AWS Public Sector Blog
University of Michigan team uses AWS to accelerate simulations and win the American Solar Challenge
The University of Michigan Solar Car Team, renowned for their innovative designs and competitive edge, has once again pushed the boundaries of solar car technology. Their recent victory in the American Solar Challenge (ASC) can be attributed, in part, to their strategic use of Amazon Web Services (AWS) for advanced simulations and design optimization. As the team gears up for the 2025 development cycle, they’re using cloud computing to accelerate their design process and gain unprecedented insights into their car’s performance.
Accelerating aerodynamic simulations
One of the team’s most significant advancements has been in the realm of aerodynamic simulations. The implementation of detached eddy simulation (DES) models has allowed for more accurate predictions of turbulent flow behavior, crucial for optimizing the car’s design. However, these sophisticated simulations are computationally intensive and can be impractical to run frequently on local systems. This advanced simulation technique was particularly crucial in designing an innovative turn signal system that met regulations while minimizing aerodynamic performance of their solar car, Astrum.
This is where AWS has made a transformative impact. By using AWS high performance computing (HPC) capabilities, the team has achieved near-linear scaling up to eight instances. The results are impressive:
- Local computation: 37 seconds per iteration
- 1 HPC node: 24.6 seconds per iteration
- 4 HPC nodes: 6.5 seconds per iteration
- 8 HPC nodes: 3.6 seconds per iteration
For a typical run of 6,000 iterations, this translates to a reduction from 2.5 days of local computing time to just 6 hours using AWS. This dramatic speedup allows the team to iterate more quickly, explore more design variations, and ultimately create a more aerodynamic vehicle.
Enhancing strategy simulations with Graviton 4
The team’s in-house strategy simulations have also seen significant improvements thanks to AWS Graviton 4 processors. Compared to traditional AMD64 instances (r7a.48xlarge), the Graviton 4 (r8g.48xlarge) instances have:
- Reduced simulation time from 1.8 hours to 1.5 hours per 100,000 simulations
- Lowered costs from $26.28 to $16.95, a nearly 40 percent reduction
This efficiency gain has enabled the team to explore a much broader design space, running sensitivity studies to understand tradeoffs between any two variables in just an hour. The ability to quickly analyze these tradeoffs has been crucial in making informed decisions during the design process.
“The tooling improvements we have made to automate this process, allowing the use of cloud computing, have dramatically improved our ability to do this analysis,” said Aidan Goettsch, the team’s engineering director at ASC.
Turn signal design for Astrum
A prime example of how AWS has enhanced the team’s capabilities is the design of turn signals for Astrum. Due to new regulations for ASC, the team needed to add turn signals in a position that didn’t exist on the original body design. This challenge presented an opportunity to showcase the power of their new simulation capabilities.
The team explored various design options, using computational fluid dynamics (CFD) to understand the aerodynamic impact of each. Initial simulations revealed that whisker-like offset lights provided the lowest drag. However, the turbulent nature of the flow necessitated more sophisticated transient simulations to fully understand the impact.
This is where the DES model, implemented and run on AWS, proved invaluable. The team was able to capture the unsteady behavior of the airflow around the turn signals with unprecedented detail. This increased understanding allowed them to validate their initial conclusions and further refine the design.
The iterative process of design refinement was significantly accelerated by AWS cluster tools. The team could run multiple simulations in parallel, distributed across many nodes, dramatically reducing the time needed to optimize the turn signal design. The result was a highly aerodynamic turn signal that met regulations without compromising the car’s performance.

Figure 1. An animation of the turbulent and wave-like airflow caused by small protruding turn signals generated by a computer simulation that characterizes turbulence. The small disturbances seen in the airflow indicate that this is an aerodynamic design. The dark red colors show areas of high drag and the blue areas show low drag.
Building for the future
The infrastructure and workflows developed for the turn signal project have laid a strong foundation for the team’s 2025 development cycle. They have extended their web app and preprocessing steps to provide more flexibility in running simulations, allowing them to make informed tradeoffs between simulation fidelity and computational cost.
AWS support has made it practical for the team to use these more expensive models routinely. Instead of running simulations a couple of times a week, as they would be limited to with local resources, they can now use cloud parallelism to cut runtime down to a few hours. This capability means they can make better decisions in less time, a crucial advantage in the competitive world of solar car racing.
Looking ahead to 2025
The University of Michigan Solar Car Team’s use of AWS demonstrates the power of cloud computing in pushing the boundaries of engineering and design. By using AWS HPC capabilities, the team has dramatically accelerated their simulation and design processes, enabling them to create a more competitive solar car and secure victory in the ASC.
As they look towards the 2025 development cycle, the team is better equipped than ever to innovate and optimize their designs. The combination of advanced simulation techniques, cloud computing power, and the team’s engineering expertise promises exciting developments in the future of solar car technology.
The success of the University of Michigan team serves as an inspiring example of how cloud computing can accelerate innovation and drive success in competitive engineering challenges. As we look to a future of sustainable transportation, collaborations between academic institutions and cloud providers such as AWS will play a crucial role in pushing the boundaries of what’s possible.
The University of Michigan team experienced flexibility, scalability, and security while developing and testing their solar car on AWS. Other automotive customers can leverage these same benefits to accelerate new vehicle development and testing. Reach out to us via the AWS Automotive website and let us help you innovate. Let’s transform the industry together.