2012 pre-season testing – Patterns in the Chaos…

Posted: March 6, 2012 in Formula 1, Pre-season tests, Teams
Tags: , , , ,

Hello guys (and girls). I have a problem. I don’t know how to start this post. So, I think I’ll just get right down to it.

I was a little bit disappointed with my last blog entry (the round-up), because it was too argumentative and not factual enough. The 2012 pre-season testing has, arguably, been one of the most difficult to understand and decipher in the history of the sport. The regulations have apparently closed the gaps between the teams, to a point where it’s difficult to say if Ferrari will be faster than, say, Force India. This ambiguity led many respected journalists to desperate measures, such as Andrew Benson’s decision to take all lap times, from all testing sessions, from both tracks (!), add them up, and produce an average as a means of comparison. I think F1 fans deserve a little bit more than that. So, immediately after I did my last blog entry, I decided to sit down and try to crunch some numbers, trying to find patterns in the chaos (don’t you just love it when the title of the post is mentioned in the post itself?), and see if I can make some sense of what we have seen in testing over the last few weeks. You should, therefore, brace yourselves for a very long post, filled with stats, assumptions and calculations. If you are here just to confirm that Red Bull is fastest or find out what the pecking order is, I suggest you continue reading no further.

So, our testing analysis, first of all, will focus only on the Barcelona tests that took place between 21/02 – 24/02 and 01/03 – 04/03. I think that we must exclude Jerez from the testing picture because (a) it adds a variable for which we simply don’t have enough data to factor in (b) not all teams were present with their 2012 cars and (c) most teams used Jerez as a springboard to get a baseline for their cars, so most laptimes are completely inconclusive. From the various lap times that the teams were performing, it was clear that the teams were following four (4) different testing scenarios:

1. Back to back comparison runs, with heavily fueled cars. I will refer to that in my analysis as “heavy fuel stints”. The stints that I have decided to include in this category and analyze were those that consisted of at least five (5) consecutive lap times, with the first 2-3 laps above the 01:28.000 mark.

2. Back to back comparison runs, usually in shorter stints, with reasonably fueled (but not very heavy cars) – I presume around the half tank mark. I will refer to that in my analysis as “medium fuel stints”. Again, the stints that I have decided to consider as “medium fuel stints” for analysis, were those that consisted of at least five (5) consecutive lap times, with the first 2-3 laps within the 01:24.000 – 01:27.999 bracket.

3. Low fuel runs, usually in very short stints. Some teams (like Red Bull, McLaren and Mercedes) were running a bit more heavy and some were going for more accurate qualifying simulations. I will refer to them in my analysis as “low fuel runs”. In order for a stint to qualify as low fuel and be part of the analysis, there is no minimum limit on the laps / stint, but there is an upper limit of 01:23.999 in terms of lap time.

4. Finally, the last testing scenario that I was able to make out was, of course, race simulations. The term “race simulation” is by itself a bit contentious, since it doesn’t mean the same to every team on the grid. Anyhow, we will see that later on.

To give you an idea, you can see in the picture below (a snapshot from my working file) a typical testing day, and how I have marked the various stints. The low fueled stints are marked orange, the medium fueled stints are marked yellow, the heavy fuel stints are light green and, finally, the race simulations are dark green.

Snapshot of 24/02 Barcelona 2012 pre-season testing

By breaking the testing into these 4 categories, it’s easier to understand how each team compares to each other, and it’s also safer (but definitely not safe) to draw conclusions. We are trying to compare apples to apples and oranges to oranges, and we managed to get some pretty interesting results. I would like to start by examining the low fuel runs, and see what we get:

You can go to any F1 website and it will give you the fastest lap times from testing and an unofficial classification. Do that, and you will find out, for example, that Lotus registered the fastest lap time from all teams with a 01:22.030 and that Mercedes were the slowest with 01:22.932, i.e. 3 tenths off Caterham’s pace. One must, however, take into consideration the amount of fuel that each team is carrying. This is an information that we don’t have. What we do have, however, is the amount of laps that they did during these low fuel runs. By removing the fuel penalty (again, I am using the 0.150 sec / lap figure) that you have to carry in order to complete all the other laps of the stint (apart from the out lap / in lap and the timed lap), we are getting some different results. You can see for example, in the table below, an example, where Red Bull’s fastest lap time was 01:22.662, but fuel corrected on an other stint they were actually able to go faster than that, with a 01:22.157. I don’t think it’s necessary to remind you that Red Bull were still probably carrying quite some fuel onboard, something that we cannot be positively claim for the smaller, midfield teams.

