2017 F1 gearing from onboard videos

[hollus]

First contact with 2017 gearing, all from Melbourne.

First a reminder of where the data is coming from. I pick pairs of rpm / speed data from onboard videos and plot them in excel, where I make linear fits (forced to go through 0,0) and remove outliers based on the correlation coefficient.
I try to gather data only under acceleration and far from kerbs, gearshifts, etc to reduce artifacts and wheelspin as much as possible. This means that data for 5th - 8th gear is very good, but data for 3rd - 4th gear not so good (wheelspin) and for 1st and 2nd gear I often have to compromise and use what I can.
This is an example, from Vettel in Melbourne.



A few things to note: Data for 6th is far from perfect, the nature of the corners in Melbourne.
4th is split in two sections. The data at faster speeds is from actual racing. The data at lower speeds is from the victory lap. In the first case, he was at full throttle or close to full, in the second case, he was barely applying any throttle. And the difference shows. I don't think that is due to wheelspin (after making many of these, you learn to recognize wheelspin), but more an artifact of how the speed and rpms are averaged/smoothed/interpolated in the software. The data might improve if get more points, but for now this is what it is. That 4th gear would have come out a bit shorter without the parade lap.
Why do I spend so much time explaining that? All data for 2nd and 1st gear are from that parade lap. 2nd is also abnormally close to 3rd. Obviously, 2nd gear would come much shorter from racing data (but they don't use 2nd gear in Melbourne!).
Also good to keep in mind that these are not absolute or perfect numbers, but hopefully they are consistent numbers that allow to compare the different teams.

Now to the data I have processed so far. 8 gears for Ferrari, only 3rd - 8th gear for Mercedes, for 2017 an compared to 2016 (in gray). The artifact with Vettel's 2nd is showing (his 3rd might also have a sniff of wheelspin).



What I see: Merceces and Ferrari have very similar 7th and 8th gears. Then Ferrari has shorter 6th and 5th, they are similar in 4th and Ferrari has a longer 3rd. The result of this is that Ferrari has much better overlap from 3rd to 8th at any speed, while Mercedes has a poorer overlap from 3rd to 5th. I can only speculate on why, and that's what I am going to do:
The trace for 5th gear from Hamilton was very clean. Also the trace for 3rd gear, considering what is normal for a 3rd gear. What I mean is that most data points fell nicely in a straight line from the get go. In contrast, Hamilton's 4th gear was a disaster. Points well above the linear trend and points well below. Lots of wheelspin, even at 180 km/h, unless it was also data artifacts (Hamilton did a lot of kerb riding in 4th).
What I think is going on is that for Mercedes, the limit of wheelspin falls in 4th gear, which is higher than I expected. So they let the driver use his foot in 4th to, well..., drive. I can hear the mechanics saying "aren't you the best in the world with your right foot? Then show it, we won't help there".
In exchange, they get relatively optimized transitions for 5th to 8th with some rpm overlap, which should allow to select the optimal rpm for power or fuel usage without any worries about wheelspin. Below 4th, that shortish 3rd makes me expect a short 2nd and a short 1st, to avoid the trap of the riduculously long 1st gear from last year which likely compromised their starts. Of course, Ferrari's 4th might be a bit shorter, which would make it similar to Mercedes´, but they would still have much better overlap to 5th and 3rd.

The other thing to notice is the change to shorter 8th gears compared to 2017, which to a point drags into 7th and also into 6th. This was expected as the cars have lower top speeds. Still, there should be no issue going to 340km/h and even 350, so we should not see any overtaking attempts stopped by short gearing, at least not from these two teams.
And lastly, the higher overlap means that the exact choice of gears should be even less important this year than last year, which makes Mercedes choices around 4th gear a bit striking.

[hollus]

Added McLaren:


The 8th gear is significantly shorter compared to Mercedes and Ferrari. It might look like a lot in the graph, but we are talking about some 5km/h differences here, so it is not world changing.
The big question is whether it is shorter because they do not expect to reach the same power as the other two teams or whether it is a patch for their resonance problem.
I am inclined to believe more in the second. Their 7th, 6th and 5th gears are also significantly shorter, it is first by 3rd gear that they fall in line, and I am sure that McLaren are expecting to have similar cornering speeds to the other teams.
Something I learnt looking at the onboards from McLaren: they are shifting up at impossibly high revs so that the car is almost never below 11000 rpm, even not after the upshift. It is only out of part throttle sections that they are at 10xxx. This goes some way to explain the lack of power and, in some way, fits with shorter gears. By shortening your gears you get to be at higher revs for the same speed. This makes intuitive sense, although I am not sure it makes sense if one thinks analytically.
Certainly they have the best overlap in the middle gears, again not sure what this means or even if it means much at all.
In any case, teams are allowed one change of ratios this season, so I was kind of expecting McLaren to have compromised their first set to cure whatever illness the engine has right now, and to introduce the second set once the vibrations are gone. There is nothing utterly bizarre there, though. Time will tell.

[mrluke]

Are you able to post a table with your calculated gear ratios per team?

Appreciate that some will be based on limited data

[hollus]

These are the ratios for the data I have processed so far:

Mercedes (3rd to 8th): ----- -----  9.812 7.965 6.572 5.645 5.072 4.472
Ferrari (1st to 8th): 13.641 10.342 9.607 7.909 6.739 5.762 5.043 4.469
McLaren (2nd to 8th): -----  12.367 9.667 8.068 6.900 5.882 5.118 4.519

I am assuming that the diameter of the wheel has not changed and is still 660mm, is that right?

