The battle for the LMGTE-Am victory at Le Mans 2023

by Boris Deshev, Ph.D.

Image: Focus Pack Media - Marius Hecker

At the centenary edition of the 24 Hours of Le Mans, the #33 car of Corvette Racing decisively won the LMGTE-Am class. On their way to victory, #33 went from first in class to last overall and then back to first in class again. This is something that occasionally happens in motor racing, and a 24-hour long race certainly offers plenty of opportunities to recover from early technical problems. However, such a performance always indicates a dominant race pace within the class. In this article, we delve into the data from the 2023 edition of the great race to examine what determined the final order in the LMGTE-Am class.

THE COMPETITION

Of the 21 cars entered into the race in the LMGTE-Am category, only 9 managed to finish. Throughout the entire 24 hours, 11 cars took turns leading the class, with 5 of them leading for more than a stint, which typically lasted 9 laps. The #33 car led for 67 laps, second only to the #56 Porsche 911 RSR-19 operated by Project 1 – AO, affectionately known as Rexy due to its inspired paint-job. Rexy led for 78 laps and remained competitive for the first 17 hours of the race but eventually slipped to 7th place in class, trailing #33 by 4 laps. The third-highest lap leader was the pink Porsche #85 of the Iron Dames, leading for 63 laps. #85 held second place in class until the penultimate hour of the race but ultimately finished fourth, one lap behind. Noteworthy leaders in the class also included #57 and #54, leading for 53 and 23 laps respectively. Only the latter of the two managed to complete the race, finishing fifth in class.

The car that secured second place, the #25 Aston Martin operated by ORT by TF team, only led for 5 laps during pit stop reshuffles. The car finishing in third place, the #86 Porsche 911 RSR-19 operated by GR Racing, led for a single stint during the third hour of the race. After all, it is an endurance race.

In this article, we will compare #33 with the other podium finishers—#25 and #86—as well as car numbers #56 and #85, as they were the most formidable competitors at any point during the race.

PIT STOPS

Table 1. Pit stop statistics for the five main competitors in the LMGTE-Am class. The final positions in the class are written in the brackets in the first column. Data property of Al Kamel Systems

Table 1 lists the pit stop statistics for the five cars we are concerned with in this article. The #85 Porsche of the Iron Dames team was the most efficient with their pit stops. They made fewer stops than any of the competitors and spent in total 19% less time in the pit than the winning car #33. Their longest stop was 2:42.528, which is slow for a normal pit stop but it also indicates lack of any major problems. Cars numbers 33, 86 and 56 all had one pit stop during which major repairs were made loosing around or more than 3 laps each. The car with the worst pit stop record was #86 which, despite that, finished third. If pit stops were the decisive factor the finishing order would have been different.

SPLITTING THE RACE INTO FAST AND SLOW LAPS

Figure 1. The working of the algorithm separating all race laps into fast and slow. On the left is the median normalised lap time for all platinum drivers. On the right is the distribution of those lap times with a Gaussian fitted to the main peak. The separation line between fast and slow is plotted in magenta

The data provided by WEC and Al Kamel Systems does not include detailed information about the track conditions during the race, such as rain or yellow flags. These factors are not indicated in the dataset.

To distinguish between fast and slow laps, an algorithm is employed to create a map of the median pace in time-space, as illustrated in Figure 1. The data contains the time stamp of every crossing of the finish line by every car. For this algorithm only the laps driven by platinum drivers are taken as their pace is more stable and thus best represents the track conditions (see for example this article). Every lap time is normalised to the fastest that the given car has achieved during the race. After that the whole ensemble of relative lap times and their corresponding time stamps are binned in bins in time as wide as the fastest overall lap. Plotted on Fig. 1 is the median relative lap time for every bin. The semi-constant fast pace is visible at the bottom.

Following that a Gaussian is fitted to the main peak of the distribution and a division point between fast and slow laps is selected as the point where the fit goes below 1 lap. The distribution and the fit are shown on the right side of Figure 1. The cutoff line is also shown as a horisontal magenta line running through both panels. Based on this cutoff every time stamp in this set is flagged as slow or fast. Those flags are then mapped onto the lap space of the entire data set for the race. For every car, a lap is flagged as slow if during its duration there is a time stamp which is flagged as slow.

One limitation of this method is its inability to differentiate between the causes of slow track conditions, whether it be due to weather conditions or caution periods. Currently, there is not much that can be done to address this issue. Of course the racing on wet track is even more difficult than on dry one. The split is made in this way just to try and highlight the factors that determined the final order in the race.

As can be seen on the right column of Figure 2 the algorithm is not perfect, as some apparently fast laps are flagged as slow, however, this applies equally to all competitors so we can still use the algorithm to analyse fast and slow laps separately.

RACE PACE

Figure 2. Distribution of fast laps (left column) and slow laps(right column) for the 5 main competitors (rows) over the 24 hours. The lap times are expressed as percentage of the fastest in class. The averages of the distributions are shown as black solid lines and value printed close to the top of each panel. The medians are shown as red solid lines. The numbers in the top right corners indicate the number of laps in the panel.

Using the algorithm described in the previous section the racing laps of the five main competitors in the LMGTE-Am class are shown on Figure 2, split into fast and slow laps. The class winning car #33 did 313 laps over the 24 hours of the race, and the main competitors ended with a lower but similar tally. Of those around 50 were either pit-in or pit-out laps (not shown on any figure), around 150 were high pace racing laps and the remaining ~110 were at somewhat reduced pace, dictated by either rain or yellow flags or safety cars. It is likely that the algorithm used to separate fast from slow laps has misidentified a small percentage. 

Few things are notable on this plot:

The slow laps contain additional information. Remember that rain affected laps are also included on the right side of Figure 2. Assuming that cars make no gain during caution periods, all the difference in the pace during slow laps must come from their rain affected laps.

Figure 3. The fast lap times during the 24h race for the five main competitors. Colors correspond to gold, silver, bronze, and pink and green are for Iron Dames and Rexy, respectively. Shown are only series of at least two consecutive fast laps.

CONCLUSIONS

Perhaps the main competitor in the LMGTE-Am class, alongside Corvette racing, were GR Racing and their Porsche 911 number 86. This was the car that spent the most time in the pit and also had a period when they could not maintain the high pace that they were otherwise capable of. #85 executed the race very professionally both in the pit and on track, however, they lacked pace to challenge for the victory.

more from kineticum.com

The Iron Dames car took pole position with ease and kept the strong pace during the race. Yet it finished fifth in class. We analyse the data from the 6 Hours of Monza, 2023 to understand why.

We analysed the way the battle for overall victory developed and what factors decided its outcome. The strategy mistakes, the predictably unpredictable weather in the Ardennes, fundamental car design concept, and outright pace all played a role.

Previous comments