Red Bull low fuel runs, fuel corrected

By applying the above logic to all the low fuel runs of all the teams from all testing sessions, we are getting the results of the following table. I have posted the standard classification to the left (i.e. the fastest lap times as recorded), and next to it I have posted the fastest lap times, after the low fuel stints were fuel corrected.

Fastest laps classification, fuel-corrected

So, there you have it. As you can see, the order changes dramatically. Does this mean that the new order is correct? No, it’s not – but it’s closer to reality. It all has to do with how much fuel was left in the tanks when the cars were returning in the garage, and we don’t have that information available. It’s safe to presume that teams like Sauber and Williams had very little fuel, and those 1:21’s are close enough to a very legitimate qualifying performance. On the other hand, we can safely say that Red Bull and McLaren apparently have a considerable amount of fuel left in the tanks, and that Mercedes are looking good, based on the above table; in fact, they look quite fast. I also assume Caterham must have attempted legitimate qualifying simulations, so the gap they have to Sauber and Williams is not very small. Unless they had at least 6-7 laps of fuel onboard, they seem to not be exactly where they wanted to be. As you understand, these are all big assumptions and we can not base any meaningful conclusions on this table alone, so let’s move to the next part of our analysis: heavy fuel stints.

As I have described to you above, in order to consider heavy fuel stints as part of my analysis they had to be at least 5-laps long and they had to be above a lap time benchmark. In this way, I was able to shed a lot of scattered and inconsistent data; noise, if you prefer. I then worked on the assumption that teams would be testing their cars with comparable (high) fuel loads, and that averaged laptimes between those stints should be a good indication of where we stand. I am afraid I am repeating myself, but I have to stress that it’s only natural that bigger teams will be testing bigger fuel loads and doing all sorts of things to blur the waters, such as doing slow and fast laps in the same stint, slowing down in a specific track sector, not operating DRS or KERS, etc. I have therefore tried not only to produce average laptimes, but I also calculated for each team the average degradation that they suffered in all their heavy stints, from all Barcelona testing sessions and I have excluded the peak times (which can be a result of a driver lifting off, facing traffic, or simply making a mistake).

Surprisingly, Lotus didn’t do any heavy fuel running at all, since they focused on shorter stints, medium fuelled and race simulations. Very surprising. We therefore have no data for them, but we have 9 other teams which are presented in the following figures:

Heavy fuel stints by Red Bull, McLaren, Ferrari and Mercedes

Heavy fuel stints by Force India, Toro Rosso, Sauber, Williams and Caterham

First of all, it’s clear to the naked eye that Ferrari’s runs don’t belong in the first group (top teams) but rather to the 2nd. Visually one can also say that the team that stands out from the 2nd group seems to be Force India, with Caterham being, again, visually slower than the other teams in the group. Let’s take a closer look then at all the numbers involved. In the following table, I have listed the pecking order, as a result of averaging the laptimes that were performed, using the above mentioned criteria. Next to this table, there’s the classification based on the calculated degradation (average), but factoring in all the heavy fuel laps, of all the stints, from all the days of testing and for all teams.

Classification based on (a) heavy fuel running lap times and (b) degradation

As you can see, McLaren was the fastest team of all and, on top of that, enjoyed a very mild degradation of 0.169 seconds per lap. Mercedes were a very close second (less than a tenth), but their degradation was the worst of the lot (0.238 seconds per lap), however such levels of degradation are not cause for concern, especially since we noticed that the Mercedes drivers were pushing more at the start of the stints, taking more life out of the tyres, in comparison to McLaren or Red Bull. McLaren’s fastest lap times are complimented by the fact that they did 11.5 laps per stint, the 2nd biggest tally after Caterham. Mercedes (with 10.8) and Red Bull (with 10.4) were close behind. Red Bull was only 4th fastest, but their degradation was insignificant (easily the best of the lot, at 0.121 seconds), indicating that the drivers weren’t pushing and there’s much more performance to come from this side. Force India’s good results, both in terms of lap times (3rd fastest, 4 tenths off McLaren’s pace) and in terms of degradation support the visual observation that they are the best of the midfield in this comparison. Ferrari’s results are somewhat disappointing. Their average lap times are only good enough for 6th, they suffered the 2nd highest degradation of all (0.227 seconds per lap) and they did the least amount of laps per stint (only 9.4). These results also confirm our visual observations that they should have been graphed with the 2nd group. One last observation is Toro Rosso’s seemingly disappointing performance with regards to lap times. I remain sceptic as to whether this is the real picture or if they have been hiding the car’s pace (more than others, that is), especially considering their decent results in the low fuel runs above.

