...and what to do about it

Yesterday, we saw Force India choose to not run in Q3 at Singapore. This is the second time this season they have taken this option. In DRStrategy, this was recommended as the solution for a team which cannot seriously compete for pole in Q3. There was no possibility of Force India managing a much higher position if it ran than if it didn't, so it was sensible to save a set of tyres to preserve strategy and perhaps pass the more profligate Mercedes team through strategy.

This may seem a rather mercenary attitude, but remember Force India is in a battle between Renault and Sauber over 5th place in the championship. Sauber is very close behind Force India, so it needs points. In addition, Renault is having a lousy weekend in Singapore, making this the best opportunity Force India is likely to have to close the gap between the two. Logically, Force India has to prioritise racing well to qualifying well.

So all is well? Not to judge from the large number of disappointed fans. The best I've seen of their reaction came from @LewisBarthaud:

something needs to be done, this 3 phase format came about to improve the show, you can't call it the "top 10 shootout" with only 7

 

You can't argue with that logic. Also, there's an emotional argument which presumably wouldn't condense into the remaining characters.

There is a massive visceral pleasure in seeing a racing car on track. When there are many racing cars doing likewise and competing powerfully with one another, the pleasure is multiplied. So to take it in reverse, removing 30% of the expected competitors will take out an average of well over 30% of the excitement (exact percentage depending on exactly who "forfeited" the session and who's watching).

This is an example of perverse incentives - the regulations, designed to provide excitement and happiness to the multitudes who watch F1 racing, create the exact opposite effect in qualifying if played out to their logical conclusion.

The first thought might be to penalise those who do not run in Q3 on purpose through a grid penalty. However, this ends up creating perverse incentives of its own. One of the main ways people have been able to overcome the zero-sum passing engendered by DRS is by having more sets of unused soft tyres. What this means is that everyone will, barring unforeseen punctures or crashes, everyone will have the same number of sets of soft tyres remaining. By insisting everyone must have the same number of unused sets of soft tyres, the racing will become even more predictable and samey - under the rippled surface of inconsequential passes. Only someone setting their fastest time with hard tyres will be able to break the cycle.

So that method of making a more exciting qualifying would make a rather dull race. However, this is not an insoluable equation. Changing the way tyre allocations work may help.

My proposal works thus:

- 3 "hard" sets given out on Friday. This will encourage race set-up and endurance work, shift tyre comparison work to Saturday, possibly encourage more teams to use unproven drivers. More importantly for the specific problem under discussion, it means a "soft" set can be awarded later in the weekend without making Pirelli bring any extra tyres. You've got to adapt to cost-cutting...

- 1 "soft" and 1 "hard" tyre set given out at the start of Saturday

-  1 "soft" and 1 "hard" tyre given out at the start of Q1. Any of the three sets of softs given out thus far may be used in qualifying.

- 1 "soft" given out at the start of Q2 - but only to drivers who set a time on the soft given out in Q1.

- 1 "soft" given out at the start of Q3 - but only to drivers who set a time on the soft given out in Q2.

- 2 "soft" and 1 "hard" set given out on Sunday - but only to drivers who set a time on the soft given out in Q3 (or Q1/2 if they were eliminated there). Q3 runners will be permitted to use this instead of the tyre they did their time on, should they be eligible to receive such tyres. Teams will hand back 3 "hard" and 2 "soft" sets of their choice, leaving them with the same number of tyres for the race as they have now.

Stewards' discretion will be used for anyone who makes a genuine attempt to qualify but has a technical issue on-track, crashes on their "sighter" Q1 run on hard tyres or during their soft-tyre run. The idea being that genuine accidents would be treated the same as people who were knocked out at the same stage but completed the soft-tyre run. Anyone of whom foul play is suspected (or had such serious problems that they  wouldn't have used up much/any of the soft tyre's longevity) will receive no such privileges and be treated the same as those who never attempted the run.

This would mean qualifying would influence how many tyres on has available in a more sensible way.

• Someone who didn't run in Q1 on soft tyres would have no unused "soft" tyres.
• Someone who chose not to run in Q2 or Q3 on soft tyres would have 1 set of unused "soft" tyres (the one they got for running in Q1 on softs).
• Someone who chose to run soft tyres in all sessions for which they were eligible would get 2 sets of unused "soft" tyres (for running in Q3).

 This should re-align incentives for the race in favour of having both an exciting qualifying and an exciting race.

