Rob Smedley reveals how AWS is powering Formula 1 insights, and helping drivers gear up to perform better
Rob Smedley, 46, is one of the most recognised faces on the F1 circuit. The engineer, who uses his knowledge of math, physics, chemistry, and design, to help drivers drive harder, faster, and safer, is now betting on data to revolutionise the motor sport.
The former Head of Vehicle Performance at the Williams Martini Racing Formula 1 team was earlier race engineer for Felipe Massa at Ferrari. Some of his radio transmissions to Massa achieved cult status, with many Ferrari fans using his messages as ringtones on their mobile phones.His most famous instruction to the Brazilian race driver was simple: “Felipe, baby, stay cool.”
After his exit from Williams Martini, Rob has signed up for a new role: expert technical consultant with Formula 1. As technical consultant, Rob is assisting F1 in the computational fluid dynamics (CFD) project and much more.
The F1 CFD project used over 12,000 hours of compute time to design the race car for the 2021 season. F1 is working with Amazon Web Services (AWS) to collect, crunch and deliver insights at speed to change the fan experience, and broadcast and viewing experiences.
Pat Symonds, Chief Technical Officer of Formula 1, has said that the project with AWS was “one of the most revolutionary in the history of Formula 1 aerodynamics”.
In an earlier news release, Matt Garman, Vice President of Compute Services at AWS, said it was exciting to be part of the design of the next generation of racing cars.
“The work Formula 1 is doing with CFD is at the leading edge of cloud usage and we are always amazed at the fascinating way that they are utilising our technologies to increase the performance of their sport and the experience they give fans. As CFD work with Formula 1 continues, we look forward to seeing the resulting car and are excited to see it on the track in 2021,” he said.
Formula 1 used AWS technologies to study how cars “perform in the wake of another, as opposed to running in clean air” and understand the “incredible aerodynamic complexities associated with multi-car simulations”.
In an exclusive interview with YourStory, Rob Smedley speaks on how what is changing in the F1 world, the need to have data in real time, and what the race car of 2021 will look like.
Edited excerpts of the interview:
YourStory: You belive in adopting science to change the way entertainment is viewed. How will the partnership with AWS change things at Formula 1?
Rob Smedley: There are 500 million people tuning into a grand prix every time. It is arguably the world’s greatest sport. In that context, we want to make F1 more entertaining, intriguing, and engaging. We have a strategic vision. In 2021 we are going to make seismic shifts and changes in the technical regulations of the sport. We engaged our fan base and asked them what they want. Our fans want to see wheel-to-wheel racing, the way drivers compete with each other. Those are iconic moments for fans.
But what will actually transform fan experience? The change will begin with aerodynamics.
Why bother about this? Let’s talk about force; teams work very hard to create it. It is what pushes the car down and gives grip; that’s how they manage the corners at high speeds. Then there is drag, which slows the car down when you are going in a straight line.In F1, all teams want to maximise downforce and reduce drag.
We know that the downforce goes up when speed increases. It is exactly the opposite in the aircraft industry, where when the speed goes up the wings lift the craft. In F1, the speed of the car keeps it on the ground. There is a lot of down force on an F1 car, for example, a car going at 100 miles per hour creates a down force of 1,000 kilos, which is about the kerb weight of the car.
How do aerodynamics enter the picture? Two teams work on this. The first one works on the CFD and the second focuses on the wind tunnel.
The wind tunnel works on a scale model of the car, about half of the size of the real car, and is an accurate representation of the car’s aerodynamics. CFD is getting more sophisticated; it has collected a lot of data on cars, flow physics, and performance. For example, the vortex structures that flow underneath the floor of the car create the grip. The aerodynamic wake, the air that flows over the car, is the one that’s bad, aerodynamics-wise. This data reveals how we need to change the weight of the car.
Building a race car isn’t easy; it takes a lot of effort and involvement. You have to do a wind tunnel test, then a CFD, and then test the car on the track. CFD is a supercomputer problem or a high-performance computing problem. We built a 20-core supercomputer to understand data and it was not scalable as there was so much data. We now use 200 cores and compute a lot of data.
YS: Tell us why the CFD project is so significant?
RS: A car that is one second behind loses 30 percent of its downforce; if it is half a second behind then it loses 40 percent of the downforce. The plan now is to use a two-car simulation and understand the aerodynamics wake effect of the car that’s trailing. Once you see that in simulation, you see how much downforce is lost. So you need to create a new car that can manage the wake effect.
In 2021, we are redesigning the car in simulation to solve this high-performance computing problem. This is what makes our partnership with AWS significant.
A single car’s simulation and testing would have taken 14 days. Even with a team technology of 200 cores, it takes 14 days; this is not agile to build a better Formula 1 car.
Now, we can do a two-car simulation in 11.5 hours; this gives us the agility to move forward. We are using 7300 cores to solve this massive problem statement. This has helped us design a 2021 challenger; a car that will show engineers how better designed vehicles can lose only 5 percent of the downforce when one second behind, instead of 30 percent as the case is today. For half a second, it will be a 7 percent downforce loss.
Fans want cars to perform at their very best, and we are working on that. With insights gained from these simulations, Formula 1 has been able to design a car with only 15 percent downforce loss at the same, one-car length distance.
The resulting car will feature a brand new bodywork design with a new front wing shape, simplified suspension, new rear end layout, underfloor tunnels, wheel wake control devices, and will run on 18-inch wheels with low-profile tyres for the first time.
YS: When did data start making a difference in F1 and when did you decide to tap it?
RS: Changes began to happen in 2006, thanks to regulations asking us to bring the testing time down. Every team got down to testing things virtually. From 2009, those who used data efficiently began to win titles.
You may remember that in 2009 Ferrari did not win the drivers and constructors championship. I have been an engineer for 25 years now and continue to give back to the sport. I worked with champion teams, and later took this opportunity because solving problems using data and computing is close to me. Teams generate so much data that can be used, from performance and broadcast point of views.
At F1, we have a very startup feel to building a high-performance car for 2021. There is so much to work on: engineering, content, and fan engagement.
YS: The world is talking about electric and hybrid engines. Do you see that impacting F1? How do AI and machine learning come into the picture?
RS: There is no single solution out there. VW has announced that it will go all electric. There are people who talk about hybrid. But engineers are iterating on what works. F1 has used V-8s and V-6 hybrids. There are also 1.6-litre engines with 1,000 horsepower that are thermally efficient. F1 will continue to change with regulations. On the software and infrastructure side, the cloud has already helped us in CFD.
Our partnership with AWS will help us scale our technology architecture, which is very important to crunch data.
Several teams are already using AI and ML capabilities. The car’s performance has to be studied, and machine learning helps us gain insights, which earlier used thousands of people, from volumes of data.
(Edited by Teja Lele Desai)