Is the secret to Mercedes F1 domination, having the best data ?

When Lewis Hamilton climbs into his car on the Formula 1 grid, a team of rockstars (armed with the best technology) have already analysed potentially billions of combinations of...
2019 Spanish Grand Prix, Saturday – LAT Images

When Lewis Hamilton climbs into his car on the Formula 1 grid, a team of rockstars (armed with the best technology) have already analysed potentially billions of combinations of car set-up. Identifying the best possible configuration and race strategy combination has the potential to make the difference between a podium or not.

Ahead of each F1 Grand Prix, Mercedes’ Vehicle Dynamic Group (VDG) team simulates and analyses a vast array configurations to optimise the car for upcoming track and weather conditions. We often hear F1 is a team sport, but most people imagine the pit crew, but these are the unsung heroes.

2019 Canadian Grand Prix, Sunday – LAT Images

The race can be won or lost before the engine bursts into life.

The simulator provides a driver’s first opportunity to test new design features and understand how they will affect performance before going on to the track. It generates data over thousands of channels simultaneously, so the engineers need to be able to interpret that data as quickly as possible, focusing on the most important scenarios that impact performance while planning for the unique characteristics of each circuit.

Data is at the core of all the simulation, which involves interactive visual analytics, data science and what-if scenarios to optimise car balance and setup parameters. Target values and parameters are tracked throughout the season. When performance in a particular race is sub-optimal there is more headroom to optimise configuration for the next race.

Mercedes-AMG Petronas depends on TIBCO Spotfire software visual analytics and TIBCO Data Science to collate a vast amount of data every second, visualise it and run machine learning algorithms for insights to make better decisions faster.

2019 Australian Grand Prix, Saturday – Wolfgang Wilhelm

A number of factors contribute to a Formula One car’s performance like chassis design, aerodynamics, configuration and strategy. F1 teams like Mercedes-AMG Petronas Motorsport, invest in simulators that accurately reproduce the real track experience and maximize the benefits of limited on-track testing time.

Graphics, seat adjustment, steering wheel, sounds, and more, are replicated to give the drivers the feeling they are on one of the circuits. The Mercedes-AMG Petronas Motorsport simulator and surrounding technology have proven to be a winning combination, and with growing importance as technology continues to rapidly advance.

Simulators have changed the F1 landscape and yet, the term “simulator” is misleading. Even though it’s not an actual track, but with numerous restrictions on testing, it’s a close substitute for Mercedes AMG-Petronas Motorsport and other teams to real-world conditions. And, it’s paying off!

2019 Canadian Grand Prix, Sunday – LAT Images

The engineers and simulator drivers that dedicate themselves to studying the potentially billions of combinations of car set-up are more than just engineers, they are rockstars, with the benefit of access to impactful data to help guide decisions for Lewis Hamilton and Valtteri Bottas’ race cars.

Equipped the full race and car speculations, including the steering wheel, race pedals, and fitted seat, simulators replicate the track environment and reproduce the feeling of being in the car. What takes place in the simulator strongly impacts what happens on the actual race-day track; the right set-up, validated by the simulator, gives the drivers what they need to start on the grid.

“We develop the simulator in a way it to help us understand strategy and set-up, giving our whole team a competitive edge over the competition,”

Ivo Marlais, Lead Simulation of the team’s Vehicle Dynamic Group (VDG), led by Loic Serra, Performance Director.

Serra works with Marlais, and Senior Performance and Simulation Engineer Michael Sansoni, to help produce the fastest car possible. They use the simulator to test components and developments, and drivers’ feedback to understand the implications on car performance and race strategy.

The VDG team is also responsible for supporting races with car simulation results. While the simulator group focus on in-depth analysis of the car’s behavior, the VDG engineers run pre-race simulation sweeps and extract optimal setup data, while providing valuable information into upcoming event strategies. It’s no small task to pull supporting simulator data comprising more than half a weeks’ worth of data— a massive amount.

It’s not all fun and games

2019 Canadian Grand Prix, Sunday – LAT Images

While the teams see tremendous benefits from the simulator and pre-event simulations, this does not come without challenges. One of the biggest challenges is handling the vast array of setup configurations and the number of parameters that can be implemented.