Having said that, there’s always a degree of uncertainty in those figures; a considerable one, for that matter. Yet I believe that although the above figures and tables don’t represent the truth, they are a good indication; they are, let’s say, a hint of where things stand at the moment. But still, it’s not enough. It’s time to move to the next stage of our analysis, the medium fuel stints. In the following figures, I am doing a graphical representation of the various stints. Before we get there though, a very surprising observation: Red Bull have decided to do absolutely no running on medium fuel levels (apart of course during the race simulations). That’s a completely bizarre decision and it makes us think that, apparently, they are trying to hide the car’s true pace. From all the lap times that I have amassed, it’s clear that Red Bull have only been focusing on very heavy fuel running (as we saw above) and on race simulations, which are also a bit “bizarre”, as we will examine later on.

Medium fuel stints by McLaren, Mercedes, Ferrari and Lotus

Medium fuel stints by Force India, Toro Rosso, Sauber, Williams and Caterham

Before we go into the analysis of the medium fuel stints, it’s worth saying that medium fuel stints are less accurate than heavy fuel stints, because there is more data scatter, so they are considerably less conclusive. The reason is that a medium fuel stint can start at 1:24.100 and end at 1:27.800, or it can start at 1:27.000 and end at 1:30.000, you therefore have more than 6 seconds of data scatter between laptimes. On the contrary, in a heavy fuel stint you start pretty heavy (at 01:28.xxx) and end up in the 1:32’s, so there’s less scope for scatter – it’s much more accurate. On that disclaimer, let’s look at the results:

Classification based on (a) heavy fuel running lap times and (b) degradation (medium stints)

It’s immediately clear that the picture is less intuitive than before. Ferrari remain solidly in the middle of the pack (5th), however they manage more laps / stint than anybody else (9.1). As you can see the degradation figures are higher, and that’s because on a lighter car, the drivers tend to push more. Also, many of the performance tests and setup evaluations are done using half tanks (medium fuel), and that’s when the drivers go for it. McLaren and Mercedes continue to look reasonable, as does Force India. Toro Rosso seem to be going very fast, but they are paying the price in tyre degradation (0.446 seconds per lap), as opposed to Force India who are both fast and consistant. Lotus is behind Ferrari here, and look relatively mediocre. However, it’s important to note that, in general, McLaren, Lotus, Force India and Caterham did must fewer medium fuel laps in comparison to the rest, so the results are a bit skewed. All in all, I think it’s better to leave the medium fuel stints and go to the final part of our analysis, the race simulations.

All teams bar Caterham attempted race simulations in Barcelona, but the way the teams approach a race simulation can vary dramatically. Others choose to stretch the legs of their cars and test the performance whereas others (the most) tend to hide stuff and prefer to test mainly reliability, consistency and procedures, such as pitstops. It makes sense. A race simulation is very easy to be read and analyzed by opponents. You know the fuel loads (you start filled to the brim and work your way through it) so you can calculate the pace, the degradation, the strategies, everything. Which is why most teams prefer to hide as many things as possible during race simulations, and usually do them to get their reliability, systems and consistency checked. I would therefore pay the least importance on the race simulation figures (it’s clear to me that the most important testing scenarios are on that order: (a) heavy fuel stints (b) low fuen stints (c) medium fuel stints and (d) race sims), but for what it’s worth I am posting the race sim results below:

Race simulations

A brief explanation on how I compiled the above numbers: Some teams did a bit less than 66 laps (theoretical Barcelona distance) and some a bit more, so I normalized all race simulations to reflect a 66 laps race, using the average lap times. Also,Β in the final race time I included the pitstops. As you may recall from our previous blogs, the average time penalty for a pitstop in Barcelona is 20 seconds, so adding all the pitstops together we finally get the total race time. As you can already see, there are some oddities. First, Williams 24/02 race simulation is the 2nd fastest recorded, whereas their 03/03 simulation is the slowest of all. McLaren, despite for their excellent tyre degradation, seem to be lacking in pace, etc. Mercedes’ race simulation is a very good one, and should provide hope to the Merc fans, but I have to stress that these race simulations is the least reliable set of data we have in our disposal.

