Beyond Math’s ‘Electronic Wind Tunnel’ Employs Physics-Driven AI Simulation for Formula 1 Vehicles

The challenge of accurately simulating the intricacies of the real world presents a significant hurdle, especially when striving for high levels of precision. This challenge has become a bottleneck for innovation within the vehicle and aerospace sectors. However, Beyond Math is leveraging artificial intelligence to revolutionize this domain by offering a novel simulation methodology that promises to dramatically reduce wait times for design teams.

Darren Garvey, a co-founder of the initiative, points out a distinct advantage in physical sciences compared to language processing: the existence of precise mathematical models for physical phenomena. “The breakthrough with machine learning isn’t just its pattern recognition capabilities but also its computational prowess,” he explains.

Beyond Math pioneers its efforts in the realm of computational fluid dynamics (CFD), a field that combines computing power with fluid physics to predict how objects will interact with fluids. These computations, essential for designing anything from aircraft to racing cars, are notoriously difficult and resource-intensive.

This necessity translates into extended waiting periods in sectors such as automotive and maritime design, where simulations precede real-world testing.

Designers traditionally engage in a time-consuming cycle of hypothesis, simulation, and validation, often facing discrepancies between simulated outcomes and wind tunnel tests. This iterative process can significantly slow down development cycles, according to Garvey.

The ambition of Beyond Math is to streamline the digital design process, minimizing the gap between conception and validation.

“Imagine the potential if, instead of waiting hours or days for simulation results, designers could experiment and receive feedback in near real-time. This could fundamentally change the pace and scope of design iterations,” Garvey suggests.

It is becoming clear that machine learning might offer a more efficient approach than traditional simulation tactics. Beyond Math’s inaugural offering is a “digital wind tunnel,” capable of simulating complex airflow interactions with unprecedented speed and accuracy.

This approach mirrors advancements seen in other scientific fields, where machine learning models expedite simulations by learning from extensive data sets of observed phenomena. Still, Garvey acknowledges the challenge of starting without a vast dataset: “How does a startup replicate the depth of traditional simulations on complex geometries without access to massive simulation databases?”

The solution, intriguingly, lies not in simulating but in understanding the underlying principles of phenomena like airflow and incorporating real-world data to enhance accuracy.

According to Garvey, this strategy aims not just to mimic existing simulations but to mirror reality itself, thereby providing more relevant data for design optimization.

This methodology opens the door for machine learning to not only analyze but also contribute creatively to the design process, much like advancements seen in image processing technologies.

Formula 1 racing teams, highly invested in the cutting-edge of CFD technology for competitive advantage, are among the early adopters exploring Beyond Math’s capabilities to refine their design and aerodynamics strategies.

Garvey is optimistic about demonstrating tangible benefits to customers soon, buoyed by a recent $8.5 million seed funding round led by UP.Partners, with Insight Partners and InMotion Ventures also participating.

This influx of investment is expected to support a significant expansion of Beyond Math’s team and computational resources, including the acquisition of Nvidia DGX 200 systems.

While F1 racing provides an initial proving ground, Beyond Math envisions broader applications of its technology, challenging the adaptability of their models across diverse design scenarios beyond high-performance vehicles.