Simulation in Formula One: The Future of Racing Technology w/ Gerson Garsed-Brand

Welcome to the Reimagine Mobility podcast series. I'm here with Gerson from SabeTechnologies and Gerson explain to the listeners what in the world is Sabe? What are you guys are doing? And then let's jump in to a very exciting field that I'm excited to talk a year today.

because we have not had, in my opinion, to many, if anybody really in your space, the racing space and us Brazilians certainly are way more into Formula One probably than than all Americans are me coming from Switzerland, even though Europe certainly Formula One is big. I'm so let's talk about all these things as together we reimagine mobility or, and have a hopefully a fun and very educational at the same time discussion. Sure. Thank you for having me, Stephan.

yeah. So I'm, I'm I'm Brazilian. I am currently based in the UK. I've been here for 15 years now, so that kind of explains my, mixed up accent.

yeah. My my background is in aerodynamics and engineering. and my passion has always been motorsport in racing car, as you said. you know, Formula one was a big thing in Brazil when I grew up in the 80s.

90s, with Senna, and with and then, yeah, I did for me my, my career that went into engineering and, very, very, very quickly I moved that into, motor racing. yeah. During university and early stage of my career, I was exposed to, various aspects of simulation, e-mobility so they could make stress but I was, fluid dynamics and computational fluid dynamics. That really made sense to me.

So that's the career. Career I followed. And then I moved to the UK, chasing my childhood dream of working formula one, which I did, for a first stage good ten years. where I, I kind of moved across I did several sideways moves in terms of started with, CFD methodology and then moved into CFD development, actually developing the car using CFD, then done some thermo analysis of brakes, and then, we did a testing and then later aero performance, track side air performance and correlation.

So several side side side moves. But I always been very curious and I think that was, you know, very interesting for me. And then five years ago, I decided to start ASAP. I think I had a, but still have a young family, but what, even younger than.

And I felt I couldn't quite, find peace with the the work environment I was in and being around my children as well. So I decided to leave and create, made my own business. Since then, I always say I probably, you know, the workload has been the same way, even higher than before, but at least I have had much more flexibility to to decide when to work and to be around my family a bit more. So that's been great.

yeah. Within same, we, we we are a fluid dynamics, experts who we are a small engineering consultancy. We have seven of us only. We are based in Silverstone.

So I'm sitting just next to the race track here. And, so we kind of cover all the fluid dynamics aspects. So we have done marine, we have done America's Cup before, but aerodynamic is where, you know, is our bread and butter and we De Niro we, we motorsport is a big tech area that then high performance, high end automotive as well. So we clearly we are looking at we become more relevant when our performance is important.

So we are looking at faster cars for that. And and also we have a food scene in the aerospace, particularly vital, sectors and, you know, kind of in depth and, and within say we kind of covered the whole. So we are not just a simulation business. We do, surfacing the entire aero develop, essentially surfacing in simulation.

we prepare experimental tests, we execute the tests, we, do performance analysis, correlation and so on. Yeah. So that's a bit of a summary. And then of a few more of what I noticed over the last year I've been working, I've been very fortunate to be working with the Andretti Formula One project again, has and I'm so he's, American team trying to get into Formula One.

I've been heading the CFD, department here in the UK as a consultant and helping them to build up the team and, and so we've been for a year now. And, yes, the plan is to carry on in to helping to set up technology in the team, and then, we'll move out of the way when when they're fully resourced. And, and I also work with the ACO, with they are the organizer of the 24 hour of the month. So I work on the motivation of the hypercars are still behind me here.

and also a performance, analysis. So, in the main events. So the main thing coming up is the 24 hour, in the beginning of June. So, yeah, getting busy for that.

So, yeah, I think that's a, a good summary of who I am. What do I do? And what say does I. Perfect summary.

So let's maybe start out from the top generally speaking simulation. Right. Again you mentioned you guys are you guys are in business AVL is in business. as well when it comes to simulation tools.

But tell me a little bit. Whereas where has formula one maybe we stick with Formula One for a moment, the big ones. And but also when has Formula One's really started to adopt and use simulation? Because again, you mentioned, you know, Ayrton Senna, right?

I mean, I remember watching him, and several other guys back in the days when I remember my dad watching Formula One races. if I look back, I would never think that they used simulation, right. As was sort of trial and error, in my opinion at least. But again, I'm not I'm not an expert.

