Big Data Analytics make Formula 1 cars go faster
An analogy of big data analytics and formula 1 car racing to understand big data in real world
Formula One (F1) motor racing is among the top 3 richest sports in the world with some F1 drivers often being the highest paid athletes internationally (Michael Schumacher reined in $59m in 2000, and listed by Forbes as the highest paid athlete in the world that year); Lewis Hamilton is currently the highest paid sportsman in Britain.
The sheer adrenalin rush of watching an F1 machine on 4 wheels hurtling down the straight at a speed in excess of 330kmh and often cornering at more than 250kmh brings in more than 100k fans to watch an F1 race; even in a populace country like India where lakhs of people congregate for events, an excess of 1 lakh paying a minimum of Rs12 000 ($165) per ticket is mind-numbing indeed. This year, 2020, is the 71st year of F1 motor racing, it’s grown into the 3rd most technologically advanced industry in the world, trailing only that of the space and weapons industries, but definitely ahead of the airline industry. In 2018 Forbes stated that the total budget of the participating 10 teams was $2.6 Billion (USD) – that is a staggering amount, some countries’ fiscal budget is less than that! It’s an industry where the difference between winning and losing is measured in nano seconds. The entire industry, not surprisingly, is now being permeated by BIG DATA ANALYTICS, which has now evolved to become the key to gaining fractions of a second advantage from race to race, from circuit to circuit, from car to car, from power unit to power unit, in different weather patterns, in different climatic conditions and atmospheric pressure
Here’s an interesting fact :
- More people have been to space than have driven an F1 car !
The use of electronic technology in F1 cars began in the ‘80s’, but from the ‘90s’ F1 began investing heavily in R&D to develop cutting edge technologies in electronics, composites, nutrition, training regimes, logistics, aerodynamics, tyre change equipment, and more, to become the leading industry to unlock performance enhancing gizmos. The most advanced and sophisticated sensors are used in F1 cars – about 200 to 300 per car during testing on a Friday and Saturday! Pirelli, the racing tyre manufacturer, the biggest supporting industry to F1, also invests heavily in R&D year after year to improve the balance between speed and tyre endurance. Up to about 1800 tyres are used for each race by the 10 competing teams. Constant innovation is the name of the F1 game such that no team is able to drop the ball even momentarily for a single event; there is a close correlation between R&D spend and performance – it is well known that the big 3 F1 performers have the biggest budget.
Did you know that a typical F1 car has about 80k components, including body parts ?
- That’s a result of data driven insights, often small pieces of body work leading to a tiny fraction of performance improvement
The Manager of IT at Mercedes-AMG, the team that has won both the driver and constructor championship for the last 5 years in a row, has said that Data Analytics has become a critical factor in Formula1 racing. In 2017 he further spoke about data analytics, that “It’s difficult, because nobody talks about it too much, everybody knows how important it’s going to be.” (Datanami.com 19/4/2018). During test runs almost every part of an F1 car is monitored by sensors. They measure lap times, tyre and brake temperatures as well wear, airflow, engine performance, gearbox performance, downforce (that magic factor that keeps the high-speed car glued to the road in the bends), driver health, driver G-force, each of the 4 suspensions, ride height, pedal movement, exhaust temperatures, clutch fluid pressure, oil and water levels, engine RPMs and so much more. On race day the car is then set up to provide what the engineers and driver deem to be the right balance between straight line speed and downforce, making small tweaks based on ongoing data gathering. Each race weekend generates TB’s of data per car in real time (includes video streaming data)! Team principals are glued to the telemetry screens during the operations of a car (tests, practice, qualifying sprints and race). In no other sport is there so much of data analytics employed as in F1 – the ability to obtain insights from the data determines changing strategies during the course of a race; the blazing speed means that there is no time to be tardy with decisions! Data is streamed live from the car to the pit wall from where it is transmitted to the factory – the role of Cloud computing in F1 – with secure feeds to factory-based mechanics and engineers. An army of Data Scientists and auto engineers sit huddled around the main dashboard, constantly communicating with the pit wall. While the tyre manufacturer may state that the hard compound will last, say, 60 laps but will run at about 1.2 seconds per lap slower than the soft compound, teams hardly ever take this at face value. This is the reason for 3 x 1.5 hr practice sessions before each and every race – 2 on a Friday and the last on Saturday morning. During the practice sessions they build up data on each type of tyre, on minor parts changes, on brakes, etc, so that during the qualifying sprint and on race day they can take strategy decisions confidently, based on the insights gleaned from real data during the test runs. Tyres are absolutely critical to performance – the drama of winning or losing is so often dependent on the tyres. Even minor changes to, say, a wing, may change the tyre life. It is the insights drawn from the data analyses from both the factory and the track tests that determines the starting strategy as well; one such strategy would be whether to sacrifice front grid start position set during qualifiers for midfield performers, understanding that for good starting positions starting on the fast tyre is axiomatic, however the fast tyre is a soft compound that degrades quickly; the medium or hard compound may be a bit slower, but it can run longer saving the time of an extra tyre change. Without reliable data, such a crucial decision would be a hit or miss affair – and when it comes to a billion-dollar sport, no team dare take such decisions based on whims and fancies – just like businesses who have to take decisions that lead to success – mistakes can lead to the ruin of a company! One driver who has made it a habit of sacrificing a good qualifier run for a good race finish is Mexican Sergio Perez (Team Racing Point, formerly Force India) – he notoriously starts below 10th place on the grid but invariably ends up with a decent haul of points at the end of a race. Data analytics also shows which drivers have a better record of saving tyre wear during a race. Cavalier drivers have to undertake more tyre changes – and depending upon the circuit, between 20 and 26 seconds may be lost during a tyre change. Note also that every practice session, qualifier and race is either televised or streamed live – this adds another dimension of data to the data lake, that of videos – the analytics workload therefore also involves mountains of video analyses – the need for the right algorithms and scripts, together with a range of analytical tools, is the very essence of Data Science in the F1 racing industry!
It isn’t only the race day real time strategies, based on the live data streaming that determines preparation, car balance and strategies. Almost every aspect of F1 racing is highly dependent upon big data analytics. Right from R&D, design, exhaustive testing of prototype (parts and whole cars – in the wind tunnel), suspension and chassis manufacture, front and back wing manufacture, power train manufacture, assembly, through to track testing – all of these aspects of F1 racing are based on reliable data analytics. The mountains of data gathered at each race has also provided the means to create virtual reality simulators that closely simulate a real qualifier and race. Midway through the 2019 season, Red Bull Racing decided to demote Pierre Gasly and promoted junior driver Alex Albon who was racing in F1 for the first season. Albon, with no previous experience in the circuits in an F1 car, practiced diligently on the simulator for each circuit – and finished very strongly (7th out of 20 positions) at the end of the year. (The simulator has also spun a new industry – Esport Racing – virtual reality simulators for gamers – yet again another industry that employs Data Scientists). It is Data Science that dictates how the cars are built, how they are driven, and how the race is run these days.
The mountains of data collected during race weekend is used to reengineer the F1 car by running the data in the simulator to improve the car’s systems. Team McLaren* claims that it can achieve more in one day of data simulation than it would in a week of on-track testing (Datanami.com 19/4/2018). Pitstop times have been drastically reduced through the analyses of videos of actual pitstops recordings – some teams are able to change 4 tyres in under 2 seconds! Here in F1, more than, perhaps, anywhere else, big data drives big decisions. As the reach and application of data analytics proliferate the wider world, F1 is bound to take advantage of whatever the tech industry throws up – with cloud computing, predictive analytics, predictive intelligence, machine learning, prescriptive intelligence, and, dare I say it, AI, all playing larger roles in the future of the sport – wonder if F1 would ever go the route of driverless cars….
Data Science is a relatively new branch of Information Technologies that has become the core function in e-commerce, supply chain industries, financial sectors, large scale manufacturing operations, the Health Care industry, and the list goes on – wherever men, machines and materials perform work, commercial or otherwise, big data analytics is bound to become the growth and performance factor, if it isn’t already. Even governments are now employing data analytics to improve multiple areas of service delivery – where it’s needed, when it’s needed, in the correct level. The demand for data Scientists has been predicted to grow exponentially: