Tyre Analysis - Valencia

sushifiesta

Champion Elect
Contributor
This is really a follow on from jez's Pitstop spreadsheet thread (definitely worth a look if you haven't seen it), but this is going to be a fairly lengthy post with quite a lot of information so I thought this was the best place for it. I got sucked in and have spent many hours since playing with various things relating to strategy and tyre wear, and after today's race I've tried to have a look into the way the Pirellis were degrading. I did something similar for the Monaco race in the original Pitstop spreadsheet thread.

I'm going to try and look at a few other drivers/strategies (more on that later - I need a bit of help) tomorrow, but for now I've looked at the top six (Vettel, Alonso, Webber, Hamilton, Massa and Button) who all used the same three stop strategy (Option - Option - Option - Prime).

In all the graphs below the horizontal axis show the number of laps in the stint, and the vertical axis shows what I'm calling the "Tyre degradation". This is the change in lap time from the start of the stint adjusted to take into account the effect of fuel burning off (0.081s/lap for Valencia according to the James Allen Strategy Report). If the tyre degradation is positive this means the lap times are slower than the first lap of that stint. Also, if a lap time is much slower than expected I don't include it, for example in many cases the first lap or two of a stint have to be disregarded or laps where a mistake has been made.

TOP 6 DRIVERS - 3 STOPS

1st Stint - Soft Tyre

In the first stint of the race the tyre wear seen be all of the top six drivers is remarkably similar, although Hamilton and Massa suffer more than the rest after around lap 10. The averaged graph shows quite a nice clear pattern with the tyre holding up for 8 laps before starting to deteriorate quickly, particularly after lap 11.

top6stint1soft.PNG


2nd Stint - Soft Tyre

The stand out feature for me from this stint is the Red Bulls who manage to go 13/14 laps on the soft tyre without ever being more than 0.5s slower than the first lap of the stint. It may be that other factors are at play - for example, this is the stint where Alonso overtook Webber - but even so it is mighty impressive. The McLaren drivers suffer the worst in this stint, with Hamilton's tyres already almost 1.5s off their initial pace before he pits after just 10 laps. The averaged graph is similar to the first stint, although the tyres seem to start losing pace straight away.

top6stint2soft.PNG


3rd Stint - Soft Tyre

This is pretty horrible viewing for Hamilton fans, frankly when you look at this it's amazing that he managed to finish 4th. In just 7 laps Lewis' tyres have well and truly fallen off a cliff and sunk into tyre hell - during this time his tyres lose almost three tenths/lap performance compared to only around half a tenth/lap for the rest of the top 6. The tyres do stabilise after that though at ~2s off peak performance. The averaged graph is skewed by Hamilton's horrendous wear during this stint, making it more linear.

top6stint3soft.PNG


4th Stint - Medium Tyre

I've included this for completeness but there's not a lot you can take from it. I had to discard a lot of lap times for various reasons, e.g. Webber's gearbox problem and Vettel cruising towards the end. Hopefully I can gather more information on the Medium tyre from some other drivers.

top6stint4medium.PNG


SOFT TYRE - AVERAGE ACROSS ALL STINTS

These graphs are the average of the 3 above. Again, Hamilton has by far the worst tyre wear, although Massa isn't too far behind towards the end of the stint. The prize for the driver that looked after his tyres the most goes to Mark Webber. I am pleasantly surprised by the very smooth pattern that arises when you average all of the top 6 across each soft stint. I used excel to try various lines of best fit and the quadratic shown seems to fit well (i.e. tyre performance drops off seems to be dropping off as the square of the number of laps it has been used for). The coefficient for the squared terms is very small however so the growth is relatively slow. I suspect that beyond 18 laps or so this relationship would caese to be valid and the performance would stabilise, like in Hamilton's 3rd stint.

top6averagesoft.PNG



Well that's it for now. Well done if you made it this far!! I am planning to look at Perez's 1 stop strategy as well as Alguersuari, Kobayashi and Schumacher who did 2 stop strategies, which hopefully will give me some longer runs on the medium tyre to look at. However, to do this I need to know which tyre they were using for each stint. Does anyone have this data or know where I can find it?

Now I need sleeeeeeeeeeep :sleeping:
 

Brogan

🦶 Leg end
Staff Member
Splendid work!
Really makes for interesting reading.

Would be great to see this for the previous and future races, although I can imagine it was quite a lot of work.
 

sushifiesta

Champion Elect
Contributor
Splendid work!
Really makes for interesting reading.

Would be great to see this for the previous and future races, although I can imagine it was quite a lot of work.

Yeah it takes quite a long time to sift through the data and get anything resembling a pattern out. I'm going to try and resist doing a lot more to be honest but I'll probably end up doing a few anyway :embarrassed:. Also, the lap times and tyre information are hard to come by.
 
