Thursday, 7 July 2016

LMP2 at Le Mans - Some Answers?

Less than three weeks have passed since the dramatic events of the 2016 Le Mans 24-hour race, but somehow it seems longer than that. I seem to have spent a lot of time discussing the emotions of the week with a lot of people. Having been to Le Mans every year since 1981, it remains a very special week, but somehow perhaps not as much of a highlight as it once was. That said, the impressions of 2016 will remain with me for a long time. I have read a lot of what has been written, and listened to a lot of what has been said in the last three weeks – but there has been so much that I have probably missed most of it. Undoubtedly the rights and wrongs of the stories from the LMP1 and GTE classes warranted the extensive coverage, and yet, somehow I feel that there are elements of the stories that might never be known – at least not in public.

But before moving on, I want to spend some time reflecting on the LMP2 class, which may have been put into the shade a little by the shenanigans in the other classes, but nevertheless holds stories of its own. It was interesting that the winning Signatech Alpine of Nicolas Lapierre, Stéphane Richelmi and Gustavo Menezes not only had a slower average lap time, but also spent longer in the pits than the second-placed G-Drive Oreca 05 of René Rast, Will Stevens and Roman Rusinov. As regular readers will know, I use the average of the best 20% of green laps to establish the ultimate performance of the car, and the results of the class were as follows.
Pos. No. Team Race Best Lap Time Race Average Lap Time
1 36 Signatech Alpine 3m 37.195s 3m 39.038s
2 26 G-Drive Oreca 3m 36.558s 3m 38.739s
3 37 SMP Racing BR01 3m 40.065s 3m 42.476s
4 42 Strakka Gibson 3m 38.795s 3m 41.968s
5 33 Eurasia Oreca 3m 38.605s 3m 41.700s
6 41 Greaves Ligier 3m 41.806s 3m 44.820s
7 27 SMP BR01 3m 39.445s 3m 42.371s
8 23 Panis Barthez Ligier 3m 39.629s 3m 42.721s
9 49 Michael Shank Ligier 3m 37.339s 3m 41.315s
10 43 RGR Morand Ligier 3m 38.734s 3m 41.406s
11 30 ESM Ligier 3m 42.146s 3m 44.037s
12 25 Algarve Pro Ligier 3m 40.450s 3m 43.218s
13 40 Krohn Ligier 3m 39.998s 3m 43.442s
14 22 SO24! Ligier 3m 43.769s 3m 47.349s
15 48 Murphy Oreca 03 3m 41.582s 3m 44.470s
16 31 ESM Ligier 3m 39.156s 3m 42.189s
17 34 Race Performance Oreca 3m 43.647s 3m 46.774s

And the time spent in the pit lane was:
Pos. No. Team No. of Pit Stops Total Time in Pits
1 36 Signatech Alpine 33 46m 53s
2 26 G-Drive Oreca 35 41m 12s
3 37 SMP Racing BR01 32 41m 05s
4 42 Strakka Gibson 33 39m 52s
5 33 Eurasia Oreca 33 52m 49s
6 41 Greaves Ligier 33 45m 07s
7 27 SMP BR01 36 1h 05m 13s
8 23 Panis Barthez Ligier 33 51m 33s
9 49 Michael Shank Ligier 32 49m 49s
10 43 RGR Morand Ligier 32 1h 16m 23s
11 30 ESM Ligier 30 45m 38s
12 25 Algarve Pro Ligier 34 1h 08m 37s
13 40 Krohn Ligier 32 59m 12s
14 22 SO24! Ligier 32 1h 39m 50s
15 48 Murphy Oreca 03 32 2h 10m 07s
16 31 ESM Ligier 31 3h 57m 50s
17 34 Race Performance Oreca 33 3h 33m 57s

There were also noteworthy performances from the Oreca 05s of both Manor and Thiriet by TDS Racing. Roberto Merhi, Matt Rao and Tor Graves held the class lead for 44 laps in the British-entered car and also set the fastest lap in LMP2, thanks to a lap of 3m 36.259s by Merhi. In the Thiriet car, Pierre Thiriet, sharing with Mathias Beche and Ryo Hirakawa, exchanged the lead for much of the race before Pierre came to grief early on Sunday morning in the gravel at Mulsanne corner. Their average lap times were: 44 (Manor) – 3m 39.448s and 46 (Thiriet by RDS Racing) – 3m 39.697s, so which compare well with the leading two cars.