There. But when it relates to simulation, again, as we reimagine mobility for simulation, certainly anywhere in the mobility space for electrification to autonomous to big data, all of that stuff, digital twins and simulation tools play a bigger and bigger role. So tell us a little bit when has formula one auto racing in general? But again, certainly the big guys formula one started using that and then as a next step, then I want to go to and where is it going and how much more progress can there be made or are we already at using the cutting edge of what simulation technology today allows in formula one?

And really in passenger vehicles? We're the ones that then benefit from what formula one is inventing or doing. And it's not the passenger vehicle guys showing the formula one guys and how to use simulation. So a lot of questions in here.

But never want to share that at So it would be I'm interested in I'm sure our listeners as well. Yeah. And I should say I'm not I'm not I don't still qualify to comment on other areas of simulation than mine. But I do hear the rumors.

And also, I think when things started in the 80s, 90s, I was, I was a child, so I don't quite know, you know, the, the background, but for sure, simulation has been around for a very long time, I think, 80s, 90s, you know, some, you know, fluid dynamics and stress. It was all there, and growing. And then from fomula one was the one of the first, to introduce composites. So carbon fiber, I think.

And then that pushed the, you know, nonlinear stress simulation and so on, on, on the air versus competition fluid dynamics. We'd be very much dependent on computational power. So, although that was, a already a push in the late 90s to, to start using us very, very limited in this area. When I started my career in from aligning the late.

So it was 2008. I was nine when I started. I still I still think the computational power available then was not good enough for what we needed. So, CFD, which stands for competition through dynamics, was very much a, a complementary to, kind of supporting tool for wind.

the development release, but that's, and that's been seen as kind of the traditional approach of forward development has been you do things, you CFD first and then what's good you move to the tunnel and then it continues to move says is good. We come and say is good and that goes to the racecar. So I totally like a linear, serial process. but over the last decade, I would say there has been quite a transformation in that sense where, we now have much, much better, CFD that simulations, but partially because of the availability of resources, but partly because of new models, better models and so on.

And, CFD safety analytics, in my view, is, is a, side by side with the wind tunnel in the aero development. So we do wind tunnel CFD has limitations for sure. So does the wind tunnel. Both are tools and they are now used in combination, to make decisions about performance.

So that's really interesting. I think from Milan is a bit of a weird environment in that we, we are regulated, in terms of, we are restricted in terms of our development. So when you think about wind tunnel, there is a limitation with how many runs we can do it and how much time we can spend and time when we think about CFD, there is a there is a limitation in terms of how many geometries we can test and computational power. So there's a, there's a metric for how much solar time you are.

You have to heat. So, because for that reason there is a bit of from a one end up converging to a slightly different balance sometimes in terms of the model accuracy you intend to end up working with, maybe at lower accuracy. Not not in the lower accuracy, but at lower accuracy than possible. The model than possible.

to be able to get more answers more quickly, evaluate the more more options, more quotations. So it's a bit of a funny one in that sense, but but yeah, it's and and I think you ask me about, how someone brings, I think, things to, to the wider world. If I think one thing from one is very good at from the one is very resourceful and the pace is super fast. So, you know, there is from which and is the one really pushing for the latest processor technology.

now we are, embarking into, which is already available is becoming more and more, common, GPU being used for CFD compilation as well. so all these technologies are very much pushed hard. And when we, we work with, you know, suppliers, cloud computing suppliers or HPC suppliers, we you know, people are we are they are used to people doing 1 or 2 simulations a day and in front of one team. And we are doing each 151 day.

So when you think about that, the, you know, the bottom max in terms of, storage bandwidth, memory, you know, at but there's a whole everything is very different when, when you are working at that scale. So the scale in the space is just unbelievable. And, and then I think so one final thing is are you talking about where we are going? I think that's, I think I could speak here for a few hours.

on that subject, I think we are in a very interesting time. I find I find interesting and exciting. I think some some people find maybe potential to be scary. which is the.

What? Where are you going with AI? and then I think, so I'm in a very fortunate position to be, talking to people and finding out what technologies are there and behind me a fair amount of resources to make those things happen. so getting what I think it is, a fact is that over the next 5 to 10 years, there will be a huge transformation on how we do things in terms of er in terms of development.