Which brings us to the conlcusion of this article. As I said, I was disappointed a bit with my previous blog entry, but I now think we have some good data to base our “conclusions”. First, and foremost, all the top teams have excelled in one part of our analysis or the other. Mercedes have been very quick in low fuel runs and also 2nd quickest in heavy fuel running, not to mention their impressive race simulation. McLaren have been the fastest in heavy fuel running and have exhibited admirably low levels of degradation throughout. Red Bull have been relatively fast in heavy stints with the tiniest of degradations under all conditions. Even Force India have shown flashes of brilliance, with good and consistant speed and no degradation issues.

Ferrari though is different. They have been strikingly mediocre, failing to shine in any of the testing scenarios we examined, and particularly the important ones (i.e. heavy and low fuel stints). Their heavy fuel pace is unspectacular and their tyre degradation at the same time is a bit poor. Their quickest lap times in low fuel mode were 2nd to last, and they have never been able to escape the middle position of the tables. If they are sandbagging, they are doing a hell of job. A lot of the people have suggested that the figures aren’t bad for Ferrari and that we are basing our negative reviews on heresay and rumours. The data, though, seem to suggest that Ferrari are behind at the moment, firmly in the midfield.

Red Bull are fast, but are clearly hiding the majority of their car’s true pace – they should be bloody hard to beat in Australia. McLaren are very fast in all situations and their degradation patterns suggest that they are also consistant. Mercedes have clearly done a much better job this year, and they are in a much better position to challenge. It seems they haveΒ inherited Ferrari’s position in the top 3 of the sport. Ferrari have fallen in the clutches of Force India, Lotus, and the rest. Lotus’ pace is a bit inconclusive, because they have done no heavy fuel stints and their fuel corrected low stints pace is a questionmark. As for the rest, they all look in very good shape, which is proof of the extremely tight battle that is developing in the midfield.

I, for one, cannot wait for the season to commence, and I hope you will be here with me to share our thoughts and experiences… I hope you liked this series of articles on testing, and do stick around! πŸ˜‰

  1. Kevin Irwin says:

    Thanks for that, I still think McLaren will be the team to beat as Red Bull must have a question mark over their spec “B” car because of how little running it got, roll on Melbourne

  2. katmen says:

    Thanks for sharing…GREAT!

  3. BodomReaper says:

    Great analysis! Thanks for all your hard work! : )

  4. Peter says:

    Sound, scientific analysis. The results are consistent with what a lot of F1 websites have speculated on, but its great to have someone data mine the laptimes for proof.

    It will be an interesting season and I wonder if we’ll see a different dynamic from the drivers if Mercedes enter in to the mix. Schui’s robust approach to defending his position last year was somewhat understandable in a car that was degrading rapidly on each new set of tyres. This year, if he and Rosberg are competitive enough to attack rather than defend against Red Bull & McLaren it could get real interesting given how aggressive Schmi, Hamilton and (when he wants to be) Webber can be.

    It’ll certainly give the strategists at Red Bull and McLaren a nightmare trying to figure out what craziness Ross Brawn is going to throw at them. We may see a fair few races where multiple cars converge on the final laps with different tyre strategies.

    I still have a worry that if the Renault engine is as fuel efficient as reported that Red Bull will be able to continue qualifying on the front row by virtue of their lower fuel load. Then create enough of a gap in the early laps to be able to manage the race while others fight it out in the hot, dirty air behind them.

    Great post though. Keep it up please! πŸ™‚

  5. ABAD says:

    Brilliant job, plenty of data and which is more important, reasonable interpretation of the stints.

    My point of view, based on the photos of the tyres pressure used by Ferrari, apart the use of “old” tyres in several stitns, plus some good info coming from paddock make me think that Ferrari is not in that bad situation they pretend. In terms of performance, they are not the first team but close.
    The programme they have been following is not only completely different to any other team (what is not strange) but completely different to any other of their pre-seasons. Especially in terms of use of tyres.

    Thank you !

  6. ABAD says:

    One of those (many) photos iΒ΄m talking about. 1,7 bars pressure!! ItΒ΄s 0,3-0,5 more than should be to get a good performance. ΒΏWhat are they testing? VΓ­a piusgasso (cam-man)


    • abu says:

      Maybe they are just testing heavy fuel loads. Normal F1 tyre pressure is around 1.1 – 1.2 bars, enough for 600kgs, so an added weight of 180kgs from a full fuel tank would be enough to adjust the pressure upwards by 0.4 bar. A normal road car is 2.2 bars, so these numbers don’t surprise me much, I don’t think they are testing anything special. Perhaps they even go to extreme ranges of pressure to test the setup and degradation effect. It should all be part of a normal process of understanding a new car philosophy.