The one thing this doesn't prevent is a team not running because they don't think they will lose anything, despite the disincentives in place. This is the frustrating position I am in with my other favourite team, AF Corse. It is in Portugul for the Le Mans Series race but due to some terrible luck involving a lorry accident, a broken tail-lift and an hastily-completed new car presenting problems, is apparently comtemplating a deliberate non-finish for its hitherto most competitive and popular car (the #51 driven by Giancarlo Fisichella and Gianmaria Bruni). It's already team champion, with the #51's drivers also confirmed as individual champions, so they not particularly worried about points. The car is in one piece and is very fast when it works, but could break at any moment and the staff all need to be in America as soon as possible because Petit Le Mans, part of an Intercontinental Le Mans Championship that the team hasn't fully secured yet, is next week and everyone on the the team needs to be there for early scrutineering. I can see the logic but am still upset with the idea a team might forefeit a race where it has somehow managed to qualify 2nd. Much as the people watching the logical withdrawal of the Force Indias from Q3 were upset about them "forfeiting" the last bit of qualifying when they were good enough to at least set representative times. 

Intellectually logical, emotionally tough to accept...

Warning! Long entry alert!

This blog entry was prompted by TEF20's comment in Formula 1 Blog, which in turn was about Ferrari opposing the 4-cylinder engine concept mooted by the FIA for 2013. About a third of the way through replying, I realised this "comment" was not only too long for a comment (2549 words are too many, even by my reckoning!) but only tangentially relevant to the topic. Therefore, it was moved to my blog.

TEF20 was considering in his comment how reducing the amount wind tunnels are used in F1 could allow more testing to happen, therefore making a lot of fans happier because they can see more of the cars (among several other benefits). It certainly sounded an intriguing idea, so I decided to try mathematically test it - with mixed results.

Firstly, it is true that wind tunnels consume colossal amounts of electricity (thus carbon) whenever they are used. We are not talking hairdryers pushing air through plastic sweet tubes onto miniature models here. The vast and vastly powerful variety used in F1 use enough energy that the electricity companies demand to be told before teams turn them on so that demand can be planned. The two tunnels in Brackley (for Mercedes and Force India) probably consume more electricity than everything else in the 12.848 km between Brackley and Silverstone put together (I have no data to confirm this, but if the steelworks in Sheffield, working at full blast, can consume more electricity than the rest of the city put together, then it is surely reasonable to suggest that two powerful industrial energy-guzzlers can use more electricity than the sum total of a few villages).

Secondly, none of the foregoing means that track tests are minor consumers of carbon either. It takes a lot of equipment to conduct a test.

Test teams take at least three lorries with them, alongside cars or buses capable of carrying 45 staff. The lorries would use 17.082 g/km per ton. If the team went in buses (probably the most ecologically sound way of taking that many people), each bus would use 10.249 g/km per passenger.

The Ecology of Test Travel

If we assume that the lorries are 20 tonnes each including their loads, then that gives us 683.28 g/km for the lorries and 461.205 g/km for the buses - or a total of 1144.485 g/km.

So taking Mercedes as an example, quartered the relatively short 12.848 km away from Silverstone. They would need 29.409 kg of carbon emissions to get to the test and back, assuming the test got rained/frozen off.

If a team like Mercedes wishes to avoid that fate (frozen/rained-off testing being completely useless in terms of data collection), it is likely to go to Spain. That means sending everything through the Eurotunnel, down the length of France and down part of north-east Spain to reach the nearest track (Barcelona). I will assume for the sake of ecological sensitivity that nothing is flown. And then return, of course. The distance is 1752.608 km, therefore the carbon emissions would amount to 4011.667 kg, or 4.012 tonnes. Again, before running or indeed the travelling between hotel(s) and test track/travel routes that would be required to make a road-only test work.

If we assume the team must go to a hotel if they wish to stay overnight in Spain, and that (as is common) they stay at the cheapest nearby hotel (Hotel Ciutat at the moment), they need to travel 17.952 km extra per overnight stay. If this is done in buses, that's another 8.280 kg per night. They also need to stay there before and after each test so that they don't waste the first few hours of testing and/or risk total collapse trying to return to the UK. This makes a typical four-day Spanish test (as generally practised in F1 these days) 4.053 tonnes per team.