With all of these combinations, the teams need to figure out optimal car configurations for upcoming track and weather conditions. These configurations are developed to suit both car and driver, and they provide a significant edge versus competition in qualifying and on race day.   

“With the setup, there can be over a billion combinations in areas we can tweak and setup parameters we can make. Each combination has the potential to be the fastest for that car on that day, that circuit, and that driver. We need to make sure we get that perfect combination, and that’s really what we leverage TIBCO for, and where we find the performance benefits.”

Serra works with Marlais, and Senior Performance and Simulation Engineer Michael Sansoni

It’s the VDG’s responsibility to make sure the team goes to each one of the circuits throughout the season with the best possible car setup, extracting the best possible performance from the car for each race weekend. And on a race-by-race basis, the setup varies tremendously. Depending on the circuit, the team has a very different focus. Knowing and planning for the unique characteristics of each circuit is part of the recipe for success.

“If you take a very short circuit, like Monaco, where the overtaking is difficult, we bias all our work towards qualifying. We want to get the maximum amount out of the car because if we qualify on pole, we think we can win the race from there.

But then you’ve got other circuits where overtaking is easier, like Baku. In those cases, we’re trying to find the best race performance. So we’re looking at the race scenario, consistent lap understanding of whether the car can perform with heavy fuel and worn tires,”

Ivo Marlais, Lead Simulation of the team’s Vehicle Dynamic Group (VDG), led by Loic Serra, Performance Director.
2019 Canadian Grand Prix, Sunday – LAT Images

Finding performance and saving milliseconds

The purpose of the simulator is to help Mercedes-AMG Petronas Motorsport setup the car to run faster, to rapidly advance car development, and to increase the team’s ability and speed to fine-tune advancements during the season.

“The main output for us is finding where we need to improve, where our competition is stronger, where we have weaknesses, and where we have performance areas to advance. Really, that’s the main aim.”

“We try to incorporate all of the things the drivers feel when they’re in the car—steering wheel, sound, and visuals.”

Ivo Marlais, Lead Simulation of the team’s Vehicle Dynamic Group (VDG), led by Loic Serra, Performance Director.

An F1 simulator also provides a driver’s first opportunity to test new design features and understand how they will affect performance before going on to the track.

Data: the fuel of simulation work

Data is at the core of all the simulation work. And, with over a billion possible setup combinations, the team needs to quickly filter through the data to attain the optimal set up. This involves interactive visual analytics, data science and what-if scenarios to optimize car balance and setup parameters. Target values and parameters are tracked throughout the season. When performance in a particular race is sub-optimal there is more headroom to optimize configuration for the next race.

“We’re talking about thousands of channels we’re monitoring as we operate the simulator. Accumulating data for each of the runs, storing it, and then accessing and analyzing it later and on the spot.”

Ivo Marlais, Lead Simulation of the team’s Vehicle Dynamic Group (VDG), led by Loic Serra, Performance Director.

Of course everyone, including the competition, would like to know how much time Mercedes-AMG Petronas Motorsport spends in the simulator, but that’s a matter of closely guarded competitive lore.

According to Marlais, one of the keys to the team’s success is understanding what areas are important and focusing on those. The team already has a good idea how the car needs to progress during the season and what it needs to react to heading into the next race. Fine-tuning previous set-ups and simulations is then of paramount importance.

“We’re generating more and more data, and we need to process and deliver it in a way that the engineer can interpret it as quickly as possible,” said Serra.

2019 Azerbaijan Grand Prix, Sunday – Steve Etherington

Squeezing the noise out of the data

To collect and understand data from the simulator, the team turns to TIBCO software, in particular TIBCO Spotfire visual analytics and TIBCO Data Science software to constantly monitoring new data that’s available for analysis. According to Marlais, the team uses MATLAB to manipulate raw simulation data and output the result to Spotfire for visual analytics. Spotfire’s data connector and data function connections to TIBCO Data Science software, provide access to all data sources, simple-to-use data science expressions, and a comprehensive graphics canvas. These capabilities enable rapid interactive visual analysis of pre-event simulation sweeps, with filtering and brush-linked markings that help generate better-informed insights about the vehicle instantly. In addition, setup parameters are monitored throughout the season to guide setup for future races.

