F1 Testing: First Race Simulation Analysis

Posted on February 23, 2012


And so, as the engines burst back into life, so the fingers flex and the dormant intelligentF1 blog shows semblance of life once more. It’s not that I’ve been idle over the winter, as there is a prototype live analysis tool in the wings (which uses JavaScript and Flash so cannot be posted here), and gentle refining of the model has been done. However, the sight of new data allied to the promise of a new season has sparked some action, and an attempt to glean something useful from testing.

There is a reason to wait for the first race simulations before trying to analyse in detail. Firstly, we are looking at a small data sample with myriad variables – therefore more data is good and isolated stints can tell you nothing about fuel loads and suchlike – just reading this blog’s analysis of Friday practice demonstrates that quite clearly. Secondly, a long run can tell you more about the consistency of the times; are the engine modes similar enough? do the paces of the stints relate to the fuel load lightening? how are the tyres performing?

Reading many sites, it seems that the fuel load and tyres being used are considered to dominate the variation in laptimes, and that a race simulation would remove much of this uncertainty. So faster is better. Unfortunately, there is no reason to believe that this is true. Consider the following thoughts:

– the race simulation might be started at any fuel level from actual fuel needed for a race distance (maybe 150kg) to full tanks (maybe 170kg) – 20kg is worth somewhere in the order of 0.6s at Barcelona. That’s a lot over 66 laps.

– the car setup will vary from circuit to circuit. With 20 races, there is little need to optimise to this track – therefore there will be some performance to find for all teams, and how much this is will vary depending on the car and the baseline setup being used. It’s likely that there will be 0.5s (or more) in this.

– there is no reason why the engine modes used, or tyres used will reflect what is done in a race. Schumacher’s stints from today (Thursday) show a few quick laps and a jump to consistent times about 1.5-2s slower. This is not likely to be tyres, and much more likely to be playing with the engine.

So the picture is actually more complex than most assessments would suggest. Does that mean that nothing can be said? Well, let’s see. The first race simulations were done by Vettel (Red Bull) and Hamilton (McLaren) at similar times in the afternoon on Wednesday. Using the laptimes, and inserting some fictional data around the pitstops, a race history chart of the two simulations can be constructed. This is shown below, with Vettel in blue and Hamilton in black. The dashed lines are intelligentF1 model fits matching the gradient in each of the stints.

The first thing that can be seen is that Vettel completes the race distance much more quickly than Hamilton, and his stints can be fitted quite well using the intelligentF1 model. Indeed, the main difference is that Vettel’s ‘race’ is consistent in terms of increase in pace with fuel burn, whereas it is less clear with Hamilton. It is worth noting that we know Vettel did not refuel during his simulation, so the fit of his ‘race’ should be expected to work. Hamilton, on the other hand, disappeared into the garage between stints, so there is no telling if refuelling took place.

It is interesting to see that the first stint-and-a-half are near identical at which point Hamilton slows – this looks like ‘phase 2’ degradation from an old (but not hard) set of tyres. This opening suggests that the cars are very closely matched subject to fuel/setup/tyre/engine considerations mentioned earlier. However, after the second stop, the story is different. For stints three, four and five, Vettel is 0.7s per lap, 0.3s per lap and 1.0s per lap respectively faster than Hamilton for an equivalent fuel load (assuming that no refuelling took place). This can be partly explained by tyres – Hamilton has been quoted as saying he was on old tyres, and the degradation curves support this – either the McLaren is eating its tyres, or they were old. There is also the possibility that the McLaren did take on some fuel at the second stop – about 20kg would make the first half of the trace consistent with the second half of the trace. Other possibilities are engine modes, or tyres. Tyres are less likely as  Vettel used faster tyres for the first half of the race (stints 3 and 4 used slower tyres – about 0.6s slower). On balance, fuel is the most likely explanation – but I don’t know why they would choose to do this.

So what does this tell us? Nothing definitive, but it does open the can of worms. Red Bull certainly gave the best indication of what they can do, McLaren are being a little more cagey, which could be either good or bad. Tomorrow’s job is to assess the runs of Webber, Button and Schumacher. The contrast of the runs by team-mates should shed some more light on the race pace of the cars.