It seems that all articles on testing at this time of year use the words ‘notoriously unreliable’ or ‘poor form guide’ before going on to tell us what the unreliable data says. The problem is that we expect the grid to be covered (up to at least the first nine teams, and probably Caterham too) by of the order of two seconds per lap, and the laptime data is much more scattered than that. Headline laptimes are indeed meaningless, but then I would also argue that is true of Friday practice at a Grand Prix weekend, and only the long runs have any significance. The predictions made from the long runs on this blog from putting physics-based curve fits to the Friday long runs are not always right, but on the whole the optimum strategy and clear form pointers (especially in the midfield) are pretty good.
Therefore, a deeper analysis than ‘average stint pace’ or ‘headline lap time’ ought to be able to tell us something more of the relative pace of the cars. We have to make a whole bunch of assumptions, but that’s part of the challenge. So here’s my set of assumptions:
– From fitting the 2011 Spanish Grand Prix, the laptime penalty per lap of fuel carried is about 0.11s – there are lots of stints in the race on similar tyres, so this should be a pretty good guess. Indeed, it makes good sense of the long run data of the cars we know didn’t refuel in their runs.
– When a car goes back into the garage between stints in the ‘race’ simulation it could have refuelled. The majority of the stints run by the cars are between 10 and 15 laps. There are small exceptions to this (Jean-Eric Vergne step forward), but this equates to about 1.1 – 1.6s pace difference. It seems odd to me that any team would put less than a stint’s worth of fuel in the car between race stint simulations, so if the pace in a stint is less than one second slower than a previous stint, I have assumed that this is not fuel related. I could be wrong – and I am sure that I will be wrong in particular instances, but in general it should be OK.
– There are many instances in the data where the pace difference between stints is of the order of 0.6s. This is an indication of the gaps between tyre compounds, but again is not conclusive.
Working with these hypotheses as a starting point, and seeing if we can make sense of the data by using them, from the first Barcelona test there are 15 race ‘simulations’ (from 9 teams), or at least fragments thereof which we can play with. I have taken this data and used the intelligentF1 model to fit the data and make the most sense I can from it using the Wednesday race simulation of Vettel (as we know he didn’t refuel, and he happens to have nice consistent tyre performance) as a baseline. The easiest way to do this is to go through the simulations and see what we find.
As detailed in previous posts, Vettel did 5 stops and his pace (relative to datum) was 0.0s, 0.0s, 0.6s, 1.2s, 0.0s, 0.0s. This shows very good consistency and different tyre compounds with a 0.6s gap. Note that this pace is what would be expected fuel corrected – if consistent this would show that there is no refuelling and the same tyre compund/age are being used.
Webber did 4 stops, and his pace was 0.0s, 0.0s, 0.6s, 0.6s, 1.2s. This is essentially equivalent to Vettel’s run backing up that Red Bull have a nice consistent baseline to work from.
This is a fragment of the race, with 2 stops. The pace is 0.0s, -0.3s, 0.0s. This suggests to me that the pace is again very similar – the middle stint could be engine map, or newer tyres, but the consistency suggests to me that the Red Bull is running a sensible (perhaps higher than necessary) fuel load at the start of these simulations, and is happy with the performance.
The McLaren pace across the three simulations available has a lot of questions in it. The pace from Button (4 stops) is +1.0s, +1.6s, +1.6s, +2.3s, +3.3s. Given the earlier criteria, it looks like the fourth and fifth stints are repeats of stint 2 and so the car was most likely refuelled. The interesting part here is that Button is a full second slower in the first stint. This is much slower than either of the other McLaren runs, and may well be testing for very high fuel – either that or McLaren are in a little trouble. Given that they have not shown their pace at the fast end of the race, I suspect they have something to hide – but it is not clear if that is a good or a bad thing.
This time Button is much faster. The four stints (three stops) have pace of -1.5s, -0.9s, +0.7s, +2.3s. The third and fourth stints are again repeats of the second stint – as he was running roughly 15 laps per stint (compared with 12 laps the previous day) the laptimes are consistent with this. If this is a test of the race’s second stint, this puts him roughly on a par with the Red Bulls. This would also be consistent with Hamilton’s simulation, but leaves Button’s Thursday run as a bit of a mystery.