  7. Varun says:

    Great work mate – I know there are many unknowns and like every year, Melbourne (or Sepang) will provide all the answers, but despite that, your analysis is far better than some journalists have produced, and they get paid for it!

    Quick question for you – I know the laptimes and stint lengths are available online, but what about fuel load estimates? How did you classify times as medium/long stints – only based on the stint times or did you have access to other info by any chance?

    Cheers, hope your data agrees with what we see in two weeks!

    • abu says:

      Hi there mate. Thanks for your kind comments…. With regards to stints, fuel loads are only assumed on the basis of the lap times. You get a general picture – a car can’t be very light if it’s doing 1:29’s and 1:30’s, in much the same way that a car that does 1:22’s isn’t heavily fueled. Of course the picture is blurred (a lot), but if you take the time and go through all the data of all the testing sessions, some “patterns” do start to appear. You just have to know where to focus and which data to throw away… πŸ˜‰

  8. Juniperus says:

    Thanks to another great analyis.

    Where did you get your lap-time informations?

  9. Mick says:

    Always a pleasure to read. Thanks so much. Lets hope the Mercs can challenge for race wins!

  10. jeanrien says:

    Once again great job, nice to have all those data, numbers and the commments along. Indeed not easy even with all this to draw any conclusion as added to fuel level, we don’t have the car spec for every run. And if an upgrade is a bad one, it can easily slow the car down quite significantly and modifying the average.

    Great to have compute those “corrected” lap time, much more relevant (if it can be ^^). And with the legendary development attitude of McLaren during the season, we could expect a very nice Champ this year. We have to hope ferrari won’t be too far out, I’m not their biggest fan but love when it’s close up front and if it can be 4 teams batteling, it would be awesome … Can’t wait to Melbourne, those 2 weeks are going to be long…

    For those searching for lap times info, you can get them here (detailled for every day) :

  11. juan mateo horrach torrens says:

    Abu, i did not have read this post before my post. Respect of your questions about the aspect of new cars, mclaren is different and much more esthetic than the rest, wich present the new front. Not very beautiful in any case, but better than seen in picture. Of all, i like Lotus (presentation, colours, shape…).
    Good work. It is not easy what you do, and your courage to express conclusions. I wait for your comments about Australia and the rest.
    Thank you very much again.

  12. TheChosen says:

    Great analysis… I think it will take at least few races to have some order as most teams are so close to each other

  13. Om Sammy says:

    Thumb up, good posting

  14. Gordon Stewart says:

    finally a website that delves into the juicy stuff. Glad I found it, now a #1 F1 bookmark. Get a press pass and some funding cause you make autosport look like a primary school newspaper.

  15. madmax says:

    What a great analysis, It is laughable comparing Benson’s posts on testing to yours.

  16. Avais Ibrahim says:

    A very thorough analysis. Thankyou for this.

    My question if you can answer. Regarding Engines, how fuel efficient are renault engines compared to mercedes engines and is the difference between the two a significant enough to give one team an edge ?


    • abu says:

      It has been reported that Red Bull can start the race with 15 – 18 kgs of fuel less than the Ferrari-powered and Mercedes-powered cars. Assuming they carry, say, 170kgs of fuel at the start, that would mean that the Renault engine is about 10% more efficient, which (as you understand) is huge.

  17. mikeyd85 says:

    First time I’ve seen this blog.

    What a lovely and refreshing change from all the “big boy” reporters out there! This is proper analysis using sound logic to try and extrapolate something useful out of what is widely acknowledged as almost useless information (see Martin Whitmarsh saying the big boys can go 2 -3s a lap faster). Brilliant work! Between you and @scarbsf1, I think I’ll be fine for my analysis.

    I shall return, for sure!

  18. Akheel says:

    Supreme post. Thanks to F1Technical.net, where I found link to your blog. I have followed all your posts and just amazed about the way you are using the data and projecting the things. I cannot agree anymore than what Gordon Stewart says. You do indeed make Autosport look like a primary school newspaper, which has been my favorite link all these days, but not anymore. πŸ™‚ Please do work on some funding and then get near the paddock to give even closer pictures of the facts. I am a great fan of Mercedes (via Schumacher πŸ™‚ ) and wish that they challenge for wins this season.

    My hunger for reading your blog increases after every post. Can’t get enough. πŸ™‚

    • abu says:

      Cheers mate – I really appreciate it… But I am not in it for the money, or the funding… I am just in it for fun. I am only doing this because I am having fun doing it, and I think this comes across and it’s why some of you have enjoyed it. I don’t know what will happen if I start to think “differently”, so I probably never will πŸ˜‰

  19. Aussie Rod says:

    I stumbled across your blog the other day and have come back each day since. Your analysis is fantastic and truly insightful. Thanks for the effort putting all this together.

