AWS for Industries
How AWS supports Scuderia Ferrari HP to optimize Formula 1® power unit assembly process
In the high-stakes world of Formula 1 (F1), where victory is measured in milliseconds, innovation is vital for continued success. For decades, teams have pushed the boundaries of automotive engineering, constantly seeking that elusive competitive edge. The collaboration between Scuderia Ferrari HP and Amazon Web Services (AWS) is redefining the assembly process with data-driven manufacturing.
Their shared quest for innovation led AWS to work with Scuderia Ferrari HP to migrate their manufacturing data to the cloud. This included the vast amounts of information generated during the preparation and assembly of individual power units—the lifeblood of any F1 car. By harnessing the capabilities of machine learning (ML) through Amazon SageMaker AI, the team built a sophisticated processing pipeline atop their newly created data lake. This approach allowed the team to examine details and correlate results with historical data, increasing their analytical capacity. The team can now process at least four times the amount of data compared to previous manual methods, while reaching insights in half the time.
Power unit adjustments for regulations and speed
Featuring the fastest road-course regulated cars, F1 is the premier level of motorsports racing and is tightly regulated by Fédération Internationale de l’Automobile (FIA). Each season, drivers receive a set allocation of elements in power units, a vital part of an F1 car that encompasses the internal combustion engine, two electric motors, and turbocharger, as well as energy store, control electronics, and exhaust.
F1 is a sport of big data, but it can be the smallest data point that can cost the team a win. Penalties are incurred each time a driver exceeds the limit for an element, forcing teams to be rigorous with testing and extremely strategic about any power unit changes before and during the season. According to the 2024 Formula One Sporting Regulations, the first time an additional element is used above the limit, a ten grid place penalty is issued, significantly reducing chances of the car achieving a podium finish.
“Power units are highly complex, so our tech team gets ahead by addressing issues that could create potential problems in a race,” said Alessio Simi, Head of Scuderia Ferrari HP Power Unit Assembly. “Our AWS implementation helps us detect anomalies that we might otherwise miss, giving us the opportunity to make adjustments far in advance of the main event.”
Figure 1: A phase of the assembly process
Supercharging engineering with machine learning and generative AI
Ahead of the 2024 racing season, the team began exploring ways to improve its approach and help engineers gain better data insights, faster. With the motorsports season typically running March through December, engineers have only a small window for pre-season data analysis, and even less time to optimize or adjust between each race. A strong data foundation with AWS helps Scuderia Ferrari HP consistently monitor assembly, helping them ensure their power units can withstand the rigors of an entire Formula 1 season without requiring excessive component changes.
AWS collaborated with Scuderia Ferrari HP to create a data lake using Amazon Simple Storage Service (Amazon S3). The team then used Amazon SageMaker AI to build a processing pipeline, consolidating data from multiple disparate sources. The comprehensive testing and measurement during manufacturing helps to identify potential weaknesses or areas prone to premature wear, allowing the team to address these issues before the components go into service.
Previously, data analysis and evaluation were done manually by individual engineers with data spread across multiple systems, making it difficult to track down the cause of any problems that could lead to long-term failures. For example, deviations as small as a bolt being consistently over-tightened could increase the risk of engine instability over time, causing issues on the track. With 300 sensors on each car, the sheer volume of data became almost insurmountable for manual review. “Our engineers’ expertise is essential, but it’s not reasonable for a human to analyze terabytes of data.” said Simi.
Once the data was consolidated, the Scuderia Ferrari HP engineering team dedicated to data analysis created a centralized dashboard view in Amazon QuickSight. This empowered engineers of any specialty to monitor assembly and observe potential deviations in near real time. With this additional access and visibility, the team has reduced their time to insights by an average of 50 percent. “Thanks to this reactivity we can intervene directly on the process and components, avoiding completing incorrect fittings or wasting material unnecessarily,” adds Simi.
One element the team monitors is drift—a gradual, unintended change in a part’s critical measurement or performance parameter over time. For example, this could be a gradual decline in power or fuel efficiency or a sensor gradually losing accuracy throughout the course of the season. “With all the information in a single data lake, we can evaluate trend and drift to cross reference them with any anomalies and to identify geometric variations in the production process.”
Figure 2: Amazon QuickSight dashboard
With new capabilities of Amazon Q in QuickSight, racing engineers will be able to discover key insights and trends with Q&A responses and even spin up their own dashboards with natural language prompts for ongoing maintenance and review. “Speed is essential in our sport. Having our data and infrastructure already in great shape with AWS helps us uncover issues and make adjustments faster and more accurately.”
Conclusion
By delegating automated data collection and processing to ML, it is now possible for Scuderia Ferrari HP to collect, analyze, and take action on their insights faster. Based on this project’s initial success for power unit manufacturing, a similar strategy is being repurposed for other applications relating to performance. “While we’re limited by regulations to introduce major upgrades during the season, our data architecture is paving the way for future innovations.” concluded Simi.
Team Scuderia Ferrari HP has been hard at work preparing for the 2025 season, set to begin at the Australian Grand Prix in Melbourne. Being able to capture any test, simulation, and real-world performance data point provides invaluable insights into areas for potential improvement. “This has allowed us to identify and prioritize key development areas for our next-generation power unit designs, giving us a significant head start in the highly competitive world of Formula 1.”