This mounts up. Each F1 team used 62.5 tonnes on average for fuel in races and tests in 2009. but most of them went to Spain for 4 tests prior to the season beginning and then did 8 local tests in-season. That would be 16.221 tonnes just for the current testing regime (races are more efficient because one Bernie Air plane can take multiple teams' stuff - even though the method of transport is more wasteful per item, less fuel is needed overall due to the airborne equivalent of carpooling).

The Ecology of Wind Tunnel Testing

The entire electricity consumption of the average team for 2009 was 365.412 tonnes. This includes the vast amounts necessary to power the CFD systems currently in place and all the manufacturing equipment as well as the windtunnel situation (teams are only allowed to use one less-than-60%-scale tunnel at a time apart from 16 hours of full-scale tunnel time per year, but there's no rule saying a team can't use one of its tunnels while the other one's going through a maintenance/calibration procedure, for example). If CFD, manufacturing and other non-wind tunnel uses for electricity are assumed to take up 50% of the electricity teams use, that puts wind tunnels on 182.706 tonnes/year. Unfortunately, the data does not exist to tease out how much wind tunnels specifically use.

Provisional Comparison Between Wind Tunnels and Track Tests

Local tests are pretty efficient uses of carbon dioxide. A team could test in its backyard every day for an entire year and not use up half as much carbon dioxide as it would doing one test in Spain. Unfortunately most teams are British and the neighbours would complain.

Testing on the continent when a team is based in England is much more of a problem. Large teams used to do three-day tests every fortnight in-season with a test team that was almost completely separate from the team that went to races, which would clearly be far more expensive than the amount used in factory testing procedures. Even if it were more ecologically efficient, it would not be possible to transfer to such an arrangement in the current climate.

How many four-day long-range tests could a team do with the carbon it would save from relinquishing the right to use a wind tunnel for six months of the year (this would allow the team to still use the tunnel to create its cars, but not to develop them)?

Six months of windtunnel on my back-of-blog calculations would be 91.353 tonnes. Divide that by the 4.053 tonnes each long-range test takes and you get 22.53. Which means that you could do just over twenty-two-and-a-half four-day Spanish tests for the price of having the wind tunnel on for 6 months.

How convenient. That's one for every race of the season with a little bit left over.

Or it would be had FOTA not set themselves a 15% carbon reduction plan to be reached by the end of 2012. So let's take 15% off that and see what we have.

19.15.

Still almost enough for one four-day test after each race of the season. Which would be an interesting idea for enhancing development while massively cutting carbon emissions. CFD-only development would be more efficient than CFD + wind tunnel development in terms of carbon and could be a step towards removing wind tunnels from the equation entirely. Fans would love to see their heroes honing their cars carefully for a relatively low fee, teams and circuit are already set up to accommodate them and it also cuts a lot of the expense connected with travelling because they're already there.

The Villain of the Piece

There is one carbon-related problem though. To test, one must have extra pieces to test. They have to be manufacturered and transported.

There are two forms of carbon source to be considered. In-house consumption of electricity for manufacturing comes from the same 365.412 tonnes of CO2 as the wind tunnel. We've already removed the wind tunnel through a back-of-blog guesstimate of 182.706 tonnes. There are three things using up the rest of that carbon dioxide - CFD, manufacturing and sundry expenses such as lighting and heating. If we say that CFD is 60% (there are some powerful computers used in design work) and the sundry is 10%, that leaves 30% of the non-wind tunnel expenditure - or 15% of the whole - as manufacturing expenditure, making it 54.812 tonnes per team. Materials manfacturing is 613.892 tonnes per year for the average team - just over 50% of the total carbon consumption. So the total carbon used for components per team, on average, is 668.704 tonnes. These do all the races as well as testing and also counts manufacturing errors that never make it to the car in any capacity whatsoever.

That Which Is Obsolete and Useless

Every non-homologated item on the car can be expected to change at least once in a year for those teams that can afford to do so and see some point in the attempt. Some teams' wings changed 10 times in 2010, but that's an extreme. Probably more accurate is to assume for the sake of calculation that everything changes an average of twice a season. When testing enables it, all of these changes will be tested.

Not everything that gets tested makes it onto the car. At the moment, it seems to be accepted that most teams bring along 4-5 major upgrade packages per season to have one of those fail, giving a minimum of 20% failure rate. Extra testing would enable teams to tease out which bit of the package is failing easier, perhaps dropping the failure rate to 10%. The problem comes when one considers the amount that can be tested in a particular test - and therefore rejected.

