Formula 1 2018 Technical review - The Chelsea Magazine Company
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
30 & 31 October 2019 | NEC, Birmingham The UK’s largest annual gathering of OEMs and engineering supply chain professionals 91% of exhibitors met their 88% of exhibitors have either 98% of exhibitors were satisfied objectives of networking booked or will be booking with their experience at with new and existing clients again for 2019 Advanced Engineering 2018 Secure your stand today “We have done this for the 5th “I found the event a great year in a row and it is a very networking opportunity to meet good event for us to meet new industrial professionals from clients.” different backgrounds with Thomas Evans, Business different products.” Development Manager, Kat Clarke, Wing Manufacturing EBS Automation Ltd Engineer, Airbus Secure your stand now www.advancedengineeringuk.com
CONTENTS M ercedes dominated the 2018 This year also saw the introduction of the 4 Unintended consequences Formula 1 season, wrapping up Halo head protection system, which has changed How an attempt to curb aero the constructors’ title as well as the the ethos of the world’s premier single seat development has had a dramatic drivers’ title for Lewis Hamilton. But, series. Open cockpits have always been at the effect on CFD in Formula 1 in 2018 this was not a walkover by any stretch. The team heart of open wheel racing, but the Halo has 14 Pressure sensors believed, as did many observers, that Ferrari had fundamentally changed that. Simon McBeath and the latest the better car in the middle of the season, and The new mounting hardware, and meeting advances in scanning technology was unable to convert that advantage into a title. some incredible crash testing targets, means 20 Halo protection We are now coming to the end of a rule-set weight. This has led to a technical challenge for Gemma Hatton on what makes the and finally we have the 2019 regulations to the Formula 1 teams, one that no doubt will be Halo such a capable safety device hand. There have been steps taken to improve refined in the 2019 racecars. 28 Safety drive overtaking, to increase the life of engines, and Further safety features will come next year, in New helmet design for 2019 revealed to reduce the array of tyre choices that are the form of new helmet design. Head protection as the FIA steps up safety campaign available for each race, while retaining the spirit has been a focal point of recent FIA investigations 34 Big wheels keep turning of competition. But this is not the end of the and only now are we starting to see the results Why Formula 1 has finally taken the story. As we discovered this year, the regulations that will benefit drivers in the near future. On a plunge and gone for 18-inch wheels surrounding the computing restrictions need lighter note, the cars will be fitted with 18-inch 42 Cool runnings work, and since we ran the article (see page 4) the wheels in future, which will have an extraordinary Sam Collins uses Caterham’s 2014 FIA has announced a change, but not one that we impact on suspension and chassis design. data to explain F1 cooling tech expected. While the computing power limitations One shadow on the horizon is the news meant a waste of money, the FIA has opened up that Porsche will not be coming into Formula 1. Produced by Racecar Engineering team those restrictions to ensure more data comes This lack of new blood is worrying; unless new its way, and therefore it will be in a position to manufacturers join, Formula 1 must survive do a more competent job on the next set of with the power unit suppliers currently in play regulations, due out in 2021. (Renault, Ferrari and Mercedes). This puts them in It was a smart move, but the costs are within a very strong negotiating position, as the loss of the reach of only the wealthiest of teams. In this one of them would have far-reaching impact on supplement, we also look at pressure sensors, and the remaining manufacturers. how they can impact the rate of development and ANDREW COTTON the lessons learned in testing. Editor FORMULA 1 · WINTER 2018 www.racecar-engineering.com 3
TECHNOLOGY – AERODYNAMICS Know your limits This year Formula 1’s CFD restriction regime has been shaken up big time as the FIA looks to cut the costs of aerodynamic development. But has it worked, and how has it changed both the tools and the process? Racecar investigates By GEMMA HATTON I n January 2018, the FIA introduced the latest evolution of aerodynamic testing restrictions for Formula 1, and with them came the biggest change in CFD restrictions since they were first introduced back in 2009. Racecar went behind the scenes with HPC specialist, Boston Ltd, to discover the impact of these changes and how Formula 1 teams have not only benchmarked new solutions, but also upgraded their CFD supercomputers. But to put these latest changes into context we need to understand the history of the restrictions, both for CFD and the wind tunnel. In 2008, aerodynamic testing was at its peak. BMW Sauber, Honda, Williams and Toyota had all invested huge sums of money in new state of the art full size wind tunnels, each costing tens of millions of pounds. All the top teams were operating in two wind tunnels simultaneously, while Toyota was not only using two wind tunnels 24/7, but each of these was full size. However, the vast majority of this wind tunnel testing utilised scale models, and over the years the scale of these models increased from 40 per cent to 50 per cent and then 60 per cent. Operating two wind tunnels full time allowed these teams to complete around 500 wind tunnel simulations UNIFI LTD per week, with each simulation incorporating approximately 20 different car attitudes. Full size wind tunnel testing was commonplace, with teams either using their own facility or a Boston worked together with UniFi and CE to benchmark the performance of new CFD technologies in customer facility such as Windshear in the USA. accordance with the 2018 regulations to see if F1 teams would be forced to upgrade their CFD capability 4 www.racecar-engineering.com FORMULA 1 · WINTER 2018
It quickly became clear that something had to be done to curb the growth of aerodynamic testing in F1, and its associated costs NOVEMBER 2018 www.racecar-engineering.com 5
TECHNOLOGY – AERODYNAMICS In 2008 teams were already using Between 2009 and 2017 the regulations evolved wind tunnel. For example, let’s assume that CFD CFD routinely as part of the aerodynamic and generally served to reduce the aerodynamic capacity allows a maximum of 12.5 TeraFLOPs. development process, and as the software resources available to the Formula 1 teams, Using the equation with the 2013 limits results and correlation improved while hardware particularly in the wind tunnel. This was done in 41.3 hours of wind on time, as shown by the costs reduced, teams began to use it more, through introducing a ‘limit line’ which is green square. In 2015, however, 12.5 TeraFLOPs integrating it further into the design cycle. defined by the following equation. would only give you 12.5 hours in the wind At that time, BMW Sauber was leading tunnel (blue square) – that’s 70 per cent less the way in CFD hardware with the Albert 3 than 2013. The exact balance between CFD and supercomputer and over 4000 Intel cores, but wind tunnel resources varies from team to team, other leading teams were not far behind. It Where: and sometimes from year to year, depending quickly became clear that something had to be WT = Wind on time on the strategic approach and technology done to curb the growth of aerodynamic testing WT_limit = 25 hours advances adopted by each team. in Formula 1, and its associated costs. CFD = TeraFLOPs usage Of course, every restriction that is The first step came into force in January CFD_limit = 25 TeraFLOPs introduced simply triggers the teams to 2009 as part of the FOTA Resource Restriction exploit the loopholes and optimise their Agreement (RRA). This controlled the Therefore, the amount of time a team chose to designs and working practices to maximise aerodynamic resources the Formula 1 teams run its CFD directly dictated how much time it their performance from the regulations. For could deploy via restrictions on the wind tunnel could utilise the wind tunnel. Equally, if a team the TFLOPS CFD restrictions, this became ‘wind on time’ (WON) and the CFD compute could complete its maximum allocation of wind an arms race as teams pushed to develop capacity, measured in TeraFLOPS (TFLOPS). tunnel runs using less wind on time then it their supercomputers to run the most CFD Wind on time was simply a measure of would have more capacity for CFD simulations. simulations per given TFLOP allowance. This led the amount of time the fan was turned on in teams to operate CFD hardware in ways which the wind tunnel with the wind speed in the Working area were quite different from the wider industry, test section above 15m/s. For CFD, TFLOPS Looking at the WT_limit and CFD_limit data with a clear focus on regulatory efficiency was effectively the number of floating point from the last few years, Figure 1 can be created. rather than financial efficiency. For example, operations completed within the designated Essentially, by plotting the maximum of each of the TFLOPS calculation naturally includes a chip eight week Aerodynamic Testing Period (ATP) these limits, you can establish the ‘working area’ clock speed term which is reported either as the and was defined by the following equation: that the teams could operate in. For example, maximum turbo clock frequency stated on the in 2013, when the maximum WT_limit was 60 CPU specification (if the turbo mode is used), hours and the maximum CFD_limit was 40 or the base clock frequency if the turbo mode TeraFLOPs, the team could operate anywhere is not used. Teams quickly established that the Where: within the green shaded area. In 2014, the turbo mode was not an efficient way to run TotFLOPs = Total number of TeraFLOPs used per second limits were 30 hours WT and 30 CFD TeraFLOPs, CFD simulations, in terms of the number of CFD MFPPC = Peak double precision floating point operations illustrated by the red shaded area, whilst simulations completed per TFLOP. This was also per cycle per core of the processing unit 2015 was limited to 25 hours WT and 25 CFD true for many higher clock speed chips. CCF = Peak processing unit clock frequency in GigaHertz TeraFLOPs, represented by the blue shaded area, Effectively, running supercomputers with NCU = Number of processing unit cores used for the run which remained the same until 2018. slow clock speed was giving teams more NSS = Number of solver wall clock seconds Since 2013, you can see that overall testing efficiency under the regulations but with the elapsed during the run has dramatically reduced, but particularly for the obvious penalty in terms of CFD simulation Figure 1: The FIA has restricted aerodynamic testing over recent years for both CFD and the wind tunnel, but particularly the latter. This graph shows the ‘working area’ that the teams have been able to operate in. Assuming a maximum CFD capacity of 12.5 TeraFLOPS you can see that wind on time has dropped by 70 per cent between 2013 and 2015 6 www.racecar-engineering.com FORMULA 1 · WINTER 2018
TECHNOLOGY – AERODYNAMICS turnaround time. Therefore, teams then had to regulation, but it was very inefficient financially, the FIA agreed to consider the rival Intel chips balance the speed with which they receive their with as much as half of the purchased HPC (Sandybridge and Ivybridge) as four dp flops/ CFD results against the total number of CFD compute cores being left idle. cycle for the purposes of the regulations rather simulations they were able to complete within The biggest issue came when one of than their rated eight dp flops/cycle. the regulatory framework. This is quite different the teams developed the Fangio chip in By 2012 AMD had been persuaded by many to the wider CFD industry, where the turbo collaboration with AMD, a chip specifically teams to produce a second limited run of Fangio mode was ‘free’ performance and quicker designed to optimise the balance between CFD chips, allowing more of the grid to upgrade clock speeds were performance gains if your case turnaround time and throughput which their supercomputers to this specification, main criteria was financial efficiency, and so gave that team a huge initial advantage. with most of the remaining teams running an the divide between the two environments was This exploited the fact that the modern Intel Ivybridge system. With the FIA unwilling underway from 2009 onwards. HPC chips were then rated at eight double to extend the flops/cycle exemption to more precision flops/cycle but commercial CFD modern Intel chips, such as the V3 Haswell Cores and effect codes were only capable of delivering CPUs which were rated at 16 dp flops/cycle, and Core under-population also became approximately one dp flop/cycle. The Fangio AMD not producing any more Fangio chips, the commonplace in Formula 1 as it delivered chip was designed to operate at two dp flops/ teams were now locked into these older systems further regulatory efficiency gains for the cycle giving a big efficiency improvement in purely by virtue of the regulations. Newer chips teams. It was efficient for the FIA TFLOPS MFPPC. Following lobbying from various teams, were simply not viable because of their high Fully correlated and complex CFD models, such as this by Simscale, are becoming an ever-increasing asset to F1 teams, with some full car models now exceeding one billion cells Every restriction simply triggers teams to exploit the loopholes, and optimise designs and working practices to maximise performance 8 www.racecar-engineering.com FORMULA 1 · WINTER 2018
TECHNOLOGY – AERODYNAMICS BOSTON LTD A typical HPC cluster from Boston. With each new generation of compute chip delivering up to 20 per cent efficiency improvement the increased capacity of modern CFD clusters means that teams can now have an extra 200 runs, as opposed to 20 back in 2009 flops/cycle rating. These older systems were WON was retained and a parallel regulation was node system based on the Intel Skylake 8176 coming to the end of their life and were no introduced with the aim of allowing teams to Platinum 28 core chip. The group also had longer supported by Intel or AMD. continue using their old systems if they wished, access to a smaller four node Intel Ivybridge Clearly the FIA had to do something, and the without too large a performance penalty – at HPC which was used to provide a baseline of the target was to introduce a new regulation which least that was the intention. performance gains that teams could achieve by aligned the Formula 1 aero departments more Boston Ltd has been specialising in high upgrading from their older systems to the new closely with the wider CFD industry as well as performance computing (HPC) in a wide range hardware. This allowed Boston to benchmark its allowing teams to upgrade to more modern, of sectors for over 25 years. In 2017 it formed own internal CFD model across a range of CFD supported technology. This resulted in the 2018 a new partnership with Tim Milne of UniFi codes with a wide variety of hardware set-ups. CFD restrictions and a move from TFLOPS to Engineering Services Ltd (UniFi) and Dr Lee The systems were all set up with the very latest Mega Allocation Unit hours (MAUh) as defined Axon of Computational Engineering Ltd (CE). in networking fabric, up to date operating by the following equation: Milne and Axon have extensive Formula 1 systems and storage solutions, ensuring that experience, most recently at Manor F1 where the results obtained would be aligned to the they were head of aerodynamics and head of expectations of the F1 teams. CFD correlation respectively. This group combined Boston’s extensive Hot chips HPC technical knowledge with UniFi’s and Following the benchmarking of the older Where: CE’s F1 aerodynamics and CFD experience to Ivybridge system, a number of options within AUh = Allocation unit hour provide the F1 teams with a comprehensive the AMD EPYC range as well as the Skylake NCU = Number of processing unit cores benchmarking of the new AMD EPYC and Intel 8176 chip were evaluated as single node tests NSS = Number of solver wall clock seconds Skylake Platforms. They were able to use all the to gain an initial assessment of the various elapsed during the run main F1 CFD codes with models aligned to F1 different chips available in each family, as CCF = Peak processing unit clock frequency in GigaHertz methodologies and HPC hardware set-ups to well as some insight into the time/iteration extract the maximum possible performance performance benefits of different options such Effectively this a very similar measure to TFLOPS from the new regulatory environment. as the turbo mode. This also ensured that a clear but without the reliance on flops per cycle, understanding of the raw performance of the hence removing the barrier to upgrading to Node to joy compute chip was gained and that the results newer, better supported, technology. The FIA The project began in August 2017, by which were not clouded by any networking issues commissioned an independent study to be time Boston Ltd was one of the first companies which could be useful later in the process when carried out in order to set the regulation limit worldwide to have invested in its own eight trying to understand the results on the larger with the intention of giving parity between the node dual socket AMD EPYC system based on scale multi-node systems. The performance old regulations and the new ones. The link to the EPYC7601 32 core chips and a similar eight gains over the older Ivybridge system were The new method that was introduced at the start of this year is a very similar measure to TFLOPS but without the reliance on flops per cycle 10 www.racecar-engineering.com FORMULA 1 · WINTER 2018
INTRODUCING... THE BOSTON INTEL® SELECT SOLUTION BUILD YOUR SUPERCOMPUTING INFRASTRUCTURE ON BOSTON'S EXTENSIVE INDUSTRY AND DESIGN EXPERTISE FOR HIGH PERFORMANCE COMPUTING APPLICATIONS. BOSTON’S CFD SERVICES THE DIFFERENCE BETWEEN WINNING AND LOSING Our team of motorsport experts are on hand to support with code profiling and optimisation, can aid in tuning meshing and solver performance, and more... VISIT: WWW.BOSTON.CO.UK/AUTOMOTIVE/CFD.ASPX WEB: WWW.BOSTON.CO.UK EMAIL: SALES@BOSTON.CO.UK PHONE: +44 (0) 1727 876 100
SIEMENS TECHNOLOGY – AERODYNAMICS Despite the efforts of the FIA to restrict the costs of CFD work, Formula 1 teams can now complete up to 1500 simulations each week on a typical model of around 200 million cells very quickly evident and it soon became clear that the teams would all be forced to For example, the CFD case will be repeated on the same HPC system but testing the It soon became clear that upgrade their HPC systems in order to remain competitive, which is the nature of Formula 1. simulation on 48, 96 and 192 cores. It was accepted that the case being run on 96 cores the teams would all be forced to upgrade their But this upgrade was extremely expensive. This will take slightly longer than half the time of is not what the FIA had been aiming for, but the case on 