As detailed before, Hamilton appears to be repeating stints. His four stops give five stints with pace 0.0s, -0.2s, +1.0s, +2.2s, +2.2s. Similarly to Button, stints 3 and 4 are repeats of stint 2. If we assume that he is running about a stint heavier than Button’s Friday run, the first and fourth stints are almost identical. It looks like McLaren are concentrating on part of the race at a time at this stage of testing. The degradation seems higher than the Red Bull, but they are running longer stints. This looks pretty close between Red Bull and McLaren – but we can’t know what they are hiding.
In complete contrast to McLaren, instead of running only heavy, Ferrari have run only light. Massa’s three stints (for 41 laps) were -3.0s, -2.4s and -2.4s. If we assume that this means that there was no refuelling, and that the car was able to run only those 41 laps, then the pace difference fuel-corrected to a 66 lap race is 2.75s. If this were the case, then this would give Ferrari a 0.25s pace advantage over Red Bull. It is likely that both of the cars were fuelled more highly, so in reality we cannot tell which is faster, but it does seem likely that the Ferrari is reasonably competitive. This evidence suggests that we will not see the Ferrari too far adrift come Melbourne.
The new Mercedes was able to perform a race simulation remarkably quickly out of the box, and Schumacher’s four stints were paced at +1.0s, +0.7s, +1.8s, +1.8s. There is a possibility of a refuel after the second stint, but this pace does seem a little slower than the first three teams.
In complete contrast to the race simulation of his team-mate, Rosberg’s is possibly the single most impressive race run of the test. It seems evident that there is no refuelling, and his four stints have pace of -0.5s, -0.5s, +0.1s, +0.1s. It may well be that this is closest to a real race of any of the top teams, but on this evidence the Mercedes has a decent shout at being a contender. It looks from the traces like the tyres are new (very fast laps after stops) and that the degradation is higher than other teams, but nevertheless, this looks good for the Silver Arrows – and is much more promising than Schumacher’s run.
Di Resta (Friday):
The runs of Di Resta on Friday were (as McLaren) simulations of a particular stint of the race. Di Resta’s three stints are -1.5s, +0.3s, +1.4s. It is not easy to tell the fuel loads, when essentially the same run is being performed, but this suggests that they are within 1.5s of the fast cars – at worst.
The Sauber trace is very different from the others – either Sauber has big tyre degradation issues (which would be a surprise given how gentle it was on tyres last year) or it used very old tyres in the race simulation. Which doesn’t help us much. His four stints are at +0.5s, +0.5s, +2.3s, +2.3s. Again it is clear that there are two stints – refuel – two stints, with stints 2 and 3 pretty much equivalent.
Vergne did one 20 lap stint which showed quite some tyre degradation. It started at the pace of Vettel and finished some 2s per lap slower. Not sure what to make of that. Soft tyres?
Four stints: +0.3s, +0.3s, +0.3s, +0.3s. I guess he didn’t refuel. This suggests that the Williams is doing OK and probably also shows that the leading teams have a bit in hand. It’s about 0.8s slower than Rosberg – I think Williams would take it if that is the gap in Melbourne.
Three stints: +0.8s, +1.8s, +0.5s. The middle stint looked to be on very old tyres as the degradation was horrible in comparison with most other stints. The other stints were OK, and suggest that Caterham may be somewhere near the midfield teams.
We have more questions than answers, and as most of the teams haven’t really shown too much, we cannot know too much. The gaps between the teams on the first stint are very small at around one second per lap. As we expect somewhere in excess of twice that, we can be sure that the fast boys have not shown their hand. However, there is hope for Mercedes and I wouldn’t write off Ferrari.
That Red Bull and Mercedes have run proper race simulations without refuelling suggests confidence to me – that Ferrari haven’t run heavy suggests that they have concerns. The wild card is McLaren – they are hiding something, but is it good?