    You’ve instantly joined JA on F1 and F1 Fanatic as my two (now three) sources of F1 info and analysis.

    Cheers mate!

  20. Freckles81 says:

    Very clever, thanks for that.

    • abu says:

      Thank you, and good luck with your blog – I see you are just getting started… πŸ™‚

      • Freckles81 says:

        Thanks, yes this is all rather brand new. Think I have rather a long way to go when I see pages like yours though – very inspiring πŸ™‚

      • abu says:

        Pace yourself. I have only started blogging on January, and I am doing this completely on my own. I think that the most “successful” blogs, are the ones that are expressing a passion, something that you feel very strongly about. I am having great fun running this blog, and even though it’s taking up lots of my free time, I wouldn’t change it. So, try to find what it is that fascinates you, and write about it…

      • Freckles81 says:

        Thanks very much for the advice πŸ™‚

  21. Georg says:

    First of all thanks for all that work you put into the pre-season analysis !
    Is it maybe possible to add the type of tyres that teams were running during the selected stints?
    e.g. rosbergs 15Lap Stint that started with 1.22.9
    Was it on soft?medium? It would be great to have an overview πŸ™‚

    Greets from Germany !

    • abu says:

      Thanks mate. Nico’s time was done on the soft (yellow marked) tyre, which is the normal qualifying tyre for Barcelona (the red super soft doesn’t really fit the track). As for the rest, I am afraid no one has info on the tyres used for all the laps at all times, so if I were to post this information, it would be scattered and wouldn’t make much sense….!

      Thanks for following and my best greetings to your country! I’ve been to Berlin and Hamburg, and loved every minute of it… πŸ™‚

  22. Hun Rob says:

    Hi Abu,

    Just like Aussie Rod, I was lucky enough to hit your site when googling for post-Barcelona analyses. Great effort and I have thoroughly enjoyed reading your posts.

    I’m a number cruncher type myself which leads me to the inevitable question: where do you get your lap times from? I have been unable to find anything other than the daily or combined best lap times offered by the more ‘orthodox’ sites.

    Once again, thanks for all the effort and the in-depth analyses. Their acccuracy, or lack of it, for that matter, will be revealed shortly πŸ™‚

    • abu says:

      Hi mate – you can try this site for laptimes: http://mclarenf-1.com – that’s where I got what I missed during the tests…

      I am pretty sure the results are inaccurate – they don’t tell the truth, they tell a story, which is up for interpretation πŸ™‚

  23. Kalle says:

    Outstanding analysis! It’s really fascinating to try to understand and even predict outcomes the way you do. Fascinating stuff, keep up the good work!

  24. Mario says:

    In the low fuel analysis, i think you forgot to consider a stint that Lewis Hamilton did on the last day of testing. It started with a 1.22.430, on soft tyres, and after it he did 5 more timed laps. If you take in consideration the 0.150 sec/lap figure, is best time should be at least a 1.21.680.

    This way, we could assume that would be the 2nd fastest lap in testing, only surpassed by Rosberg’s one.

  25. Mario says:

    I’m afraid he did….see it here: http://f1tests.co.cc/archive.php

    On 4th of March, Hamilton did the following times:

    lap 46 – out lap
    lap 47 – 1.22.430 (his fastest time of the day)
    lap 48 – 1.25.899
    lap 49 – 1.25.231
    lap 50 – 1.25.195
    lap 51 – 1.28.301
    lap 52 – 1.26.582
    lap 53 – in lap

    • abu says:

      Mix up of posts! πŸ™‚

      • Mario says:

        πŸ™‚ Anyway, great piece of analysis! There’s some really interesting conclusions… I’ve seen nothing with this kind of depth on the net!

        I’m looking forward for your next posts….


      • abu says:

        Thank you, and apologies for the mix-up – entirely my fault of course πŸ™‚

  26. Interesting article. It’s nice to see a real attempt at treating the data properly. When we have 1.5s covering the grid, and 1s of time between soft and hard tyres and at least one second in engine modes, it makes the signal-to-noise very low (and opens questions to which ‘bin’ to put the data into). And then there’s the age of the tyres. You can see why I dodge a lot of this data – in some ways I think it’s quite brave to put it all in.