48 cores and slightly less than HPC systems in order to reflects how quickly the HPC industry moves half the case being run on 192 cores – so forward with the Formula 1 environment forced there is an element of inefficiency by running to follow suit to remain competitive. on an increasing number of cores. However, Once testing migrated onto the full, multi- it is in the teams’ interest to complete their remain competitive node systems the full optimisation process CFD simulations quickly in order to allow could begin. This involved running the same their iterative aerodynamic development model over a wide range of different set-ups, programmes to continue as quickly as possible delivered performance gains which would including options for memory bandwidth – so it’s a trade off and one which was vitally enable the F1 teams to run approximately twice per core used and process bindings. The key important for the Boston group to understand. as many CFD simulations per week in 2018 than at this stage was for the group to develop an they had been able to in 2017 (for the same understanding of the efficiency vs performance Core values wind tunnel operation). Furthermore, the teams of each compute system – ideally each compute The next step was to understand the impact of would complete each of these simulations in chip in each family from Intel and AMD. leaving some of the compute cores dormant, approximately half the time that was required In reality UniFi and CE were able to use their as previously mentioned. This is an approach under the 2017 regulations. experience in the industry to limit the testing quite alien to most of the CFD industry (why to the most likely candidates for Formula 1 would you buy compute cores and then not Formula 1 specific operations and Boston used its extensive links use them?) but something that was already well Much of this optimisation is not relevant to in the HPC industry to gain access to relevant known to deliver regulatory efficiency in the F1 the wider CFD industry, but is now considered systems for benchmark testing. Once a small environment, if you could afford it. basic within the Formula 1 teams. The next step range of AMD and Intel compute chips had Tests were completed leaving a range was for the Boston group to really exploit the been selected, the focus was on understanding of the cores dormant in order to give less expertise available from the UniFi/CE group. how they performed against the Formula 1 operational cores per memory channel, and The details of this remain confidential, but it regulations. This required repeating the CFD thus increasingly improving the memory enabled the group to develop solutions which simulation of their Formula 1 car on a range of bandwidth available to the CFD simulation. delivered even more performance for the F1 different HPC sizes and set-ups. The conclusion of this benchmarking study teams, and a further 20 per cent reduction Much of this optimisation is not relevant to the wider CFD industry, but it is now considered basic within the Formula 1 teams 12 www.racecar-engineering.com FORMULA 1 · WINTER 2018
The FIA focus remains on reducing wind tunnel reliance and delivering greater CFD capacity in exchange, and the current regulations deliver that The benchmarking study concluded that teams would gain a huge performance advantage if they purchased a new multi- million pound system because they would have twice the CFD capacity of 2017 – this was not the aim of the regulations in solve times was extracted from the same the huge performance advantage available by replacing their CFD clusters. In 2018, with the CFD set-up, which also increased the CFD purchasing a new multi-million pound system, massive increase in capacity, the same 10 to throughput by the same 20 per cent. which was not the aim of the new regulations. 20 per cent improvement available from each Finally, as the benchmarking study neared Furthermore, the impact of the increase evolution of compute chip technology is 100 its conclusion Boston worked with AMD to in CFD capacity available to the teams under to 200 runs – that is the same as the total further optimise for the requirements of F1 these new regulations only serves to increase capacity of the systems in 2009. by increasing the memory bandwidth whilst the financial pressure on the teams and in Is this a bad thing? Arguably not. HPC retaining a relatively low base clock speed. particular the pressure to increase headcount systems are much cheaper now than they ‘AMD EPYC delivers exceptional levels within the aerodynamics departments as the were back in 2009. The FIA focus remains on of performance in a number of workloads, CFD capacity available increases. Not only reducing wind tunnel reliance and delivering including high performance computing CFD have they effectively