How Amazon processed data for the new F1 car design
The International Automobile Federation recently unveiled the Formula One car for the upcoming 2022 season. The car has derived its shape from the numerical analysis of airflow through the car as well as behind it. F1 found an unlikely partner in Amazon Web Services (AWS) to provide for the computational power required during the design process. Here's how AWS helped with the number crunching.
New design helps reduce 'dirty air' in car's trailing wake
The F1 car's curvaceous new design marks a departure from the straight edges we see on the front and rear wings of the current-generation car. The primary design objective was to reduce the "dirty air" a car left in its wake that caused cars behind to suffer downforce losses and destabilize. The redesign reduces these losses in a bid to encourage closer racing.
Engineers studied impact of a car's aerodynamics on one following
The new car's design process required engineers to study and learn from the disturbances in the air created by a leading car and its impact on the downforce losses and other aspects of a car following close behind. The final design unveiled recently uses 18-inch wheels with low-profile tires, wheel wake control devices, a new front wing shape, simplified suspension, and underbody Venturi tunnels.
Aerodynamics, wake flow analyzed using computational fluid dynamics
In conversation with TechCrunch, former Ferrari and Williams engineer, now serving as F1's Director of Data Systems, Rob Smedley revealed that the new car has been in development since 2018. The aerodynamics of a vehicle are analyzed using a mathematical method called computational fluid dynamics (CFD), an inherently complex method due to the number of process variables and environmental factors that affect results.
AWS cloud computing helped F1 engineers run simulations faster
Computer simulations of CFD problems can be processed faster with more processing cores. While the most powerful commercially available desktop processors have around 64 cores, AWS was able to lend F1 engineers 2,500 cores. Since engineers improve designs based on analysis results (called a cycle in an iterative design process), running each simulation faster on AWS's cores accelerated the design process.
AWS claims it cut time per simulation by 80%
In a blog post, Amazon claimed that F1 was able to cut average simulation runtime by 80% — from 60 hours to just 12 hours per simulation. Smedley explained that despite the heavy investment into the technology behind F1 if the engineers used the computational power that individual teams are allowed, it would have taken four days per compute cycle.
Amazon says AWS also helped cut costs by 30%
Smedley shared that at one point last year, his team was processing over 12 iterations concurrently with over 550 million data points made possible by AWS's 7,500 cores. F1 CTO Pat Symonds said that running simulations in the cloud removed the barriers of time and compute capacity. Moreover, Amazon claims that F1 cut the cost of running these workloads by 30%.
Ferrari signs pact with AWS for global partnership across businesses
Separately, Scuderia Ferrari has announced a global collaboration with AWS for cloud, artificial intelligence, and machine learning services that will power the development of the Italian marque's road cars, GT race cars, and Ferrari Challenge cars besides the F1 team. This exercise of redesigning the F1 car yet again demonstrates the technological advancements backing teams and giving drivers the confidence to push harder.