I think I'll talk from for my field in areo dynamics, but I do believe that will translate to other areas of simulation, where the way we do things will change very dramatically. And when things change there's always winners and losers. And so there will be surprises coming. But, yeah, we what we are experience now, I think is there are some I mean, AI is a very broad term.

I think there, you know, a linear regression is a form of machine learning. But, and it's been done for many, many years. But I think some of the technologies coming through now will vastly transform the way we do things in terms of, so taking as an example, CFD, where we are talking about very significant costs to run a simulation and maybe, you know, from one environment, 1 hour to 2 hours to get a full car simulation, we're going to be talking about doing that in less than a second.

And therefore, instead of doing each shared analysis, doing three a day, there now can be minutes a day. And then the statistics of that are unbelievable. So if you think about millions a day and you should think, well, even if 99.9% are not very good, there's 2.

1%, which is a huge number of good options coming out. you know, so this is where we are going. And for that reason, I think, the entire concept of what tonight aerodynamicist is or what that engineer developing, whatever they are developing will change the way they do. Things will have to change.

People. Interesting. So. Well, let's take it.

Maybe let's take it a a step up and just look at racing as a whole. Right? I mean, to me again, a I don't watch racing. I love the technology, but I'm not necessarily a racing guy.

I've gone to maybe 3 or 4 races, in my life. My son loves it, but to me it seems you know where F1 is, the big ones, right? And then maybe in the US, it's certainly NASCAR, right? When we look at technology, when we look at things that you guys are doing, simulation tools, fluid dynamics, whatever it is, let's keep it as generic as you want or as specific as you want.

Now, in the automotive space, you have different players that use simulation to a different degree, maybe because it's they're used to it or they need to because faster development, because of cost cutting, whatever it might be. Explain a little bit how is racing today and where do you see go is really the formula one teams and maybe the NASCAR guys that use it and all the other, because there's so many different racing series that that I'm at least aware of, and it's probably even more than I don't even know.

There may be more more basic, still more doing it, the quote unquote traditional way. But paint a little picture on how it looks like as it relates to technology and specifically simulation technology adoption in the different types of areas of of racing. Yeah. I mean, it raises an interesting one.

So it's, you know, obviously the more resources you have, the more money you spend, the faster you go. So so if left unchecked, every category will become, you know, there'll be a bit of a Cold war going on and the spending war and the richest we win. So. So I think in general, categories realized that they, they have to put some restrictions to, to limit that.

I think in more basic, basic categories, they try to really constrain that. So you, you only get a car or you can do is to change the suspension set up and tire pressure even or come, you know, very basic changes in win angle. But you you really don't have much. much more to do.

You cannot develop new car and, and that's it very much intentional to to keep the costs down and, and keep it. I try to I mean, I worked with, Lima. so the cars and the equivalents in in U.S would be the IMSa.

So the LMD cars, sorry, nine IMSa is the, I forget the name of the championship where whether, figured out. But, anyway, the cars racing there which is the equivalent and a, that those cars are, you know, they are constrained by a set of regulations, but fundamentally they are motivated, so they are obligated to win tunnel. And then once that is done, you cannot change the car. And until you come from a real motivation.

So the again, the idea there is to freeze the design. And then also, in that in those categories you have a balance of performance. So I will differences change the difference in performance. Then it balanced out by weight with weight and power limitations.

So that again is trying to keep the cost down. And competitiveness. I cannot comment so much about, NASCAR and IndyCar because I my, my understanding is I think it would be wrong for me to comment and my understanding is limited in formula one, though, although we do have these limitations in terms of budget cost and we pay spending, the scope of work is too unbelievably interesting. So, you know, even with the restrictions, we have teams running wind tunnels, five days a week, two shifts.

That's 12, 16 hours if anyway, they're doing and, and, you know, data gathering is almost unlimited. So we have say, say in aerodynamics, we, we have, pressure topics in the car. So we are talking about now, hundreds to potentially thousands of measurements of pressure on the race car and therefore the data sitting back and, you know, and then various various different tools and methodologies to correlate that to CFD, you know, and understand if the car is doing the same thing.

yeah. So technology wise, if you, and I think that formula one needs to pushing the limits there and just and it is simply a case of resources. So the the amount of resources available and the, you know, the money behind it is unbelievable. So it is great.

