Formula One car designers and mechanics now rely on IT-based performance-analysis tools to give their cars a winning edge.
By the time the current Formula One season finishes in Sao Paulo at the end of November it will have witnessed competing teams generate, collect and analyse more information about the performance of their cars than ever before in the sport's history. To handle the considerable processing load, F1 engineering squads have beefed up their back-end IT systems to crunch the many terabytes of information as quickly and accurately as possible in the hope it can earn the vital split-second advantage that could win a single race, andeventhechampionship.
Competing teams have left no computing technology untouched, and no IT expense spared when it comes to forcing the last torque of out of their cars, while maximising driver safety. High-performance computing (HPC) clusters are used to process huge complex CAD (computer-aided design) images, while computational fluid dynamics (CFD), which some teams are using to augment (or completely replace) physical wind tunnel testing, requires massive number-crunching. Mini-mobile data centres can be flown to any circuit in the world to enable trackside telemetry and communications networks to convey real-time data to mechanics during races.
Bill Peters is head of IT at Team Lotus, formerly Lotus Racing, for whom drivers Heikki Kovalainen and Jarno Trulli are this season battling it out on the track with rival Williams, Force India, Sauber and Scuderia Toro Rosso teams in T128 cars powered by Renault engines. 'There are essentially four phases to developing an F1 car: conception; design; construction; and running it,' says Peters. 'IT helps us across all areas, helping us meet FIA (Federation International de l'Automobile) requirements, then simulating and building virtual cars.'
Engineers use CAD software at the initial design phase, as well as finite-element (FE) stress analysis software, which puts artificial loads on virtual mechanical components with pre-defined properties to make sure they are sufficiently robust. Product lifecycle management tools also play their part in'managing the design and construction process, as do enterprise resource planning (ERP), and computer-aided manufacturing and inspection applications.
Computational fluid dynamics
It is the use of computational fluid dynamics (CFD) software, which builds computer simulated models of whole cars in order to improve their streamlining, that has attracted the most investment. Indeed, the FIA has introduced rules that limit the number of hours of CFD modelling each team can perform within its data centres to a maximum of 40 teraflops in any one eight week period, dependent on how many hours the car spends being tested in a physical wind tunnel.
Lesmana Djayapertape is head of CFD at Team Lotus, which recently equipped its Norfolk data centre with high-performance computing (HPC) equipment based on Dell rack and blade servers and 16TB of storage arrays.
'CFD will show you how fluid or air behaves when it flows around an object, and then tell you where the down-force is coming from,' Djayapertape explains. It also allows engineers to examine specific parts of the car from as many angles as are required, to see the effect on airflow of mounting a driver camera just before the cockpit. For example: 'You can also see the track situation with a 50 per cent model in CFD then take it to the wind tunnel to see how things change when you take a full-size car to the track.'
Other teams, like Virgin Marussia, use CFD exclusively but, Djayapertape says, the technology is not perfect; it struggles to predict accurately unsteadiness around the car caused by rear brakes or diffusers, for example. As such, Lotus prefers to deploy CFD as more of a filter that can help direct further aerodynamic and fault-testing in a physical wind tunnel.
'There was a different attitude in F1 last year, and we [Marussia Virgin] had a lot of people saying it is never going to work,' says Mark Davis, marketing services director at CSC, the IT services company that supplies the Marussia Virgin team with an HPC cluster consisting of around 10,000 compute nodes that runs its full 40Tflops of CFD modelling and testing allowance at a new Banbury data centre; 'but the team has made tangible improvements throughout the season, and we are beginning to see how CFD capabilities can transform what we are doing.'
Away from data-centre CFD processing, all F1 teams ship a variety of IT equipment to race locations around the world, including servers, workstations, network switches, storage appliances and all the security, monitoring, voice over IP (VoIP), messaging and collaboration applications that go with them, plus the other kit and caboodle that IT engineers need to have to hand. Team Lotus also keeps some services running in the cloud as a back-up in case the trackside IT infrastructure goes down for whatever reason, with pitwall electricity supplies notoriously unreliable in some locations.