The amount varies massively from test to test because the agenda could be anything from back-to-back testing of a complete new package against the current one - a process that generally results in one package or the other being fully rejected - to incremental honing and systems checks, both of which involve negligible part rejection. In the most through version of the procedures, it takes three back-to-back runs - 9 laps or about 15 minutes excluding analysis and preparation time - to ascertain whether a component worked. Of course the teams do quite a bit of analysis before accepting or rejecting components, but usually other things would be tested during at least part of that analysis. However, there's simply no way round the fact that parts take time to fit and remove, or that the car will need a wipe-down at the very least after each run to maximise data parity. The preparation figure ranges from about 5 minutes if it's simply a wipe-down plus fitting a new front wing to the best part of half a day for certain combinations of internal component (especially if removal requires the fuel tank to be adjusted in any way). Half an hour average preparation is probably a reasonable ballpark figure, with the caveat that it varies massively according to the components involved.

There are about 80,000 components in a F1 car, but that's not the best figure to use for this set of calculations. Some things (like the engine, which accounts for many of the components) are homologated and therefore can't be modified by the teams. Also, some components (such as fuel tanks) couldn't meaningfully be tested by the teams in this way - they'd be tested using other methods, most often by the third-party suppliers who produce such specialised kit. Finally, some components are typically tested in clusters (come on, have you ever heard of a team taking a gearbox, putting in a different 4th-gear dog ring and then running the car again just to see if the new dog ring is an improvement on the last one?)

Time Is Carbon

Taking another angle, if it takes about 45 minutes to do the running and preparing to test a typical component/cluster and a test typically runs for 7 hours (9-5 am excluding an hour of lunch), then it would be possible to test 9 components/clusters in a day, or 36 per 4-day test. 3.6 of these will prove failures (in that they won't be better than the previous versions).

It's not so simple to derive the average carbon cost of a component because it's not clear what proportion of manufactured components don't make it onto a car. However, if a 1% failure rate is assumed for manufacture (probably an underestimate) and it is considered that teams typically use six chassis in a season, replacing an additional component every 20 minutes a car runs, then a rough estimate can be derived.

The six chassis would be 480,000 components. There were 17 races in 2009, 4 four-day "long-range" Spanish tests and 8 one-day local tests. A race consists of 4 hours of practise, 1 of qualifying and up to 2 hours of racing, so it would be fair to say it has the same mileage as a one-day test, just spread over a longer period of time. So that would be 41 days equivalent of running, with 369 components/clusters tested (and therefore 36.9 proving "failures") and 861 components needing replacement - on top of the components already on the cars. That's a minimum of 481,230 components produced per season excluding failures, meaning at least 4812 components which failed at manufacturing. So the total amount of carbon used in manufacturing must take 486,042 components into account.

From this, the average carbon estimate per component is 668.704 / 486,042 = 0.001375815 tonnes/component or 1.376 kg/component. To put this into perspective, each extra component needed for a F1 team is just under a third of the carbon consumed for travelling to a local test.

This is significant.

The Components of (Climate) Change

If you assume that the six chassis and associated extra components were meant to do exactly the 2009 season and no more, then an adjustment to the carbon figure of extra testing becomes possible.

486,042 / 41 = 11854.683 components used per day.

11854.683 * 1.376 kg/component = 16312.044 kg/day = 16.312 tonnes/day.

16.312 + 0.029 = 16.341 tonnes/test (for a local test)

16.312 * 4 (for a 4-day "long-range" test) = 65.248 tonnes/4-day test.

65.248 + 4.053 = 69.201 tonnes/test (for a long-range test)

So how much testing can now be fitted into that six months of wind tunnel time (in terms of carbon emissions)?

91.353 / 69.201 = 1.320 four-day "long-range" tests. Let's call it 1 and a remainder to see if we can get any local tests out of the remainder:

22.152 / 16.341 = 1.355 local tests. That would be another 1.

At this point, converting windtunnel usage into extra testing time ceases to make much ecological sense.

Conclusions

We've established (albeit with rather more back-of-blog calculation than I'm entirely comfortable with doing) that wind tunnel testing is more efficient carbon-wise than track testing and that the reason is not, as one might suppose, travel, but the manufacture of components to enable that testing to occur. The average team, under normal circumstances, surely gets more out of six months of wind tunnel testing than five days of track testing. The question is: do we, the fans, get as much out of those five testing days as out of the six months of wind tunnel work?