Yahoo Autos An inside look at F1’s Sim City
Last summer, a tweet from NASCAR driver Landon Cassill shed some light on how simulation lays the groundwork for what happens in the real world. It was something we covered here at RACER at the time, but with simulation being such a major part of motorsports today — particularly in Formula 1, where a car may change entirely from race to race — we wanted to delve even deeper.
To do that, I headed to Bristol in the UK, where Dynisma Limited is based. Dynisma is the class leader in simulation technology, with Scuderia Ferrari among its clients.
It was founded in 2017 by former McLaren and Ferrari F1 simulation lead Ash Warne, and in an era where real-world testing is confined to a few days a year, the company provides a vital service.
“It’s the main tool now to be able to allow them to develop the cars, and automotive companies are doing the same,” Warne tells RACER. “They want to make fewer prototypes. They want to be able to go straight to the production line without having to spend time and money building mule cars, building prototypes.”
Dynisma aims to blur the lines between virtual and reality by developing the most accurate simulation technology around.
“Simulators have been around for a while, but what we’ve got is a step change in the technology that allows drivers to feel the movement of the thing that we’re simulating and how that’s moving more accurately,” Warne says.
That’s been done by reducing the latency — the delay between what a driver inputs and what they experience. It’s been simulation’s key weakness in the past.
“Our first major contribution is about latency, and we’ve got that latency down to less than five milliseconds, which is basically imperceptible,” Warne says. “Typically, with other simulators, you can’t drive them the same way as you can drive our simulator, because you have to put in more understanding to deal with the latency. So that’s kind of critical.
“If you’re suddenly increasing the human reaction time by 25, 50 percent then — especially in the context of an elite athlete — this vehicle is going to completely wreck the simulation.
“In a simulator where the latency is too high, what basically happens is the driver gets frustrated because they’re saying, ‘Well, I can’t get around a lap. I would be able to drive this on the track, but I can’t drive it here. So I’m going to take out two turns of front wing to make the car more stable. I’m going to put a softer rear bar on to give me more grip, to make the car more stable.’ So now all of a sudden, you’ve got the wrong baseline. You’re not carrying out the experiments you wanted to.”
Dynisma’s Warne explains to Wilde how the system lets drivers feel as well as see how their sim efforts are going.
Couple that vastly improved latency with motion technology unlike any other, and Dynisma’s simulators can give drivers the sort of seat time that almost mirrors the real world.
“Fundamentally, our Motion Generator technology allows the driver to feel the car in a way that other simulators don’t,” Warne says. “That means the driver is able to drive the simulator in the same way as they are the real car.
“If you imagine the recent world champions and everybody on the current F1 grid and F2, F3, all of those drivers have typically been driving cars or karts of some description since they were really, really young, and they learned to be at one with those machines. They’re connected to them directly, and they feel every little movement, every vibration that’s going on in that car. And even if they can’t describe how they’re responding to it, they have learned to respond to all of those stimuli.
“They hear it, but they also feel it. So the driver really feels the feedback from the motion system in two ways. One is through touch; a bit of pressure or vibrations they’re feeling through their backside, and then the other is through the vestibular system – these are the organs in your inner ear that help you balance, but they also help you detect movement.”
Dynisma’s technology means that simulation is at a near-perfect level, to the point where it serves as the intermediary between two physical elements: the wind tunnel and the track itself.
Results from developments prototyped with scale models are replicated in the simulator to validate them before the car hits the track. Once it does, the real world performance is back-to-backed with the expectations from the sim to correlate the data, ensuring that all lines up as expected.
“Basically, the wind tunnel is only ever an approximation,” Warne says. “So the simulator is a tool that brings together all the data that the team has about the car and allows the driver to drive it and to do experiments on what’s coming up next, or look forward to next year.
“But it can only ever be as correct as the data that’s put into it, and that’s why teams will have tens of people in the aerodynamics department, tens of people in vehicle dynamics departments and vehicle modeling, and a big part of their role is to understand the physics of the car, understand how it behaves, and in part, to be able to put that in the simulator.”
All of that data creates what is known as the “vehicle model.” And since we’re talking about sims, you might assume they mean a literal 3D model much like you’d see in a video game. But no, the reality is the “model” is a list of numbers that are put into the system.
“This is the bit of software that describes exactly how the car works,” Warne explains. “How stiff the springs are, how much downforce you get off the rear wings, how much drag the car creates. So not the literal 3D model, but the sort of under the skin, mathematical model. A description of everything that matters to car performance about the car, how downforce changes with the ride height of the car, when it goes over bumps or when it gets compressed due to downforce, how the downforce changes and how the drag changes, how the balance of the car changes.
“F1 teams are right at the forefront of how to do this, and they’ve had their vehicle models, their mathematical description of how the car works, for several years. So every year it’s an increment — they start with the model of this year’s car and they do a lot of correlation work where they compare the results that they get out of the simulator with the data that they get on the track.”
Naturally, building upon last year is a logical step. We see it every season when most teams unveil an evolution of the previous year’s car that can be seen, felt, and driven. But for seasons like 2022 — or the upcoming 2026 campaign — where there are wholesale changes to the regulations, you can’t evolve. Or can you?
“It’s still the same methodology, but you’re taking bigger leaps,” Warne says. “They’ll start with the aero data — they would have their wind tunnel model of the previous year, and they would have their wind tunnel model of next year’s car. Early on it wouldn’t be very representative, and the downforce level will be really low.
“But fundamentally, they’re still doing wind tunnel testing. They know from historic data what their likely rate of development in the wind tunnel is going to be. So they will actually bolt that benefit onto a less-mature car so that they can predict what they’re going to be like at the start.
