UCI Track: 'Data analysis has become increasingly important in recent years'

It is enormous. Reducing aerodynamic drag is essential to save power and attain maximum speeds. There are two ways to combat drag: in relation to other cyclists and in relation to yourself. First, cyclists can alter their track position to take advantage of the air displaced by the cyclist in front, who leaves a pocket of air in their wake. By positioning themselves in this wake, and assisted by the aerodynamic suits they wear, riders are able to shield themselves in this pocket, saving valuable power (watts). Second, riders can decrease their frontal area in order to reduce the amount of air that they have to displace themselves. This can include changing their positioning on the bike and optimizing the design of their equipment.

Of course, drag is proportional to the velocity squared, meaning the faster the cyclist pedals, the more aerodynamic drag they create. To summarize, cycling is a balance of forces: those the riders apply to the pedals to go fast and those that oppose this motion in terms of the rolling resistance and the aerodynamic drag.

Absolutely. Since the 1980s and 1990s, they have helped to improve aerodynamics. In concrete terms, when a cyclist wants to reduce their aerodynamic drag, they either change their position on the bike or the bike itself. The wind tunnel can examine both, often with very good results. But these tests also have their drawbacks: they are extremely energy-intensive and very expensive, costing several thousand euros for a day of testing. Another disadvantage is that these wind tunnels will determine an air resistance figure but they will not indicate what the precise cause is. However, computational fluid dynamics (CFD) allows you to virtually test aerodynamic flow through numerical simulations that you can host on the cloud. These virtual tests have the advantage of providing very detailed results, allowing both engineers and riders to pinpoint the origin of the drag and optimize the bike’s design and/or rider’s position accordingly.

AWS is the Official Cloud Infrastructure Provider for the UCI Track Champions League. During live races, AWS logs, processes and analyzes data streams using its Amazon Kinesis and Amazon DynamoDB Database services in real time, providing real-time insights to fans who attend the event in the velodrome or follow on TV or via a dedicated mobile app.

There are 2 categories of data: acceleration, speed and position on the track, and biometric data, like power, cadence and rider heart rate. Warner Bros. Discovery Sports has trialed the AWS SageMaker data service to perform data analytics and build machine learning models to extract valuable insights from the data.

Yes and no. In both disciplines riders or cars are fighting against aerodynamic drag. However, Formula 1 is more concerned with downforce, which affects a car’s ability to stick to the road, whereas aerodynamic drag is a cyclist’s main concern.

Physics is the fundamental core of cycling. Winning a competition is not just down to the cyclist themselves, but also their equipment, such as their bike and helmet, which can separate two cyclists of equal strength and ability. But it is true that data analysis has become increasingly important in recent years and that everyone is trying to exploit this data to the maximum. This is why we now see many coaches and technical teams dedicated solely to aerodynamics. But I still think that the power and instinct of a rider continue to play an equally important role. In a race, placement remains essential. You need to make sure you are sheltered to avoid taking too much air while being ready to react in case of an attack.

When a rider is in the lead, they displace a small pocket of low-pressure air in their wake called a slipstream. If you are right behind them in this pocket, drafting in the slipstream, you will save 40% of your power! Reducing air resistance by exploiting a slipstream therefore rests on the rider’s position. At AWS, we are developing methods to analyze all this data in real time in order to communicate to fans the different strategies for saving power by taking advantage of the slipstream. We will thus recover all the information on the riders’ acceleration, speed, weight, the pressure exerted in the corners and even the inclination of the track according to the rider’s position and transcribe all this into a physical model.

But, upstream of these calculations we have run hundreds of computational fluid dynamics simulations, placing riders in different slipstream configurations based on their speed and relative position. A machine learning model then combines the results of these simulations with the physical model to determine exactly how many watts a rider is saving by drafting. As you can see in the visual above, which is the result of using AWS’ SageMaker service, the rider at the back is saving 149.8 watts by being in the rider in front’s slipstream and reducing his drag by 11.7%.

CFD analyses are computationally expensive and time consuming. Previously, the hundreds of CFD simulations required to perform this study would not only be costly but, if you were to do it on-site, you would need to set up your own high-performance cluster. AWS offers a quick way to deploy and turn around CFD workloads at any scale without the need for your own infrastructure. You can run jobs that were once in the realm of national labs or large industry. In just an hour or two, you can deploy CFD software, upload input files, launch compute nodes, and complete jobs on a large number of cores. Once your job is done, results can be visualized and downloaded, and then all resources can be ended – so you only pay for what you use.

Due to the inherent scalability of AWS, you have the option to run multiple cases simultaneously with a dedicated cluster for each case. AWS allows you to centralize all the data and produce these analyses, which would have taken weeks on site, in less than a day.

Not only does this represent an extremely valuable saving of both time and money, this project is a great example of how we can combine artificial intelligence, computational fluid dynamics and physical models to derive valuable insights. This information helps explain to fans the importance of aerodynamics and drafting in the competitive world of track cycling.