10,000 simulations for one Gen4 component: How Stellantis develops its Formula E car

Long before Stellantis' first Gen4 car had even turned a wheel, its engineers had already sent it out on track virtually. Since January 2024, the development team has been working with mathematical models of the new Formula E car and asking them ever more detailed questions: How does acceleration change with more power? How do weight, tyre grip or a different powertrain configuration affect handling and lap time? At one point, around 10,000 simulations were run within a single month just to define one component.

Stellantis simulation engineer Teena Gade explained the scale of this virtual development work to e-Formula.news during Citroen's private Gen4 testing.

Put simply, the engineers recreate the real car on a computer using mathematical equations. They can then change individual parameters and calculate how the car would behave under those new conditions. As development progresses, these models become increasingly detailed.

One particularly complex task was defining the rear motor-generator unit (MGU), which propels the car and also recovers electrical energy under braking.

"Just to come up with the specification we wanted for the MGU, there was one month where we ran about 10,000 simulations to arrive at our answer. And that's just one component," Gade says, illustrating the scale of the development work involved.

Those 10,000 simulations for a single component are only a small part of the overall development programme. Even Gade cannot put a number on how many simulations Stellantis has carried out since the project began.

"I wouldn't be able to tell you how many simulations we've run, but if by the time we get to the end of the Gen4 project we're into the billions, it wouldn't surprise me at all."

Computing power sets limits on virtual development

Behind those figures are models of vastly different levels of complexity. While some questions can be answered with comparatively simple calculations, the most advanced vehicle simulations consist of millions of equations. How detailed those models can be and how quickly they can be calculated therefore depends heavily on the computing power available.

"Simulation never ends. There are multiple different types of simulations. Obviously, the biggest limitation with simulation is computing power."

Can AI replace Formula E engineers?

Asked by e-Formula.news what role artificial intelligence already plays in her work, Gade explains: "AI has its uses, but one of the biggest issues is that we're bound by the world of physics. We're bounded by Newton's laws, and we know what those laws are. With AI, what happens in the middle is a black box."

Gade nevertheless already uses AI regularly for certain tasks, albeit more as a tool than as a replacement for conventional simulation. "Right now, certainly for what I do, it's my copilot. If I want to write analysis code quickly, it's great for that. I can tell it that I need to draw all these graphs and plot the information in a certain way. Superb. It writes the code for me."

"Can it replace me in my job? Not yet," Gade says, before adding with a wink: "Yet."

AI could become even more interesting in areas where the focus is less on physically simulating a component and more on analysing large amounts of data for previously unnoticed connections. One example is strategy simulation, in which engineers repeatedly run races on a computer with different scenarios to determine which strategy is most promising under a given set of conditions.

"If we take something like a strategy simulation, we might be able to say: let me feed you the last 10 years of every lap time for every car on the Formula E grid. Maybe it can make connections between different pieces of information that we haven't made ourselves."

More insights from our conversation with Teena Gade will follow on e-Formula.news in the coming days.