How Cloud Supercomputing Helped Design a 2022 F1 Car That Promises More Exciting Racing


When Formula 1 management unveiled the first car designed to 2022 specs ahead of the British Grand Prix last week, they told fans they wanted to encourage closer and more exciting races.

Regulations, or “formulas,” are updated every few years, providing a framework for teams and engineers to push the boundaries and extract every drop of performance – without breaking the rules.

The aesthetics of the car are striking. Major changes include bigger tires, larger front and rear fenders, as well as the return of “ground effect” – a feature that first appeared in the 1970s to improve downforce before quarrying. ‘it was not banned in the 1980s.

Aerodynamic changes

From the start, it was obvious that aerodynamic changes were the key to encouraging wheel-to-wheel racing.

“When a car closely follows another car it loses a lot of downforce,” Rob Smedley, former race engineer and current head of vehicle performance in Formula 1, tells me. clear [the leading car] is a washing weight that causes a lot of turbulent air.

“This means that when the next car rolls in that air, its aerodynamics are distributed and it loses downforce … Pilots would tell us it was easy to stop behind a car, but in about a second, it was getting quite difficult as the car would start to slip and the tires would overheat.

This loss of support is staggering. If a car rolls three lengths behind the one in front, then the loss is 35%. If the car is in one length, it goes up to 47%. The aerodynamic advantage offered by “ground effect” – which more efficiently channels the air under the car – and the less complex front spoiler – reduces these losses to 4% and 18% respectively.

These changes also mean that the “dirty air” emitted by a car will be less likely to affect the car behind it.

Cloud-based supercomputing

However, the new design would not have been possible without the use of cloud-based supercomputers which performed thousands of simulations to find a solution. Traditionally this would have involved the exclusive use of a physical wind tunnel, but this is a long and expensive process that would not have allowed for the large number of incremental changes required.

Computational fluid dynamics have become increasingly common in F1 circles over the past few decades, allowing teams to perform ‘virtual tunnel’ tests alongside conventional wind tunnels – something all the more important taking into account the measures taken to control costs in sport. However, even a virtual tunnel relying on on-site equipment to power (CFD) would have struggled.

“Even if you put half a car in a CFD environment, it already requires a huge amount of computation,” adds Smedley. “If you have two million data points, the simulation takes five hours. When we switched to a two-car simulation and 95-550 million data points, it takes about four days for each design interaction.

Instead, Formula 1 turned to its technical partner Amazon Web Services (AWS). Its cloud infrastructure provides massive computing capabilities that allow it to run a large number of simulations much faster. It could even operate up to 20 instances simultaneously, further increasing the time required.

In total, Formula 1 ran 7,500 simulations that took 16.5 million base hours to solve, generating half a petabyte of data. It would take almost 500 years for a high-end laptop to do the same – although you would have hoped mainstream technology would improve at least a little during that time, too!

Better than the real thing

Formula 1 has always been a technologically advanced and data-hungry sport where even a hundredth of a second can be the difference between victory and defeat. However, much of this innovation has been done in isolation. After all, many teams say they were doing “big data” before it was a thing.

Now, the hardware and software used in the business world is increasingly adopted by the sport to find any possible advantage on the track.

“The technical complexities of F1 run deep because they’ve been around for a long time,” suggests Smedley. “But having access to an extremely powerful calculation only helps to solve [technical] problems faster.

But even the latter project involved at least some physical testing in the Sauber wind tunnel in Switzerland. So, can the virtual wind tunnel ever replace the real thing?

“As technology advances, the answer will be yes,” he says. “I don’t mean to say we’re far from that point, but we’re not.”

Source link

Leave A Reply

Your email address will not be published.