Great. You push the boundaries. yeah. I don't know if I quite answer your question on that one.

you know, I think I think it's good. I mean, again, I guess, you know, where I'm getting that is like from our business in in racing. Right? I got to keep it general here.

But from our business in racing at AVL, I see a very, very competitive space. Right. Very, very competitive. when I compare that with passenger vehicle OEMs and their competitiveness, it just doesn't.

It's really not on the same scale. It's just not on the same level. Right. So then the question is how much or or some formula I see with formula one or again to racing that, you know, the monde or the formula one that, that you're very familiar with.

What are the different levels? Different companies are using simulation because in passenger vehicles or the heavy duty truck space that do, we do most of our work, right. That we can definitely point out which ones are the ones that are adopting and embracing simulation technology way, way more than the auto ones, though, right? There's the leaders, there's a large field in the middle.

And then you have, let's call in the lagers maybe right. Is formula one similar though, or does anybody in formula one recognize if you don't do it full blast out, you're not even going to have a chance of winning because you you need to take advantage of every little thing or the the stack up of the differences make it impossible for the driver to make up that difference in technology and capability. So share a little bit there. And then also again with always looking into the future, what does the future look like.

Right. So if for example, everybody right now is on the same level, is simulation never going to be a differentiator? Is the next best tool the differentiator and the ones that are really eating are just always on the cutting edge of the next better tool. Yeah.

in In formula one, everyone is pushing every single limit. So yeah, everyone is into simulation one way or another. I think, the quality of their models, is probably will vary. And, you know, someone like, like myself wouldn't have the visibility of what other teams has in terms of quality.

But everyone is pushing I think is everyone agrees that simulation is is a super powerful tool, both in terms of, saving costs and, you know, reducing, reducing, the life cycle and, and, and also look, so the on this aspect of simulation is obviously understanding what's happening. And you have much more visibility, much more information about the problem you are trying to solve. I think what, what, from a one is really, different is, because of the amount of cost, amount of resources and testing we do, we have a lot of data to correlate.

You. So I think the models are very much tweaked and, over time have been adjusted and, and accuracy improved. and then again, I'll come to I now as in, I think another area that is being exploited very heavily is how do I integrate experimental data to, computational data. I think anyone in the I think I think all of us in the simulation environment have been, we are used to the people questioning the quality of our models.

I always felt. But as he. That's definitely the case in CFD, the thing certainly has been on the back side over for a long time. but I think very few people over the years crash the experimental data as well.

And how that does that represents the track. Why. So now now I think where we're coming in my field, we are coming to a point where we have a good understanding of the limitations of both sides. But we are saying, hold on a second.

We have all this information. My simulation might explain very well these relationships. My experimental model gives me this quite good, accurate information of those points, but only those. How can I couple those two now and, and, you know, make a better prediction of what's going to happen in reality.

So so, to give you an example, here it is again. I'll try to stay on my side, but, my, my area, but, tire shape is a big thing for us. So we we have a car, we're on the track, huge deformations of the tire and from on from there, then on this use, that's crucial to understand how the tire, the forms and model, the shape and the safety model and the wind tunnel. And that's a big source of correlation across tools.

and yeah, you can go in a, in a, in a test rig and, you know, apply some preload to some natural lobe and measure that. But that's not the race car. Right. The, the, the racing environment.

So it and then you can go and create a simulation model that tire and you know that exploit different different loading that you cannot do experimentally. But then you have all the limitations in terms of accuracy. So now where we are coming to is, you know, I can combine those two very well and have a much better prediction of weapons on track. so I think that that is a very powerful aspect of what's happening now in terms of enhancing simulation.

Well, this I think in terms of how it goes to automotive. I work with automotive as well. I think, everyone, you know, at least from the high end engineering, you know, fast cars out there and luxury cars, you know, the usual million dollar cars that kind of thing. You know, people making those cars, they obviously they they do have their digital twins.

That's the intention. I think everyone understand the value of simulation. but obviously the resources are much more limited. So, you know, like how they run the simulations, the quality of the model, the correlation levels are reduced.

But I think throws off to the idea is the same. Interesting. So maybe one more question generic to to simulation then. Yeah.