“Spotfire analytics are used to filter and quantify data in a more understandable manner. It helps us filter out some of the noise and gives us metrics, which we can use to understand the statistical variation.”

Ivo Marlais, Lead Simulation of the team’s Vehicle Dynamic Group (VDG), led by Loic Serra, Performance Director.

Spotfire software provides results in just minutes. The old way of analyzing the data meant waiting for the engineer to take a screenshot of it, overlay the comparison, then circulate it via email and wait for analysis. It was critical time wasted on manual tasks that the Spotfire solution quickly performs. Spotfire’s data streams and connections to TIBCO Data Science are used to engineer features, predict parameters and visualize results on streaming events. The software enables engineers to rapidly sense, respond, and adjust the focus to the important combinations of parameters.

Spotfire software also helps the team streamline the process, providing a centralized location for racing intelligence, rapid visualization, and interrogation of all current and accumulated data. Spotfire helps the team filter out mistakes in the simulation phase and establish what the driver does consistently, lap after lap. The more the team explores the car setup running in the simulator, the more it can understand the behavior of the car. And in understanding the car, and digging into the details using the TIBCO Spotfire and Data Science environment, the team can figure out where to improve performance.

“Spotfire has made our processes more efficient. It has made us view our data more clearly, focusing on the details, and distributing and sharing our findings across the company,” said Marlais.

Collaboration across teams is vital

Spotfire ensures that everyone on the team with an interest in setup is viewing the data to bulletproof and validate the same conclusions. Every development to the car needs to derive from the data that is analyzed and the conclusions made among the group of specialists.

With everyone looking at the same view and validating the same conclusions, setup decisions can be implemented instantly by the pit crew and the mechanics. The process enables the team to be more proactive and react faster to uncertain situations, such as a broken front wing.

“We need to be able to cover that 0.1 percent of uncertainty, and that’s what TIBCO provides us. It provides us with the ability to react quickly and to make those changes almost instantaneously to help the car progress to be the quickest come Sunday afternoon of race day.”

Serra works with Marlais, and Senior Performance and Simulation Engineer Michael Sansoni

One of the biggest simulation testing regulations the team must deal with is the Friday curfew before a race day. F1 teams are prohibited from operating at the track for a period of eight hours overnight. Because the team has already performed two to three practice sessions, it has to be very quick in its analysis and conclusions from those sessions to give the mechanics a chance to setup and improve the car before the next morning.

The team combats these regulations with intelligent and succinct collaboration. Because the team is limited in the number of personnel who can work at the track, it relies on various remote sub-teams, all specialists in their unique areas. Teams stationed at the factory who are simultaneously analyzing the data enables the trackside team to make better improvements. Everyone is looking at the same data in unison—no matter their role or where they are located.

Management of the simulator is also a team effort. There’s the test engineer who defines the program and the test items. There’s the engineer who is looking at the data live and processing it as it comes off the car. There’s the operations team that is ensuring accuracy between what the car is doing and the data being collected. And then there are the drivers.

But it’s the interaction between the engineer and the operations team that evolves the car throughout the day. Collaboration is essential to guarantee the right results are communicated, and Spotfire software is one of the leading conduits to ensure this happens.

More cross-departmental collaboration takes place when the wind tunnel team provides the VDG team with aerodynamic data inputs to the simulator. The simulator team takes the recommended car setup prescribed by the aerodynamics team, evaluates it, and tests it on the spot to determine the potential success. Defining target values for aerodynamics and high pressure processing (HPP), and monitoring these, is increasingly important as the season progresses. Spotfire enables tracking and comparing these parameters across track setups and optimizing the parameters throughout the season.

“Before we can even run a simulation, we have to validate each one of these inputs, ensure that the accuracy is correct, and ensure that it’s correctly representing what the physical component is doing.