However, to do this I need to know which tyre they were using for each stint. Does anyone have this data or know where I can find it?



The best source is Pirelli, here:

http://www.pirelli.com/tyre/ww/en/news/2011/06/26/p-zero-tyres-show-pace-and-durability-in-valencia/
The earlier races are also there, but note that the first race (Australia) is incomplete, and included drive-throughs
as though they were normal pitstops. I did point that out to Pirelli and they acknowledged the errors but
decided against correcting it.
 

Brogan

🦶 Leg end
Staff Member

sushifiesta

Champion Elect
Contributor
Thanks Bro/Brian. I used the FIA document to get the lap times but they don't have previous races.... that's probably a good excuse not to spend time looking at them LOL. I've got quite a lot of free time over the next couple of months so I'll try and do another race or two at some point - maybe Silverstone and the Nurburgring. Unless I can find away to speed things up I don't think I'll be able to continue it throughout the year but I could upload the spreadsheets etc. I've used so others can try if they want.
 

sushifiesta

Champion Elect
Contributor
I've updated the graphs for stint 3 and the average as I noticed a mistake. Button now appears to have similar degradation to Massa whereas before it was less.
 

sushifiesta

Champion Elect
Contributor
Ok, I've had a look at some data for the medium stints, it's not as extensive as what I did above though. Before/if I do any more I definitely need to think about the method a bit and how I can speed up the process because it's a bit tedious at the minute...

What struck me straight away was just how reluctant pretty much everyone was to use the medium tyre. Even Perez, who one stopped, did his longer stint on the soft tyre! Frankly I think Pirelli need to do a better job in this regard - the prime tyres they're taking to races just aren't competitive enough at the minute and if it wasn't for the rule that forces the drivers to use both compounds no one would be using them.

There are two graphs below. The left hand graph shows the average of the stints on the medium tyre done by Hamilton, Perez, Kobayashi, Alguersuari, Schumacher, Heidfeld, Barichello and Petrov. The right hand graph shows the three longest stints that were done on the medium tyre by Perez, Kobayashi and Schumacher.

medium.PNG


Although the averaged graph shows a decent pattern on the whole the graphs for individual drivers varied quite a lot. For example, Barichello completed a 16 lap stint with all the fuel adjusted lap times oscillating around the same sort of time apart from a couple of laps at the beginning and end of the stint. On the other hand Petrov's 14 lap stint was U shaped, with the fastest laps in the middle of the stint and the lap times at the end being similar to the times at the start of his stint (similar to the first 10 or so laps of Kobayashi's stint on the graph above).

Like I say when everything is averaged out the result is ok. It looks like the tyres maintain their initial pace for eight or so laps before losing performance at a fairly constant rate of the order of a tenth per lap, or maybe a bit more (the actual values vary from half a tenth to two and a half tenths).

I'll do one further post at some point to compare the soft and medium tyre directly, but a quick glance at the averaged graph for both tyres shows that the soft tyre is around 1.9s slower after 14 laps whereas the medium tyre is just 1.4s slower after 23 laps. This is a significant difference so the initial performance of the medium tyre must be significantly slower than the soft tyre for the teams to be so reluctant to use it. I remember Martin Brundle quoting 1.3s or something during commentary. I might see if I can extract something like this from the data as well but I think it will be difficult to find a decent point of comparison.
 

jez101

Bookies drive nice cars because of people like me
Contributor
Great stuff sushi, as always mate.

I did pick up the lap times but the problem is the cleaning process from text file to useable data (this comment is more for other readers, as you will no doubt know this acutely!) I have had a busy week so far so haven't taken it further yet, but perhaps what I can do is write some formulas to do this quickly. Can you post your sheet so I can output into your analysis format?

If I can do this we can probably look back and do future analysis a bit quicker?

The Italian site would be great Bro, but I think the times are stored in an image rather that as text...?
 

sushifiesta

Champion Elect
Contributor
Attached is the excel spreadsheet I've been using. It's all a bit of a mess because I wasn't expecting to take things this far. Getting lap times in takes time like you say, partly due to copying and pasting and partly because it's a bit of a pain dealing with times in excel so I've been stripping off the minute and just dealing in the seconds, so 1m47.234 just becomes 47.234s.

What's really slowing it down is that I haven't thought about how to set it out properly so I'm having to copy/paste data all over the place along with formulas etc. Getting rid of times that are much slower (or in some cases faster) than expected takes time too. Like I say after I've finished with Valencia I would put some thought into how to improve all of this before trying anything else.

P.S. Did Perez really start on the Medium tyre? At the minute it's looking like he got less degradation in the first 22 laps of his soft tyre stint than his 22 lap medium tyre stint which is odd to say the least.
 

