Next to the F-22 fighter plane and the Space Shuttle, a Formula 1 car may well be the most sophisticated piece of high-performance machinery humans make.
Although its performance has actually been limited by the Federation Internationale de l’Automobile, the sport’s governing authority, the typical car is capable of 230 mph. Even more remarkable than its straight-line speed is its quickness and agility. Zero to 60 mph in as little as 1.7 seconds. Zero to 100 mph and back to zero in less than five seconds. Gear changes take place in 40 milliseconds, literally the blink of an eye.
If you leave the planet in the space shuttle, you’ll feel 3 Gs, three times the force of gravity. The forces created by a Formula 1 car cause the driver to experience corners and braking at more than 6 Gs.
Airplanes use wings to generate lift. Formula 1 cars invert the wings to create a “downforce” that presses the car into the track. This downforce can be as much as three times the car’s weight. You could stick it to the ceiling. To squeeze the most out of the performance of their cars, the design and combination of wings are tailored and tuned by the teams for each individual racetrack.
Even the pit crews are remarkable. All four tires changed in 1.9 seconds.
The rules and regulations governing the sport evolve regularly and cars are updated accordingly. It’s not unusual for the major manufacturers—Ferrari, McLaren, Mercedes, Renault—to introduce a new car roughly every 12 months.
And it’s all made possible with data analytics. The cars and drivers and pit crews are equipped with a dizzying array of sensors, which keep track of hundreds of key performance indicators. Analysts will pull 2.5 terabytes of data from car, driver, and crew over a race weekend. That’s more data than all the books in the world’s biggest library, the US Library of Congress.
Of course, that data is recorded for future use—performance analysis and improvement, car redesign—but it’s also analyzed with the help of machine learning in real time. At the track, there are 15-20 engineers and data analysts working for each team and a hundred or more monitoring the race (at a .17-second delay) a world away at company headquarters. In an environment where winning and losing is measured in fractions of a second, any tiny gain in mechanical or human performance, any advantage earned in race strategy, is the difference between winning and losing. The analysts assess data to know exactly how the car is performing at each instant; they predict performance, run simulations, and prescribe variations in strategy. The Mercedes team upgrades a part for the car or makes a performance improvement, on average, every 20 minutes.
As impressive as Formula 1 analytics may be, what’s the relevance to business? The short answer is—everything. According to Joe Mariani, a research manager at Deloitte, “Formula One race teams are already experiencing today the technological and management shifts that mainline manufacturers will likely see in 5–10 years’ time.” By no means are the humans excluded from the process, but every decision is data-driven. The teams that make these decisions are highly collaborative and multidisciplinary, and the design and manufacturing process has been tweaked to the nth degree.
Simon Roberts, chief operating officer at McLaren Racing: “Compared to most automotive industries, we do ourselves every year what most big automotive companies will do every three to five years.” The racing season ends in November, but teams will start the design of a new car in September. By December the team has designed the 16,000 components that must be manufactured for the new car. In February, the first car is being built. Roberts again:
“Once we have built the car and start testing and racing it, we change the car about once every 10 minutes. Every 10 minutes we get a new CAD drawing out. That is a kind of relentless upgrade of everything. Normal carryover from year to year of about 3–10 percent is typical. But by the time we get to the end of the year, it is about 0 percent. The entire car is new.”
It’s Smart Manufacturing.
Today, it’s not unusual for a design team to create a product, then throw it over the wall to the manufacturing engineers, who throw it over the wall to the production personnel, who then push it out to product testing and eventually to the customer. Each part of the process is siloed, and there’s rarely a formal feedback loop. In Smart Manufacturing, designers, manufacturing engineers, and production personnel all participate in the creation of a detailed 3D model and this “digital twin” is evaluated through simulation, both on its performance and ease of manufacture.
And today, for the most part, we must put the experts and the physical product in the same location to troubleshoot issues of design, performance, and production. However, the digital twin and its manufacturing processes (also a twin) can be evaluated by experts anywhere in the world and improved upon literally in process. Once individual components are built, the actual performance can be tested and compared to the digital simulation and the simulations refined. It’s a continuous feedback loop and the continuous digital track of the component from design through manufacture and performance is called a digital thread.
McLaren Racing doesn’t sell cars or components to customers, but they do track each component from raw material through manufacturing to installation on the car. And, of course, they continue to collect data on the components as the race car is driving. Roberts:
“. . . our two race cars are actually stock locations on the system. So if you sat and watched our stock system on a race weekend, you can actually see parts booking onto the car or off of the car as mechanics make changes at the track. That also auto-records mileage, the number of starts, the time that part has been on a car to make sure that no part exceeds its life span or design limits. It is a bit like aircraft from that point of view.”
Before McLaren even gets to the track, they run digital twin models of the car around the track, testing nearly every mechanical and aerodynamic variable. Once the digital twin has been optimized, the driver steps into the simulation, rehearses his performance, and further tunes the twin to his style and preference. Those settings are folded into the racecar, and tested again on the practice and qualifying days before the race.