Once we have all of those components, we can begin simulations to understand how the car will behave once you put all of the individual, unique components together.” 

Serra works with Marlais, and Senior Performance and Simulation Engineer Michael Sansoni

Hence, informed by all the fresh data, the new aerodynamic platforms it receives, and the optimum performance from suspension and power units. All of these components need to be combined to ensure that out of the multitude of combinations, the fastest structure can be chosen.

Mercedes-AMG Petronas Motorsport is always looking for the next level of advancement in driver immersion and process improvement. The goal is to seamlessly emulate the track environment and respond quicker than in the past. TIBCO Spotfire and TIBCO Data Science software have given the team the understanding and ability to rapidly and comprehensively analyze and visualize the data it needs to evolve performance.

With each circuit providing a unique set of characteristics and data to analyze, the team can learn a lot by leveraging previous configurations, understand how those setups work, and identify shortcomings. This information, in addition to the magnitude of simulation data, allows the team to push its limits and keep on improving and innovating.

“We can apply predictive algorithms to understand what changes we made at previous events to learn and predict what we’ll do at future events. Using similar examples we’ve come across in previous seasons or circuits gives us direction,”

Serra works with Marlais, and Senior Performance and Simulation Engineer Michael Sansoni

Applying a digital twin to your company/real-world business scenarios

You might be wondering how can the work being done in a simulator for a high stakes F1 competition apply to my company? Well, the lesson learned is one of measure and response, often articulated as sense-and-respond and this applies to many industry sectors and use cases.

What the F1 simulator and all of its surrounding technology demonstrate is that If you are a technology or business leader, you can take a vast amount of data in a short period of time, visualize it and run machine learning algorithms to derive insights and patterns that enable collaborative tech and business teams to make more informed decisions faster. This improves the odds of getting ahead in a very competitive business world.

For every asset and product, one can develop virtual replicas, in software, of the same functionality – that is the concept of digital twins and one reason why they are now consistently used to produce better results and deeper insights for operations optimization. More and more companies are adopting digital twins to survive their ever-growing competitive markets.

Digital twins help companies like Mercedes-AMG Petronas Motorsport address their biggest data challenges. The end results include process optimization, insights into predictive and condition-based maintenance, and optimal business action on the event stream, with the potential to lead the team to its sixth Constructors’ Championship.

Winning with TIBCO’s digital twin capabilities

TIBCO has made a unique contribution to digital twin technology, especially in the industrial internet and specifically in the high-tech manufacturing and energy sectors. The TIBCO approach includes visual analytics and data science to glean insights and leading indicators from recent data, and applying these insights into current event streams to monitor equipment performance and identify anomalies. Anomalies are case-managed to resolution and learnings are fed back into the analytics and data science.

TIBCO also has solutions for data unification, including master data management and integration on the event stream and with accumulated data sources. The combination of analytics and data management is especially powerful, enabling management of master data, metadata and reference data. Such data management and modeling sets up a comprehensive system for rapid data model updates that flow through the analytics seamlessly.

TIBCO makes its digital twin and analytics solutions available on premises and in the TIBCO Connected Intelligence Cloud in the form of accelerators and starters that can be freely downloaded by interested users. These TIBCO technologies and starter solutions increase interconnectivity, augment the intelligence of the IoT, and expand the edge of digital business.

Categories
GeneralSports

Creator of techAU, Jason has spent the dozen+ years covering technology in Australia and around the world. Bringing a background in multimedia and passion for technology to the job, Cartwright delivers detailed product reviews, event coverage and industry news on a daily basis.
One Comment

Leave a Reply

  • Ellerton
    19 July 2019 at 10:37 pm

    Driver input and driver integration into this mammoth data driven environment are mostly missing from this dissertation. At the end of the day, we the F1 fan and enthusiast see the driver and car on track and are always curious about how they work together. I think the piece would have been more interesting if the drivers’ inputs, reactions, etc. were part of the presentation. What happens, after all the simulation and data work, when the driver goes on track and returns to the pit and says: this set up is shit..

    Leave a Reply
  • Related Posts