Excuse me. You've been in this space for a while. You said you've been in the simulation space and just formula or not formula one but just racing in general I would say again and stay with the ones you're certainly familiar with. What areas do you see specifically with AI pushing into not only development but during race, probably data analysis and Post data analysis and all of this stuff.

What areas in racing, in your opinion, is simulation technology or AI, or the analysis of data that we may have or may not even know we have? Where are we not pushing simulation technologies yet in in in the racing space where we are not pushing? That's an interesting question. really I mean, I guess I give you an idea riding in passenger vehicles, we're pushing.

I am, from my perspective, slowly but surely into assessing with the use of AI and the data we're collecting from driver monitoring cameras, for example, the behavior of the reaction behavior and reaction of people. That is not the purpose of a driver monitoring camera necessarily, but it gives us information as relates to the way the car breaks, the way the car accelerates to maybe how annoying or not annoying, asserted a certain audible or visual warning or sign is right. that is really, I would say, somewhat of, of a new area.

Right. It's sort of an area that's being trailblazer now. And I think we'll come out on, there's other areas where we use data from a driver for more from a, from a medical perspective. Right.

And how well is his or her heart rate doing, in certain situations? should some will be allowed to, drive level three autonomous on the road when he or she is upset, out of breath, drunken impaired right, whatever it is. Those are some areas we're pushing into. That simulation is only, I think, at, at, not at the forefront, but at, sort of at the beginning of what we can do.

And I'm interested that there's got to be areas too. I would be surprised if those not in, in racing where, where you see the opportunity with the data, we either have or could get the we need a create models to simulate some of these things for the better obviously then improve the technology to speed of the vehicle, maybe to safety. or also areas where you could see we could collect data that then really would allow us to create new products, maybe for your company, for other companies or new areas to really, again, not only reimagine mobility, but reimagine maybe how a driver, a race car driver drives the race car.

So that's what I'm about here. yeah, I think I, yeah. Okay. I would start with racing, then I'd like to move into aerospace, which I think would be more relevant in these ones.

I think, one area that I cannot say is not being pushed is, teams have been trying to push, but I think is an area where there is still a steep learning curve going on. It is, coupling, experimental data with simulation through, machine learning models. So, I'd say one thing is, so, for example, how we characterize the aerodynamic performance of a car. And that applies for both cars as well.

But it's just that doing that kind of analysis is a little bit less relevant in, in a race car. So we go we don't know that. We know where there is, you know, a quasi static or, slow motion. And we measured in different conditions.

And then we create these maps, air maps that, you know, explain the, you know, kind of give the, the performance over several conditions. So on, and likewise, you CFD what we tend to do is we do this quasi, quasi static simulation where we are running steady state in various points. So we don't have the transient behavior. We can we could do it.

But these very expensive units and one recent case from Milan, the the recent regulation that occurs is the effect of porpoising. So we have this coupling between the diffuser and so the car, the teams want to run the car as low as possible. But then the car priest so much downforce he sucks the car down into stalls and then pulls up again. So you have this dynamic.

I have dynamic coupling. and, you know, it's essentially a transient effect that has to be models and, and, you know, the tools are not quite there to do that easily. And, and again, I need to be careful here because there are all the teams maybe might watch these and say, hey, we've been doing this for four years and I'm aware that they, I would say is in the early stages of, of that development. And I think there is a there are opportunities here.

If were using, existing or experimental data or race car data from the track and, and then combine, to simulation data and make predictions about what would be the transient behavior. Now moving to an aerospace example, which I can talk about, which I just my company, developed. So we, we recently had been working on a, a new concept of an electric propeller, for a eVTOLs is targeting new eVTOLs, but essentially could be applied for. And so with the idea is that the electric is that the motors would be moved to the rim.

So we don't have the motor at the center anymore. And there are several advantages. You know, it runs in low rpm, side torque, lower rpm. So it's quieter and, and, and there is an efficiency gain because of the, air losses that the, the tips from north there.

and one, one interesting aspect of that one is, we have done that parallel project using AI. and the idea was say, well, if you imagine I'm flying my, my aircraft, my craft, let's call it, it could be different shapes for my craft and fly my craft. And, and then I have a few discrete sensors in my craft, pressure sensors that tell me. So let's say I have, 1050, you know, that a reasonable number of tool and you don't have a full map of what's happening next.

so we then developed the technology where we, we're using CFD data, where we have all the pressure on the surface, force various conditions, and then we can create relationships that say, well, if my the pressure at that point where I have my sensors here, here, here and there is telling me, give these numbers, then I can. So then the pressure everywhere else would be such and and what that allows is in principle. And I think that could be interesting in terms of sustainability in the future, is that you can trim your craft or you can predict behaviors, you know, you can in theory, you can predict early signs of a stall from that.