Attachments

  • valencia tyre analysis.xlsx
    121.9 KB · Views: 90

f1fansp

Points Scorer
I did pick up the lap times but the problem is the cleaning process from text file to useable data (this comment is more for other readers, as you will no doubt know this acutely!)

Are you having to enter the numbers manually?

I've been trying to imitate you folks in the statistical/graphical analysis (nowhere near as useful as your stuff, hence not posted), but it didn't seem too bad transferring from PDF.

From the 'Lap analysis' PDF, Alt-Right click lets you drag each column individually, then Copy-Paste to Excel sheet.

As it's in mm: ss.0 , if you *86400 (I think, 3600*24 anyway) in the next column, you get ss.000.

One thing I wasn't sure of was if it rounds off the decimal part as its only to 1 dp.

I've got an Excel sheet with the lap times (only for the top 5 teams though) if anyone wants it with the boring bit done.
 

Brogan

🦶 Leg end
Staff Member
You can custom format Excel cells to mm:ss.000 for example, or any other flavour you want.

As f1fansp says, you should be able to copy and paste directly from PDF documents into Excel.
 

sushifiesta

Champion Elect
Contributor
That's basically what I've been doing f1fansp, though I've been converting into ss.000 in a different way that's probably slower so I'll try your method next time.

You can custom format Excel cells to mm:ss.000 for example, or any other flavour you want.
.

This is what I've been doing but just putting ss.000 as the custom format. The problem then is although it displays 45.230 or whatever the number it's using behind the scenes is some small decimal so I get into a mess with other formulas etc. Thinking about it now maybe I just need to make sure the whole spredsheet is set to use that custom formatting.......
 

jez101

Bookies drive nice cars because of people like me
Contributor
Are you having to enter the numbers manually?

Um... No! :)

What I am trying to do is make it as easy as possible to analyse any data set, so basically double click on the word "Race" of "Race Lap Analysis", right click, select all text, copy paste into a notepad file.

That gets you all the raw data in one shot that you can import into excel (delimited, without any delimiters if you see what I mean). That would then be pasted as a column into the sheet I have attached.

You then have to clean it and split out what is a lap number, a lap time, a P for pitstop etc, which the rest of the columns do.

You are spot on with the *86400 though - I hadn't thought of that. Maybe excel 2007 is better at recognising a time format as such rather than text as I have struggled with this before when I had the 2003 version (I think).

Now that I have the ability to quickly clean the data, I can put sushi's formulas in and connect the two sheets.

Tomorrow...;)
 

Attachments

  • val laps.xlsx
    195.2 KB · Views: 79

jez101

Bookies drive nice cars because of people like me
Contributor
The best source is Pirelli, here:

http://www.pirelli.com/tyre/ww/en/news/2011/06/26/p-zero-tyres-show-pace-and-durability-in-valencia/
The earlier races are also there, but note that the first race (Australia) is incomplete, and included drive-throughs
as though they were normal pitstops. I did point that out to Pirelli and they acknowledged the errors but
decided against correcting it.
Nice one Brian! Been looking for that - and it says whether the tyre was new or used!!
 

sushifiesta

Champion Elect
Contributor
I'll post what I have got fot the soft vs medium comparison, the initial results are interesting but if I want to get anything else out I need to seriously declutter my spreadsheet lol.

Another two graphs for you:
mediumsoftselected.PNG

mediumsoftsauber.PNG


The top graph shows a comparison between the soft stints of the top six drivers to the medium stints of the eight drivers I used previously. The resulting picture is probably what you would expect, with the soft tyre seemingly dropping off the cliff around lap 14 whilst the medium compound seemingly stabilises at around 1.2-1.4s off peak performance beyond this point. Before lap 14 the degradation rates of the two tyres doesn't seem too different however. I am very uneasy about comparing different drivers from different teams in this graph though so this comparison isn't really valid.

For this reason I had a look at the Saubers as they both did lengthy stints on both tyre compounds. The result is very surprising with the medium tyre seemingly degrading faster than the soft tyre! I should point out that I've only included up to the lap where both drivers had done a stint of that length - this means the last few laps of Perez's soft stint aren't included, and he did suffer high degradation in this time so beyond lap 20 the soft line would quickly rise to around 2.5s by lap 26. We don't know what the Medium tyre would do at that time though, and this certainly isn't what should be happening at any rate!

I think this is worth devoting a bit more time to so I will try and analyse the Soft stints of the 8 drivers I've used for the Medium tyre in the top graph. It may be a while before I get this done though...

EDIT: Note the different vertical axis on the bottom graph. The lap times used are still fuel adjusted as before but I think comparing to the fastest time in the stint may be a better measure than comparing to the start of the stint.
 
Top Bottom