    A few questions to help my understanding of your analysis:
    For the degradation, do you do a straight line fit through the lap times? This is effectively what I do, but in a different way.
    Have you removed fuel effect or are your degradation numbers actually (degradation-fuel)? (Note I think fuel effect is about 0.1s per lap if you look at stints where just a pitstop was made – 15 laps gives about 1.5s pace difference).
    Both McLaren and Mercedes have refuelled during their ‘races’ according to my model. Your data shows a Mercedes race on 4 hard tyre stints as the fastest – faster than all the runs which used mainly softs. Not sure I buy that – but I agree that Mercedes is fast.
    When adding laps to the ends of races, did you allow for the fact that the laptimes are faster (up to 5s) at the end of the race due to fuel effect? If so, how did you decide how much time to take off?

    Anyway, nice stuff. I’ll try to find time to get in touch.

    • abu says:

      Yeah, it may be brave or it may be foolish…. πŸ˜‰

      I loved your article too – very focused and precise, I hope it turns out to be accurate. To answer your questions:

      1. Yes, I the linear approach, and then calculate the maximum / minimum points on the line to get the Delta.
      2. No, I have not removed the fuel effect, so what I am presenting is not the tyre degradation per se, but it’s the combined (and opposing) effect of tyres going off and fuel going down, working on the assumption that the fuel effect is the same for all teams, or close enough to actually disregard.
      3. I don’t trust race simulations, as I have explained. I think all teams were refuelling in-between stints, and the only question is by how much. We don’t know that, which is why I said that race sims is the worse set of data in our hands.
      4. Let’s say a 63 lap race has a 1:29.000 average laptime, so the race time would be 63 x 1:29.000 + pitstops time. To calculate the 66 laps race I use the same average lap time, i.e. 66 x 1:29.000 + pitstops time, which is close enough, as long as I don’t do it in significantly shorter race sims (e.g. 55 laps). If I had simply added 3 laps of, say, 1:25’s, it might have been a more accurate, but (a) it wouldn’t change the overall picture dramatically and (b) I wouldn’t know what kind of lap times to reliably add. But more importantly, it would just be too much work just to cover an area (race sims) that is the most unreliable and difficult to read. I didn’t think it was worth the bother…

      Do get in touch, anytime you like πŸ™‚

      • Thanks for the response. It’s helpful. I’m sure there is some truth in our analyses, but knowing which bits are reliable is the difficult part. To answer the answers:

        1. I found that the last laps of a longish stint were often ‘phase 2’ degradation (there is a different optimal fit if you include them) so I tended to ignore them on the assumption that they would pit as soon as they saw this in a race. It doesn’t skew the numbers too much, but it does give me lower degradation numbers than you. However, I agree that Sauber (oddly), Williams and Ferrari are struggling with degradation. So maybe we have something there.
        2. If it is about 10% fuel load difference between cars and I think I am about +/-10% on fuel effect, then I can’t pick differences between cars. I think you’re safe with this assumption. A fifteen lap stint would be 0.15s difference in pace at the end (so less than 0.1s per lap average over a stint).
        3. I like multiple stints as you can pick an idea of fuel effect, and thus fuel levels from them. So I concentrate there. The nice thing is that (generally) our different approaches give reasonably similar (if somewhat blurry) pictures of what is going on.
        4. OK – if you only added a few laps, then no problem. The easiest way to get something ‘close enough’ would be to repeat the last representative laptime you have for the last few laps. But I agree, it wouldn’t change the story.

        Oh, and apart from the analysis, which is the real interest for me, you have a good way of writing things.

        Do you have any plans for live analysis, or post-race data analysis?

      • abu says:

        I definitely agree that the difficult part is understanding which bits of the analysis are correct, and which point to the entirely wrong direction. I did this analysis mainly to give some substance to some on-track observations by the so called expert journalists, such as the claim that Ferrari is struggling. I think both our analysis show that, yours even more than mine’s. Maybe I tried to stretch my legs too much, and get too much data in – it would have worked much better if I had solid tyre information, and it would help to know which teams were refuelling between stints and which were not. The whole race simulations thing is a charade – I’ve never seen a team bullshiting us as much as McLaren, in their race simulations. Ah well, with regards to the specific points you raise:
        1. Yes, my degradation figures don’t remove the phase 2 degradation, although I tended to disregard extreme “peak” times. If, for example, a stint would be in the 1:25s, and then you had a 1:26 and a 1:28, I would disregard the 1:28 laptime from the degradation calculation. However I wasn’t consinstant in doing that, in the sense that I didn’t have a set methodology (i.e. disregards laps above a certain set degradation figure). It’s all room for improvement, I guess!
        2. Agree.
        3. I like multiple stints too but, as I said, maybe I stretched the whole analysis a bit too much – it’s very interesting indeed that we are getting similar results. I was surprised by the kind of speed advantage you were able to attribute to Mercedes, and I do feel that there’s more to Mercedes’ pace than meets the eye. To phrase this differently: if Red Bull were getting Mercedes’ testing laptimes, we’d be talking about possible domination at the moment.
        4. An even more “close enough” method would be to see the degradation pattern in low fuel running, and use that (against the fuel effect) to calculate the last 2-3 lap times of the last stint, but then, again, I didn’t have that much time in my disposal. I do have a day job! πŸ˜‰