'The trackside IT environment is like a travelling circus that has to be put up and taken down every race weekend,' says Team Lotus IT head Bill Peters. 'It is very important that if we lost power at the trackside we could continue operations autonomously.'
CSC also provides trackside IT services and data management to Marussia Virgin, deploying a core infrastructure in every race around the world, and assigning two IT engineers to the team with others working back at the research facility. Team members sitting in the pits use a variety of workstations and laptops to collect gigabytes (sometimes terabytes) of information each car generates in a single qualifying lap from the thousands of on-board sensors monitoring each component.
These include anything from oil, water, exhaust and tyre temperatures to speed, engine revs per minute (RPM), clutch fluid pressure and G-force, all recorded at multiple points on the circuit. All of that data is then analysed and digested by the mechanics so that they can make changes for the next day's racing. Data is also compared to historical information that is stored at trackside in storage arrays and storage area networks (SANs), such as those supplied to the Force India team by Infortrend. 'We are staggered to see how many hundreds of thousands of changes are made throughout the season,' admits Davis.
'Running the car is the exciting bit – by monitoring our cars throughout the season using analysis and simulation tools we can really make a difference in terms of competitive advantage,' adds Bill Peters. The Vodafone McLaren Mercedes team has taken this one step further by developing an application for 2010 called Race 1.0b Live Data Viewer that shows not only details collected by the car's sensors but also comments from the pitwall and mission control, and live GPS circuit map data to fans via the Internet.
All that data places inevitable strain on the network bandwidth available at each track, which due to their location can often be limited – one reason why the Williams F1 team signed a deal in April 2011 to use telco AT&T's wide area network (WAN) acceleration service to help handle the estimated 30GB per event it transmits back to base engineers during races.
How telemetry systems work
While the FIA has called for the standardisation of F1 telemetry systems, they remain a mixed bunch with lots of companies supplying the radio communication equipment which forms the telemetry networks. These can be based on a range of different wireless and wired technologies, ranging from Wi-Fi and mobile radio (PMR) technologies to gigabit Ethernet freespace optics and satellite links, with teams often reluctant to reveal exactly what they are using for competitive reasons.
Microsoft supplies the software that runs in the McLaren Electronic Systems (MES) engine control units (ECU) now used as standard in most F1 cars. This is the small box that controls car behaviour – like engine sequencing and spark-plug timing – but also connects the car's onboard wireless sensors to a receiving station back at the pit wall and the garage.
Some, but not all, F1 cars racing this year will be fitted with energy-recovery kinetic energy recovery systems (KERS), and FIA has approved new energy-efficient small engine formula for 2013. The more popular version of KERS uses a super-capacitor battery (such as those used on Toyota Prius car) to store the energy created during breaking, then release that stored power to the driven wheels when required.
The KERS system uses an electronic control unit that manages the behaviour of the electrical motor and links to the main McLaren/Microsoft ECU, information about which is also relayed to the pit wall via telemetry. The data is transmitted to the receiving station using whichever frequency is permitted for use by the local authorities (allocation varies from one country to another), often the 1.5GHz waveband. The data is compressed in order to minimise the bandwidth used, and the transmitter antennae is usually housed in the nose of each car.
The cars also have an on-board storage system that improves performance by buffering the most recent data and resilience by storing the information for a short period if the wireless signal is unavailable for whatever reason (say, blind spots or tunnels).
Information from the car can also be downloaded whenever the car enters the pit lane. The receiving station, or box, connects to the engineer computers a trackside usually via Wi-Fi, and then onto the Internet through wired broadband connections, but depending on the resources available at individual circuits, in some cases teams use satellite links to additionally relay the information back to the team HQ datacentre where higher volumes of historical data on the car's performance are held. *