have checked the data from the 2005 video before.


Finally, a few direct answers to quotes, hoping I didn’t miss any direct question (if so, let me know).
These results I post come from calculations done, making/uploading images for all though would take too much time and make even longer an already very long post, so here I just post the resulting numbers, if required I can post more detailed results later.

Mario Theissen already said they recorded 6Gs in 130R(around the 2004 era) and Barrichello already said they reached 6Gs in Barcelona's T3 in 97.
[...]
For instance, Suzuka's 130R was taken at 290-295kmh in 2001/2002(before it got slightly modified), which required a 5G lateral acceleration.

The name 130R refers to the (original) geometrical radius of the centerline, racing line is obviously not the centerline, its cornering radius is lot larger, something like 210-215m or so now (the 2014’s 310+ from Rosberg mean close to 3.5g), 195-200 before the reshaping of early ‘00s.
Already “just” 5g is absurd as would need a speed higher than these cars could reach at Monza’s speed trap, let alone 6.

Similar reasoning for Barrichello’s claim of 6g at Catalunya T3 in 1997.
Both MS in 2004 and Webber in 2010 reached there a peak acceleration of roughly 4g at about 250km/h, full throttle for Webber, MS slowing down a bit from the 260-265 reached shortly before the apex (apex of T3 is relatively close to T2 exit, not leaving much room to accelerate). 6g would require speed at the apex in excess of 300km/h. Hard to believe, to say the least.

I just had quick at the qualifying lap of Hamilton at Sochi. At the end of turn 3, he does 286km/h. With an estimated corner radius of 180m taken from Google Earth this gives a lateral acceleration of:

(286/3,6 m/s)^2 / 180 m = 35,06 m/s² = 3,57 G

I too get 3.2-3.3g basically constant for the whole corner. It’s in absolute the highest lateral acceleration on that track, but a not much lower level of roughly 3g is seen in other parts of lap, at some 160km/h already, like in turns 7 and 8.

It was during one of those exchanges that he told me those 5g's were overestimated and that anything higher than 3,5G (sustained) was unlikely. Years later, again IIRC, i mailed a pirelli F1 engineer which told me about the same thing.

RB6 last corner, 4+ pure lateral sustained troughout with 1 hand on the f duct.

https://www.youtube.com/watch?v=CiNIDaC ... a76rouC_SA

From google earth I get a radius of ~105m for the last turn cant see the speed for the RB6 but the RB9 exits at 245-250km/h

(245/3.6 m/s)^2 / 105 m = 44.11 m/s² = 4.5 G

Using the myth busters scale I would go with "plausible"

Also in my analysis of the 2013 pole the lateral acceleration in various instances thru the lap touches 4 and bit more.

4 and above is/was definitively feasible, I’ve seen it plenty of times in my estimates. Actually this year too, with all limitations of higher weight, power not necessarily available all time, limited aero rules, level in the range 3.5-4 is reachable.

Way more suspicious are the sometimes said “easily over 5”, let alone 6, especially when people mention that level in reference to the “old” era of ‘90s.

The more back in time we go then, the more we should be wary of the “fisherman effect”... be it for lateral acceleration, downforce, power or any other figure, especially as less “sure” data were available back then so the reported data are based on arbitrary recollection and as such tend to be enhanced/misremembered (accidentally or not...)

It’s like the story of the BMW 4-cyl turbo engine, it’s been out of duty since decades yet it keeps gaining power as time goes by... any day now we will probably celebrate it breaking the 2000hp barrier.