“There will always be a difference between what you measure in the wind tunnel and what you get in reality. And there’s lots of reasons for this. One of them is that the wind tunnel model is a scale model. It’s smaller, and it’s run at constant wind speed. Also, just by virtue of the fact that you can’t have curved flow as you can in a real car — you’re going around a circle, going around a high-speed corner on a real car, the angular instance of the the oncoming air at the front and the back is different, whereas you can’t actually do that [in the wind tunnel]”
By the time the drivers are plugged in, the simulations are typically well underway.
Before a driver climbs into the simulator — which is usually built using a genuine car monocoque for absolute authenticity — the car model theoretically laps a track without a real driver, not just as another data collection exercise, but as a cost-saving measure as well.
“It’s actually relatively expensive to run a simulator session, because you need a driver, you need an engineer to run it, you need the simulator, and so you can only run that for 10 hours a day,” Warne says.
“They have offline simulations, which may have a virtual driver, or they may have an optimization algorithm that’s working out what the right driver inputs are to drive the car around the track.
“So typically, before a test has got as far as the race driver in the simulator, the week before that, it was being tested in the simulator by a test driver. But in the time before that, it was probably being tested by engineers in the office using offline simulation tools.”
The words simulation and development go hand-in-hand, but the tool doesn’t fall to the wayside once the car arrives. Of course, it is used to continually refine the car over a course of a season, but it is also used on a race weekend and during its build-up to prepare for a grand prix.
“Typically on race weekend, and more often than not starting the weekend before, the race drivers will have been in the simulator to prepare for the next race, or, if it’s a flyaway, maybe they would have been in it preparing for the next two or three races,” says Warne, who was McLaren’s simulator test engineer and simulator team leader in the early 2010s before heading to Ferrari to serve as Ferrari’s senior vehicle dynamics engineer.
“They will have sat and worked in the simulator with their race engineer, much as they would work with them at the track. They’d spend a day in the simulator going through a run plan, trying out different setup options that are well suited for that specific track, or they might be evaluating an upgrade package.
“Come the Friday, the first thing that will happen is that you’re waiting for data to start coming back from the track, and that’s as soon as the first run has ended. That data will immediately arrive back in the factory, and engineers will immediately be overlaying the data that’s come from that first run with the data that they generated in the simulator session before the driver flew out.
“For most of P1, what the sim team back at the factory — including the third driver or the simulator driver — is doing, is just checking correlation against what we found at the track. So the first session, the practice session, is all about ensuring that you have correlation.
“Then you’ll get towards the end of the session, they’ll do a debrief at the track after the first session, and the driver and the engineer will be saying, Okay, well, these are the problems that we’re faced with. How are we going to fix those? Then those questions will go back to the factory and then in the simulator, the simulator driver and the engineers will be carrying out experiments, as requested by the various engineers at the track to work out how they can improve the setup — maybe for P2 but there’s probably not enough time, so they’re more likely looking forward to the Saturday morning at that point.”
Warne says the reliability of modern F1 cars means that that sort of simulation is more important than ever.
“Improved reliability has made every small improvement you can make so much more important, because there was much more of a lottery effect in the past that somebody might just drop out,” he explains. “Whereas now, especially when the rules are a couple of years old and everybody started to converge on similar sorts of pace, every tenth of a second, even every millisecond is so critical.
“It’s those very small differences that turn out to deliver big, big differences at the track, and yet they’re still very closed with regard to what they will reveal about how they do things.”
Behind the wheel…
An element of simulation that we’ve not yet covered is that simulation provides rookie drivers the opportunity to get some seat time, forgoing the huge costs of renting a track and running a car. The writer of this article is a rookie driver… You can probably guess where this is going.
RACER’s rookie straps in…
As well as learning about how simulation is used in Formula 1 today, I was also put in the simulator for two sessions. Firstly, I drove at Circuit de Barcelona-Catalunya, F1’s go-to test track. But simulators are just glorified video games, right? How hard can it be? Well, as we’ve established, that couldn’t be further from the truth. And I can now vouch for it, too.
I was thrust right into the deep end, driving a representative 2019 F1 car, and after gradually dialing up the power and dialing down the grip, I was let loose for a handful of laps with the real (virtual) thing. Far from being a game, it was incredibly demanding, both physically and mentally, with the tiniest of mistakes taking several laps to recover from.
There’s no rewind button either, and because gratuitous use of the curbs or spins (only a couple, thank you) deliver physical feedback, they require a full reset of the sim to protect the machine — and the squishy organic thing sat in the driving seat — from damage.
Barcelona looks relatively simple on paper, but there’s a reason it’s used for testing. The mix of high- and low-speed corners and long and short straights provides a good bank of baseline data. And for someone who writes about cars because he was never rich enough to race them, it provides a well-rounded feel of what an F1 driver goes through.
Interestingly, though, while I absolutely won’t be giving Max Verstappen a run for his money any time soon, I found the sim much more accessible than a console game. It sounds ridiculous, but the immediate feel, the visual and physical feedback means that you aren’t just using your hands and eyes — your entire body reacts. The simulator gives you much more, but you have much more of yourself to put to use.
I got out of the sim physically spent, aching all over, wishing I’d slept more the night before, not gone to McDonald’s on the way there, and not given up playing football while I was at university so I had better fitness levels. Curling up under a blanket might’ve been a good idea at this point, but the Dynisma team had another idea: Spa-Francorchamps.
But while it was a daunting prospect at first, after a couple of laps I was way more comfortable — and that was despite running with the full complement of power and lowest amount of grip from the start this time around. Eau Rouge was flat out by lap two; Pouhon the same a few laps later, and by the end of my session I was around 1.6s off last place on the 2019 Belgian Grand Prix grid. Not bad for a journo whose only previous race car experience had been from the passenger seat.
I was proud of my effort, but I think I’ll stick to writing for the time being.