Or you can, you can optimize, you can trim your craft to run a slightly more efficient, angle of attack and can to condition based on using, you know, that that very raw, very simple piece of information isn't correct. So I, I see this problem on the water based, very incomplete set of information coming from the, the, the real life application. But he hadn't seen it with all the computation and the simulation data. So that that's an area that he that is growing.

I think some people are exploring, but it's not quite there yet. It's good, it's good, it's good, it's good. It's interesting. So we have two questions to wrap it up.

One for you. Gerson, what excites you most that's going to happen in the next five years and let's leave it racing. What excites you the most? Is it the new requirements that are coming out with Formula One, where you now also have a battery in it?

So you kind of have a hybrid now is it is it the pure EV races that are popping up more and more and gaining in popularity, is it more different, more and or different technologies? What excites you the most about the space? yeah. Well, I, I'm sorry for being, going on the same, same note again, but I would say I think I'm actually quite excited about how everything would change with AI.

And I think, you know, fundamentally, I think, motorsport would be the first layer to see that, how significant the change will be. and I just think, you know, we we, we are we talked to here at, you know, be particular about simulation, but we think the simulation works. Really. There are other areas, such as you'd run a simulation, you need to make a decision about that.

And, the decision accelerating decision making is another very important area. So, you know, can I run my simulation and get a very detailed report already telling me why something happened? And instead of spending an hour doing that and in the formula one workflow, is that super important new shaving those minutes of time, I think I think, it's an exciting area. I think I, I feel slightly I think probably most people feel slightly scared about what's going to happen because there's a change coming and he's never, you know, comfortable.

but I chose to embrace it. And I'm, I think I'm really excited about the changes that are coming in terms of, at the courtesy, I think I think what we should be seeing is, in, in, in some industries and more which it, say is talking about mobility now thinking about, like, commercial aircraft. I think they, you know, the concept of wing, a car, a aircraft where the we get stabilizer. It's been there forever.

There's a lot of data, so it's unlikely to change very quickly. But you think about EV e-tolls, for example, is a completely new thing. And and that people are question what what is what shape should that be. And I think that's where it again where you can generate a million options.

And overnight we are going to be some very wacky shapes coming and things that a human potentially wouldn't think about. so, so you know, that that area, I think there are some, some, some key areas where, where I think we going to see more on, on the automotive, side and, and transportation and automotive transportation. I think in the, in the end of the day, there's a practicality element. So we want cars which are can see a certain number of people.

You can fit your shopping bags in the boot and you can, you know, like a truck has to be boxy to fit boxes. So there's only so much you can do in terms of changing the shape. But still, I think I think still we're going to see subtle changes. And I think, yeah, that would be quite interesting to see.

Good, good. Last question for you. Nothing to do with racing. Well maybe it has to do with racing, but I, I would doubt it for a second.

But what's what's the next car you're going to buy and why. And maybe it is a race car, I don't know, I, you know, I'm a very boring person for that one. I know, high performance engineering, but my cars have always been very practical. I currently drive a Ford Transit van because I have a family, and I have, my lifestyle is about going company and carry on with bikes and my board in it.

So I drive what I need. I just bought this van, and I'm very happy, so I'm not planning to change. But I think one thing I told my friends is that once you have a van, you are not not having your van again, because I can carry it on my holidays. So my, my next car will be a run again, I think.

Very good. Well, it's like my family, you know, we we never wanted to buy a minivan. Once we had a minivan, we have to family. We never wanted to give it up because we realized how great it is, you know?

But I was the black sheep in Formula one. you know, I used to go. I come in the car park and all these people with this sports car that I'll come with the Toyota Yaris and park there. As.

Nothing wrong with that. Very good. Thank you so much for your time. And really highlighting an area of, of the mobility space that I, and I think many of our listeners may know where near ev as much inside or a perspective to as in many other spaces of mobility.

Again, particularly here racing. So thank you so much for educating us and and giving us some great insights and things to ponder about.