The hundreds of engineers and analysts, both trackside and at company headquarters in Woking, England, coordinate as a team, but predictably there is a division of labor. Trackside has an operational focus. Say the car strays too far onto the infield and suffers some damage. The trackside crew will look at the damage and the loads and determine, first, whether the car is safe and, second, whether a minor repair is possible. Back at headquarters, the analysts are running scenarios to determine how the damage will affect race strategy.
In a quirky parallel, things are much the same at another organization that relies heavily on data analytics, Coca-Cola.
The beverage company operates in 200 countries and over 23 million retail outlets, with 3,500 products and 700,000 associates. The scale is huge, and selling soft drinks worldwide is not a one-size-fits-all operation. There are variations in packaging, flavoring, sugar, and calorie content. Just like the
McLaren strategists in Woking, England, stay in touch with the race through the sensors on the car, the strategists at Coke stay in touch with their customers through the sensors on their vending machines.
Over the past several years, consumer tastes have been changing, leaning towards healthier alternatives. To keep pace with this trend, companies must be aware of and responsive to consumer preferences. Normally, a company the size of Coca-Cola might have some difficulty drilling down to the consumer level and responding to change with agility. But Coke has leveraged machine learning.
You may have noticed vending machines have changed. Whether they’re at the gym, or the mall, or in the restaurant, they’ve got touch screens and displays that are appropriate to the setting—different ads and products are promoted at the gym than at the mall. Behind the slick displays, those machines collect data on what we’re buying and how we combine a little lemonade with our iced tea. That’s how they identified the market for Cherry Sprite. There were no focus groups necessary, no extensive taste testing because consumers had already voiced their demand by what they bought, and those same consumers even helped the company zero in on the best recipe. The smart machines allow Coke to track buying trends at various locations—worldwide—recommending the most marketable combinations to be stocked, and tailoring their screens with ads to promote those combinations.
Coke will soon roll out AI bots—like Siri and Alexa. These bots can recognize the customer and personalize the user experience. The machine will remember individual preferences and on request blend a drink exactly to taste. The AI also enables the vending machine to present different faces and different voices tailored to the environment—a happy, energetic machine at a busy mall, and a quieter, more reserved machine in a hospital waiting room.
At the bottling plant, Coke has experimented with augmented reality. A headset superimposes graphics over what the technician sees in the real world, allowing a real-time status and performance report. If the machine in question requires repair, the technology can spot a problem and connect the technician with service experts who may be located on a different continent. Sounds like McLaren racing.
Harbert graduate Steve Mann is a systems manager at Coke. “With all of the data running through systems today,” he says, “plant managers and line specialists are looking for better ways to run things more efficiently. You have to be able to react to this data.”
Whether you’re an elite Formula 1 competitor or one of the world’s largest multi-national corporations, you’d better act fast on the best data if you want to stay in the lead.
Andrew Rains’ customers operate at high speed. Really high speed.
Rains, who earned a marketing degree from the Harbert College of Business in 2015, developed APEX Pro, a device that relays real-time data to motorsports enthusiasts to help them understand what their cars are capable of doing on the track.
In the racing business, everything happens in a hurry. You hit the gas and the torque associated with rapid acceleration pushes your body backward into the seat. Objects to your left and right become a blur.
A hairpin corner looms. Decisions must be made. How soon and how hard should you brake? What’s the best angle to take the corner without compromising speed? A wrong decision here could cost seconds and even spell disaster. Drivers need to understand fully what their vehicle can and can’t do in order to make such decisions.
Rains was a driver and team owner while still a college student. His device mounts on the dash of the car and uses GPS and accelerometers to measure the accelerations and decelerations of the car. It’s the only product on the market that relays information from sensors directly to a user’s cell phone.
For all that data, however, a lot of human involvement remains. That’s true in other businesses, where human insight and experience are a critical part of a successful operation.
“When you ignore the data, just like in any business, and you’re making decisions based off customer information, reports, or surveys, there has to be an element of intuition,” Rains says. “Customers are saying this, the board wants this, and investors are telling you this. But if you’re in the seat and you have to make the decisions, then you have to aggregate that information and not just go with what you think or feel is best. That’s exactly how we work on the racetrack. We use the data as one element. But as a driving coach, I use it to support what I want to communicate to the drivers.”
Don’t over-analyze, Rains says. Even though staggering amounts of data can be collected, having too much of it can cloud the decision-making process. Some of it may be only marginally relevant to the core business problem management is trying to address. It’s easy to assume that more data is always better, but it isn’t.
“If you are trying to record every single decision you make or action you take, in business or in racing, and aggregate all of that data constantly, you’re going to end up with data overload and make poor decisions,” he says. “Even in this age of machine learning, humans ultimately still make the decisions—on the racetrack and in the board room.”