However, it seems to me that not only the battle between Signatech and G-Drive is worth a more detailed look, but also the fight for the final step on the podium between the SMP Racing BR01 of Kirill Ladygin, Victor Shaitar and Vitaly Petrov and the Strakka Racing Gibson of Danny Watts, Jonny Kane and Nick Leventis.

It seems counter-intuitive that a car spending less time in the pits and with a faster average lap time (the G-Drive Oreca) should lose out to one (Signatech) that spends more time in the pits and has a slower average lap time. Inevitably, and obviously, something else is going on.

Examine the “Rising Lap Time” graph, shown below. This sorts the lap times for each car into ascending order and the plots them, best to worst, from left to right. Whichever line is closer to the x-axis is faster.

Hopefully this shows clearly enough – click on the graph to make it bigger – that the Signatech (blue line) is above (i.e. slower than) the G-Drive (brown line) at the left hand (fast) end of the range, but below for the larger, right hand end of the range. The conclusion is that although G-Drive was quicker for the fastest 80 or so laps, Signatech was quicker for the rest of the time.

The graph does not show – but I am not convinced it is relevant – that the no. 36 Signatech pitted during each of the three SC periods (ignoring the first SC period at the start of the race), shortening the planned stint as it did so… the 26 G-Drive didn’t come in at all during SC periods, hence its shorter time in the pit lane. The average pit stop time for a ‘normal’ stop was 1m 11s for #36 and 1m 12s for #26. Note that G-Drive also had a drive through penalty which would have cost it about 28 secs.

It is interesting that Slow Zones / Safety Cars, nor which driver was at the wheel, seem to have much impact on this pattern. What can be established is that the majority of G-Drive’s quicker times came in the final eight hours of the race, whereas the Signatech Alpine set its quicker times earlier in the race – when the track temperatures were cooler. All this merely goes to show that looking at the average of the best 20% may show the true potential of the car, but doesn’t always reflect its performance over a 24-hour period.

It is a very relevant feature of the LMP2 class that the crew composition must include at least one silver or bronze driver, of course, and in addition, each driver must be at the wheel for a minimum of six hours. To a large extent, this explains the fact that the all-Russian crew in the SMP Racing BR01 was able to bring their car home onto the third step of the podium ahead of the all-British Gibson 015S.

The following table shows the driver comparison for the first four cars in the class.
36 - Signatech Alpine
Name Grade Driving Time Best Lap Average Lap
Gustavo Menezes Silver 7h 08m 53s 3m 37.452s 3m 38.8s
Nicolas Lapierre Platinum 9h 07m 13s 3m 37.195s 3m 38.8s
Stéphane Richelmi Gold 6h 57m 39s 3m 38.112s 3m 40.2s

26 - G-Drive Oreca
Name Grade Driving Time Best Lap Average Lap
Roman Rusinov Silver 6h 54m 13s 3m 36.558s 3m 39.0s
Will Stevens Platinum 6h 54m 11s 3m 36.891s 3m 39.7s
René Rast Platinum 9h 33m 43s 3m 36.563s 3m 38.2s

37 - SMP Racing BR01
Name Grade Driving Time Best Lap Average Lap
Vitaly Petrov Platinum 9h 30m 32s 3m 40.065s 3m 41.8s
Victor Shaitar Silver 7h 37m 02s 3m 41.268s 3m 42.5s
Kirill Ladygin Gold 6h 14m 36s 3m 43.489s 3m 45.1s

42 - Strakka Gibson
Name Grade Driving Time Best Lap Average Lap
Jonny Kane Platinum 9h 25m 28s 3m 38.795s 3m 40.7s
Danny Watts Platinum 7h 53m 38s 3m 41.860s 3m 42.9s
Nick Leventis Silver 6h 02m 55s 3m 45.090s 3m 47.8s

The Average Lap time above is calculated as the average of the best 25 laps achieved, and it is clear that the star drivers are Rast, Lapierre (no surprises there) and Menezes, performing particularly well on his début. But G-Drive car is quicker (when it is quick), no matter who is driving, than Signatech. The French squad won as a result of being quicker when the competition was slower - and that was when the track was cooler. Twenty-four hour racing is all about consistency.

Just go back and look at that “Rising Lap Time” chart!


  1. Just because I could, while I was having my own fun in the .csv, I collected an average lap time for cars for every lap that wasn't balked by slow zones, safety cars, inlaps or outlaps. It's a silly thing to have, really, because it was loads of manual labour with plenty of potential for mistakes to tell me very little that was new.

    One thing I did find is that while the gap between G-Drive and Signatech was smaller using the full set of green laps, it was still there (my data shows #26 averaging 3:42.031 and #36 3:42.191). Inlaps and outlaps are relevant and I ought to figure out a method to include them without including the put time (averaging sector times over a stint, maybe?), but that suggests to me that even over the course of the entire race and including the stints where the Alpine was faster, G-Drive had the edge.

    I alluded to it in my last comment, but I think the safety car pit stops made the difference. It helped that the timing was perfect for the #36; between their five-minute repair and the split of the final couple of stint, the #36 never ran fewer than 8 laps on a stint, allowing them to not have to make an extra pit stop.

    I’d like to challenge the notion that pitting under safety car conditions is ever really a disadvantage for anything slower than an LMP1 (so long as the car has been running on its tyres for more than a couple of laps). Each stop, in theory, costs exactly one-third of a lap assuming everything with the stop goes well. The time loss is different for every category, of course, because one-third of a lap in a GT takes longer than in an LMP. In LMP2, the time loss, using my completely infallible, reliable average time data, is roughly 1:14. On average, it took Signatech roughly 25 seconds to change tyres, 27 seconds to refuel and the pitlane transit time was around 33 seconds. These are all rough numbers, but it means that, were it accurate, Signatech in this instance would need to refuel ~60% of their car’s tank to break even on time, or have run seven laps or so. It’s a free tyre change if they have done those seven laps, which they had every time the safety car came out, so even if the car hasn’t run even a full stint on a set of tyres, it’s advantageous to swap them anyway because they are gaining time on a normal tyre change.

    While that may cost some time in the short term (having a 1:14 time loss instead of a 01:00 time loss by waiting until after the safety car and when the tank is depleted when the tyres can do two more stints), that time could be made back up by running an extra stint on fresh tyres and introduces the possibility of skipping a tyre stop later. Something to think about, anyway, although I’m sure you could set me straight.

    The other thing is that 5:07 of Signatech’s pit time was spent conducting some sort of repair under safety car conditions. Under green, that would have meant losing around a lap and two-fifths, but under safety car meant losing two-thirds of a lap (making the assumption based off of the eight minute safety car laps). I did some more super precise assuming and throwing numbers in different places and came up with a 2:39 “lay off” as it were as it’s not really time gained, but it’s time not lost. I’m tempted to go further into the files and see how much more of the time saves I could attribute to safety cars, but I also want to sleep sometime today.

    As icing on the cake, I dug up a neat little bit about the slow zones while trying to understand the LMP2 race fully. Apparently, G-Drive hit a third-sector an extra time on two separate occasions. And they still could have raced the #36 to the line if they didn’t eat two minutes of penalties. There’s a lesson in there somewhere.

  2. LMP2 is perhaps too much about finding the most underated silver driver. Can anyone explain how Rusinov can still be a silver driver under the driver ranking definitions?

    1. He was a gold last year. I don't have stint data from last year (never put the time into sorting out the .csv), but Canal was routinely faster than Rusinov was (outside of Le Mans). Somebody in the process must have realized something was wrong, but rather than upgrading Canal to gold, Rusinov was downgraded to silver.

      This year, Rusinov has been far more competitive. Maybe he was planning carefully, maybe he found a better driver coach, but he was something like 1.5 seconds a lap faster this year than he was last year at Le Mans.

  3. Hi Paul. I heard in some post-race discussions that Alpine was able to do more laps on a tank of fuel than G-Drive. Is there anything to back that statement?