Williams Martini crunch biometrics for Formula 1's peak pit-stop performance

One of the sport's most important races takes place off the track. And it's data that decides who wins.

It took less time than it will take you to finish this sentence.

At the Grand Prix of Europe in Azerbaijan last year, the Williams Martini team officially recorded the fastest ever pit stop in Formula 1 history.

Pulling into the pit box at the end of lap seven, Felipe Massa came to a stop. Less than two seconds later he was speeding away on a fresh set of tyres. Just 1.92 seconds to lift the car on two jacks, unscrew each tyre, remove each tyre, replace each tyre, screw them firm and release and remove both jacks.

By the end of the season Williams had clocked the fastest pit time in all but six of the season’s 21 races.

But that’s not good enough. The team wants to better its speed and retain its lead in the pits. The solution lies not in mechanical improvements, says team CIO Graeme Hackland, but in analysing the crew’s biometric data.

“I’ve heard Felipe Massa say ‘no matter what I do in that car the engineer knows’,” Hackland explains at the Williams Martini garage at the Australian Grand Prix in Melbourne on Thursday morning.

“We know that car inside out,” he says. “The car is completely instrumented, but we’ve never put any instrumentation on the humans – up until now.”

The fastest stop

Williams Martini was one of the slowest pit crews in the 2012/13 season, and the team considered it their Achilles heel. Often the pit turnaround was taking them one-and-a-half to two-and-a-half seconds longer than the competition.

Improvements were made to the cars suspension system to make it easier to remove tyres, and the crew and drivers dedicated themselves to practising the process and hired experts to help them better orchestrate the now seamless movement. As driver Valtteri Bottas put it: "Formula 1 is definitely a team sport. The pit stop is the finest example of that."

The work paid off and last year the team won the DHL Fastest Pit Stop Award.

“It really helped in a season where we under-performed on the track. It helped to show that we are still a good team, we’re still working hard, we can still be good at something even if the car wasn’t always that great,” Hackland says.

Other teams are closing in fast. There are rumours going round the paddock this morning that a rival group has set-up a project specifically to topple Williams reign in the pit this year. Retaining the title, Williams believes, will come through fine tuning the individual members on the crew.

“A car has 200 sensors logging over 1,000 parameters, and we wanted to expand the data available to us by understanding the science behind the pit crew,” says Williams’ human performance specialist, Gemma Fisher.

During the 2016 season, the pit crew wore bio-harnesses underneath their fire-proof suits to track a range of biometric data including heart rate, breathing rate, temperature, and peak acceleration.

Measurements were uploaded to a database hosted in Microsoft Azure and processed with Microsoft Power BI data visualisation tools.

Microsoft solutions and managed services provider Avanade, a partner and sponsor of the Williams Martini team, developed a dashboard to display the biometrics analytics. The aim is to provide insight into the physiology behind the perfect pit stop.

Initial findings have suggested that stress and caffeine intake have a significant effect on pit-stop performance. The information has also helped Fisher to draw up each crew member's fitness training objectives and diet.

Eventually, bespoke reports will recommend a particular heart rate for a crew member to reach by the time a car comes into the pit, to give an optimum performance and reduce the cardiovascular recovery period.

“It will enable us to continue to push the boundaries of pit stop potential and help us use additional data streams to ensure we stay a step ahead of our rivals,” Fisher said.

Predictive makes perfect

By using predictive analytics, crew members could potentially be pulled out of proceedings if a substitute was in better form, says Hackland. However, that will require consideration of their feelings.

“They’re used to doing it in football – when a player gets into the red zone they get subbed – some don’t like it and they’ll kick the water bottle as they go past. What’s it going to be like for these guys who train and absolutely want to be there? At the point that we get to the predictive piece, where we say ‘well we think you’re going to have a bad pit stop, so were substituting you’ is going to be interesting to see,” he says.

“These guys train constantly. If you’re the right front wheel gun man, that’s your job and you practice and practice and practice. Making that decision to sub someone out is going to be tough.”

As well as predictive physiology, the team is also exploring the use of machine learning during races on the pit wall – to aid engineer and strategist decision-making. In each case, allowing machines to take authority will be down to trust Hackland says.

“It needs to get to the point where they trust it. That it’s making the right decision more often than they are. And we’ll be able to prove that in the data, and then they’ll trust it,” he says.

This season, and increasingly each year, data has become the decider, says Hackland.

“The race is not just on the track it’s everything else that we do. Data is the new battleground – I think the use of data is the biggest differentiation between the teams. The more resource you can put to that the more successful you’ll be.”

George Nott travelled to the Australian Grand Prix in Melbourne on Thursday courtesy of Avanade.

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags predictive analyticsMelbournebiometricsdata analysisdata analyticsFormula 1Formula OneWilliams MartiniWilliams Martini Racing

More about Microsoft

Show Comments
[]