I intend to process the other teams eventually, but it will be slow, specially seeing that probably there is nothing earth shattering to be found. For once the rules did what they were supposed to do and gearing is a small performance differentiator.

[henry]

These are the ratios for the data I have processed so far:

Mercedes (3rd to 8th): ----- -----  9.812 7.965 6.572 5.645 5.072 4.472
Ferrari (1st to 8th): 13.641 10.342 9.607 7.909 6.739 5.762 5.043 4.469
McLaren (2nd to 8th): -----  12.367 9.667 8.068 6.900 5.882 5.118 4.519

I am assuming that the diameter of the wheel has not changed and is still 660mm, is that right?

I intend to process the other teams eventually, but it will be slow, specially seeing that probably there is nothing earth shattering to be found. For once the rules did what they were supposed to do and gearing is a small performance differentiator.

The diameter increased 10mm to 670. 680 for rain tyres.

[Nuvolari]

Added McLaren:
http://i.imgur.com/rZvTECZ.png

The 8th gear is significantly shorter compared to Mercedes and Ferrari. It might look like a lot in the graph, but we are talking about some 5km/h differences here, so it is not world changing.
The big question is whether it is shorter because they do not expect to reach the same power as the other two teams or whether it is a patch for their resonance problem.
I am inclined to believe more in the second. Their 7th, 6th and 5th gears are also significantly shorter, it is first by 3rd gear that they fall in line, and I am sure that McLaren are expecting to have similar cornering speeds to the other teams.
Something I learnt looking at the onboards from McLaren: they are shifting up at impossibly high revs so that the car is almost never below 11000 rpm, even not after the upshift. It is only out of part throttle sections that they are at 10xxx. This goes some way to explain the lack of power and, in some way, fits with shorter gears. By shortening your gears you get to be at higher revs for the same speed. This makes intuitive sense, although I am not sure it makes sense if one thinks analytically.
Certainly they have the best overlap in the middle gears, again not sure what this means or even if it means much at all.
In any case, teams are allowed one change of ratios this season, so I was kind of expecting McLaren to have compromised their first set to cure whatever illness the engine has right now, and to introduce the second set once the vibrations are gone. There is nothing utterly bizarre there, though. Time will tell.

Thanks for this info. I posted this at another thread, but it looks like what I found fits your idea of a compromised 5th, 6th, 7th and 8th gearing as a patch for the resonance issues.

This is a plot of acceleration along the 1.2 km straight at China.



The data that supports your idea is that the Renault only starts to out-accelerate the McLaren after 230 km/h...which roughly coincides with the start of 6th gear according to my data.

[hollus]

A small update after Ferrari changed the ratios in Austria (at least for Vettel's car). This is with (limited) data from Hungaroring and Spa. The changes are so small that I will not bother with graphs.

I am getting the new 7th and 8th gears 0.3% shorter, with 5th gear 0.1% shorter in my data. This is likely within the error of the method, plus it could come from internal recalibrations etc. My guess is that those did not change at all, why would you change your top speed by 1km/h? In any case, if they changed, it is probably irrelevant.
6th gear is coming 1% shorter in the data after the change. This is most likely real, but just a small optimization, we are talking about 2.5km/h difference at the same RPM. Probably just a minor adjustment.
Note: everything below is contaminated with wheelspin to some degree.
4th and 2nd gear also appear sorter in the new data, by approx. 2.5% and 3.5%. A massive caveat, though: Most of 2nd and 4th came, in my old set, from a parade lap, which would naturally tend to produce longer gears in my graphs (zero chance of wheelslip unlike when racing), so chances are they are unchanged in reality.
For 3rd gear I get 2.5% longer, but since 3rd gear data tends to suck, this is again, likely, within the margin of error of my method.
And I did not collect data for 1st gear.

My guess, there is a fair chance that nothing changed at all, or that only 1st gear changed (Toro Rosso did this in 2014). 1st gear matters essentially for the starts, so...
Other gear ratios appear unchanged or minimally changed.

No gear drama in 2017, bah!

[hollus]

@Mark4211 posted a video in the Ferrari SF70H thread that gives me 1st gear data plus 2nd and 4th gear data from a formation lap, similar to the old parade lap for Vettel before the gear ratio change. So those values should be comparable now.

After adding the new data, 4th gear has "changed" to be almost identical (99.8%) of the old value. Within error, one has to assume that it is indeed identical. Thus, all gears from 4th to 8th are, likely, not changed. The change compared to a value obtained in race conditions is a reminder that these cars spin the wheels all the way into 5th gear!

No good data for 3rd gear here either.

Second gear remains as it appeared, 3.4% shorter than at the beginning of the season.

1st gear, with a surprisingly nice linear graph, returns 5.0% shorter than at the beginning of the season, that is, the car travels 5% more slowly for the same RPMs.

So here is my best guess at Ferrari's new gear set:

1st has been shortened by 5%. This is mostly a start gear, although it comes into use in races occasionally.
2nd has been made about 3% shorter as a transition.
3rd is then likely as the old gear (within the measurement error anyway), but might have been made a bit longer.
4th to 8th appear to be identical, with the possible exception of a 1% adjustment in 6th gear.

So the change seems to be focused in 1st and 2nd only, and maybe 3rd. I refer everyone to the opening post to see what this would mean in graphic terms, but the spacing between 2nd and 3rd makes a lot more sense after shortening 2nd and, maybe, lengthening 3rd, even if not by very much.