        Thanks for the compliments; I really enjoy your articles too. They are scientific, well researched and to the point. Maybe we can pit our strengths together!

        As for the races… Yes, I am going to probably do some analysis, but I haven’t figured it out in my head yet…….. Open for ideas!


  27. patsevtsev says:

    Thank you for doing this blog. It is better than any other site for F1 and believe me I know them very well. I’ll put you in my phone as favourite bookmark . I hope MS wins again. I so desperate want to see the old invincible Schumacher.

  28. tifosiferrari says:

    phenomenal analysis , hopeful Ferrari did not appear true pace

  29. kanne says:

    your claim about lotus nt running any heavy fuel run is a bit lack of claims…….james allison recently stated in MTV3 interview that he amaze by kimi as he nvr run any full tank b4..>~!!

    • abu says:

      No, I don’t “claim” anything. Lotus didn’t do any back to back heavy fuel running, like the other teams. Apparently Allison is referring to the heavy fuel running during the race simulations that Lotus did.

  30. juan mateo says:

    I have been reading your last posts, intelligentf1 and Abu. Is very interesting all both you said, and taking in count that i am very interested but i have not time enough to work more on the fingers and data available, i would like to thank your effors and work.
    However, thinking about the results and making the qualitative hipotesis:

    1) Red Bull is cheating more than anybody his potential, based on the last years own experiences in pre season tests, and my personal view in Barcelona 1 and 2 of mars.
    2) In general, the differences are lower than last year, because of the regulations changes.

    My final conclusion about pre season test results is based more in qualitative items than quantitative.
    That is to say, the feelings looking the cars at the track, the way in every team makes tests, the experience in the last years, comments made by drivers, thechnical managers and managers, gives me more clues than the fingers. And, please, dont understant in that a disrespect over your work, which seems to me fantastic in any case. Far from that. But taking in count degradation tyres gradients, variability in age of tyres and quality of compounds, quantity of fuel and refuel operations, engine mapping, drivers intensity and skills, different track conditions; all unknown, the noise of all together is very contaminant for math analysis.

    So, at this moment, my feeling is:

    Red Bull, Mclaren, Mercedes, Lotus, Ferrari, Sauber, Force India, Toro Rosso, Williams, Catherham, all in less than 1,5 sec gap, and then 2 sec far or more, HRT, Marussia.

    Best regards

  31. jjpm says:

    Excellent job! thanks and some stuff that you may find interesting…


  32. Pop says:

    Raikkonen did the fastest lap in Barcelona with 20 kg of fuel

  33. Pop says:

    Last part of the analysis is nice but it’s not a surprise that Lotus did not long runs; J.Allison said on the first day of Barcelona 2 that Lotus will do short setup runs and race simulations because they lost 4 days of running.

  34. VJ says:

    Thank you very much for unbelievably good analysis! πŸ™‚ this is great, very well written(easy to understand) and, IMO, very informative and precise analysis. Keep up the good work!

  35. Gybrush Threepwood says:

    Ummm blogger, you forgot to take into account the different tyre types and the effect they have on lap times. It’s not only about fuel loads.

    Great write-up though.

    • abu says:

      Hi – no I didn’t forget to do that. It’s all a matter of methodology. You have to accept that you can’t have complete info on tyres usage for all testing days, for all teams, for all laps. That’s completely impossible, so you have to go on without that info. However, considering that the teams tend to test the same tyres during the entire pre-season (they are allocated with the exact sets of tyres, including un-marked test tyres), we can assume that over the entire pre-season testing, this tyre effect tends to “even out”. That’s a large assumption to make, of course, and I am sure it doesn’t apply in all cases and all teams. The margin of error is big, but it’s the best assumption we can make, otherwise we can’t make any comparison whatsoever.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s