Forget the track, F1 is about the data
We take a look inside the digital world of the Renault F1 team with Microsoft. By Damien O’Carroll.
More than any other, Formula 1 is a sport of data. The racing matters, the people matter, the competitive spirit matters, but without the data none of that would happen.
F1 operates on the cutting edge of what cars and people are capable of, so getting that data and making the most of the information it provides is the lifeblood of the sport in this day and age.
While an F1 car wouldn’t run without fuel, wouldn’t go around a corner without a driver, or wouldn’t even start without an engineer, without data it wouldn’t win.
The sheer amount of data a F1 car generates over the course of a race weekend is mind boggling – there are more than 200 sensors on the car reporting billions of separate data points back over the three days of an F1 race weekend.
Interpreting, sharing and storing this data is a huge job, and one that most people don’t even think of when they are huddled on the couch at an obscene hour of the morning watching Mercedes win yet again in a distant part of the world.
So how do they do it? Well, we’ve just had a first hand look at that process, thanks to Microsoft and the Renault Sport F1 Team at the Singapore Grand Prix.
Microsoft first partnered up with Renault back in 2012 when the French manufacturer was only an engine supplier, supplying the Lotus F1 team.
The deal initially was for Microsoft to supply an ERP (Enterprise Resource Platform, basically business management software) to the team but it has grown considerably since then with Microsoft now providing the team with a wide range of software and computing solutions, but it is the latest step that has everyone at Renault Sport and Microsoft rather excited – Digital Transformation.
Digital transformation can be thought of as the ‘‘third stage’’ of embracing digital technologies, with the first two being digital competence and digital usage. "The most difficult element of the car to simulate are the tyres. Those pieces of rubber behave in a very, very tricky manner. They change depending on the temperature, on how aggressive the driver is, the weather conditions and all of this quickly becomes difficult to enter into the simulator." Pierre d'Imbleval Think of it as going paperless but on a far larger, more ambitious scale.
For the Renault Sport F1 Team, this means embracing the cloud in a big way.
‘‘Data is everywhere in the world of F1,’’ says Pierre d’Imbleval, chief information officer of the Renault Sport F1 Team.
‘‘On a race weekend, our cars generate more than 35 billion data points equivalent to about 30GB worth of data. Our supercomputer generates 60TB of data per week – that’s enough storage for 25,000 hours’ worth of movies.’’
Over the past two years, Renault and Microsoft have been working on cloud-based solutions for dealing with all that data, with everything from products like Office 365 and Skype (which saw travel between the Renault Sport facilities in France and the UK drop by 50 per cent and phone bill savings of more than 30 per cent) up to the heavy hitting business software like Power BI and Dynamics 365, which will shortly be used by the team to track every single part of the car throughout its entire life.
While getting the vast amounts of data that the team produces each weekend into the cloud drastically improves collaboration between the three sites (the Technology Centre in the UK, the Hybrid Power Units Centre in France and the track), allowing engineers at head office to access data in real-time while cars are on the track, it is the massive potential to improve machines using data alone that is going to be driving F1 relentlessly forward in the next few years.
‘‘Because we are not allowed to do much physical testing, we now have to tackle that with the simulator and we have to set it up to be as realistic we can,’’ says d’Imbleval.
‘‘There are elements that are relatively easy to simulate, like the track. We have all that data by scanning the tracks on a regular basis.
‘‘The most difficult element of the car to simulate are the tyres. Those pieces of rubber behave in a very, very tricky manner. They change depending on the temperature, on how aggressive the driver is, the weather conditions and all of this quickly becomes difficult to enter into the simulator.
‘‘So we are able to learn over time how the tyres degrade by pushing two entire seasons of data into the cloud and start to compute that data, not to reproduce what happened last year, but to determine some generic behaviour of the tyre degradation to put into the simulator for the driver to get some sense of it.
‘‘It’s to predict with a machine learning algorithm how a tyre will degrade with a specific driver at a specific track.’’
Because F1 doesn’t yet place any restrictions on computer usage away from the races, teams are able use ‘‘from one to a thousand’’ servers to run simulations during the course of a race weekend.
D’Imbleval says that since the move to the cloud, the team has gone from running 5000 simulations a day over a race weekend to a simply staggering 75,000 simulations a day.
But while all this technology is helping F1 cars go faster, there is one thing that d’Imbleval is certain it will never replace, and that is the driver.
‘‘There is much talk about this so-called ‘robo-race’ where you have computers driving the cars. It is the ideal world, with cars running on the track with no drivers,’’ he says.
‘‘But where is the sport? Sport is about human beings. We want to keep F1 as a sport, and to be a sport you need human beings.’’