5 replies to this topic

[paulrobs]

Sorry if this has been done before but it is something that has been bugging me for a while now, something that I think McLaren suffer from more than most too.

I think too many teams are relying too much on historical data and simulations and other data rather than reading the race in front of them. They seem to get locked into a specific strategy prior to a race and seem unable to deviate from that. I can't help feeling, certainly in McLaren's case, that India should have been a two stop race. Get the tyres off asap that don't work and get the tyres on that do work and go like hell and change them again when (before) they start to really drop off. The tyres are on such a knife edge that endless simulations simply cannot allow for the massive drop off in performance when slightly outside of the operating window in race conditions (later in the weekend) when conditions are so different to when they do their long runs on the Friday to gather the data that they use to simulate various race strategies and decide the correct one.

So many times this season I've been shouting at the screen to pit someone when they are losing lots of time and yet they stay out. Conversely, I've been surprised at some pitstops when pace and tyres seem ok and they come in. I guess the computer has already defined the pitstops. Ok, I know testing is banned and teams are having to rely on simulations more and more but I'm sure race strategy is too pre-defined nowadays.

I'm a McLaren fan but I have this sinking feeling that one day, in say 100 years, they'll be unbeatable because they'll have seen every single possible combination of factors and will have developed processes, procedures and programs to optimise all of this and come up with the right answer every single time. Unfortunately they will have lost many many races up to that point simply because processes and programs simply hadn't allowed for what was happening in front of their very eyes and they trusted the computer too much to change it on the fly. Seems unbelievable to me to enter a race (India) knowing that you're starting on a tyre that you haven't been able to get working all weekend and you then run lots and lots of laps on it in a very heavy car when track conditions (in terms of the race) are at their worst.

One thing's for certain though, if Ferrari do exactly what RBR do for the next three races then RBR will win because all things being equal RBR have the better car. So mix it up, try something different and let a driver really go for it.

[mariner]

I have no idea what the F1 teams actually do in terms of mid-race decisions but simulation models to assess options using artiifcial intelligence from actual data have been available for litearlly decades. Originaly the necessary computing power limited field usage but with modern PC power and data uplinks I can't see there would be any computing limit to lap by lap simulation.

You need historic data to make it work (GIGO) but you can use that to generate ranges of options beyond simple look up of exactly the same situation from before.

Given the brain and computing power inside F1 teams I am pretty sure they simulate " on the hoof" as it were . However all that has to be actioned by a human being ( so far) and thats a judgement call. Some people are better at those , espeically in real time, than others.

Military generals would be a parallel - all the UK ones went through Sandhurst and all the US ones through West Point. The same sort of field communications exist for all of them but some turn out to be much better at winning battles than others.

[Greg Locock]

Originaly the necessary computing power limited field usage but with modern PC power and data uplinks I can't see there would be any computing limit to lap by lap simulation.

Full physics full vehicle models of full complexity won't run in real time on one PC, although if you had a custom solver and dedicated one cpu to each tire you might get close. I was told by my predecessor here that some Nascar teams used to run such rigs, but (a) he used to work for some racing teams and (b) race team members make stuff up.

But the great news is that that is overkill. I'd be amazed if any second rank teams bothered with anything more subtle than the usual trends based on fuel weight and tire condition. If your laptimes don't follow the expected trend then start to worry. if that is correct then you could use excel for your modelling. ack.

I don't have the faintest idea how the in-race pitstop strategy stuff is done, that would be interesting.

[Ross Stonefeld]

Some over-simplified stuff here http://www.wired.co....erview?page=all

[packapoo]

Too much data? Possibly.
Lack of testing, real time that is, forces teams to simulate I feel.

[BillBald]

I think that part of the problem is that teams have to set gearing (and downforce levels) according to whether they expect to be doing any overtaking.

In India the McLarens were not fast enough in a straight line, so I don't think that 2-stopping would have worked well.

On the other hand, if the tyres are in good shape and the pace is good, it is almost never going to be a good idea to pit just because that is the strategy you started with. The later you can leave your tyre change, the less likely you are to drop into traffic, and the more you can shorten (or even in some cases eliminate) the final stint. In this case, changing your strategy will almost always work out well.