been required to invest greater CFD capacity in exchange, and the applications,’ explains Roger Benson, the senior in new HPC architecture in order to remain current regulations deliver that. director of the Datacenter Group, EMEA, AMD. competitive, but the incentive to adopt future However, does it help to level the playing ‘We are excited to be working with Boston on improvements in chip technology has now only field between the high budget teams and their automotive engineering focused platforms increased. How so? The benchmarking work the low budget teams? Does it help to and improving the efficiency of aerodynamic completed by Boston suggests that teams are encourage new teams into the sport? And testing for their customers.’ now able to complete between 1000 and 1500 does it make the working practices within the CFD simulations per week based on a typical Formula 1 aero departments more aligned The results CFD model of around 200 million cells. Teams to the wider CFD industry? The stated targets of the FIA for this change may elect to ‘trade’ some of this capacity for With AMD releasing its second generation in regulations was to enable the F1 teams larger models (some teams run CFD models of EPYC chip in 2019, the reaction of the to upgrade from their Fangio and Ivybridge approaching one billion cells) or better quality teams will be interesting. Will they all upgrade systems to the latest technology available, but models (transient simulations rather than steady immediately? Or will the well-funded teams without a clear performance pressure to do state). But the key point is that the F1 HPC take the opportunity to get a performance so, and with the aim of better aligning the F1 regulations have now given the teams twice as advantage from the new technology that industry with the wider CFD industry. much capacity to play with than in 2017. the smaller teams cannot afford? Firstly, it is clear that all the F1 teams have upgraded to a new system, with most teams Step change Boston, UniFi and CE continue to develop having done so ahead of the regulatory change Typically each generation of compute chip their partnership with a focus on the F1, date of 1 January 2018. So, the first aim has that is released by AMD/Intel delivers around motorsport and automotive industries across been achieved – the Fangio and Ivybridge 10 to 20 per cent improvement in efficiency. all CFD codes and working practices. For systems that the teams were operating are now Back in 2009 this would give the teams an extra more informaition visit the websites at: www. obsolete. However, the benchmarking work 10 to 20 CFD runs per week, and therefore boston.co.uk; www.unifimotorsport.com; completed by Boston clearly demonstrates would not easily justify the large cost in www.computationalengineering.co.uk When AMD releases its second generation of EYPC chip in 2019 the reaction of the Formula 1 teams is going to be very interesting FORMULA 1 · WINTER 2018 www.racecar-engineering.com 13
TECHNOLOGY – PRESSURE SENSORS Pressure relief Thanks to a tiny new device, measuring aerodynamic pressures on and around cars may have just become a whole lot easier, and in some areas actually possible for the first time. Racecar investigates By SIMON McBEATH P hotographs of pre-season themselves, via tiny pressure port tunnel as well, to accurately monitor calibration of pressure sensors and pre-race testing often tappings over and under the major and control flow conditions. for many years, and have worked show top race teams, downforce-generating surfaces. But how are these measurements closely with a concern that might especially in F1, measuring Using these techniques enables made? And what are the practicalities? accurately be described as the aerodynamic pressures around their aerodynamicists to gather pressure We visited Evolution Measurement sector founder, Scanivalve (see cars. The most obvious manifestation data around and on the car’s surfaces Ltd (see sidebar), based in Andover in sidebar) since 2001. With an intimate of this is the pressure sensor array, like that can be used to correlate with southern England, to learn more about knowledge of the available products a two-dimensional rake with pressure CFD and wind tunnel data, as well as the challenges involved, and to see an and, importantly, a firm grasp of sensors arranged over the area of the to calculate the real forces acting on ultra-compact new pressure sensor the customers’ needs, Evolution rake, positioned on the raceccar in key individual components. the company has come up with, that Measurement has come up with the areas where what might be described Pressure measurement is an will open up exciting new possibilities. EvoScann P Series pressure scanner as partial plane pressure plots can be everyday critical part of wind tunnel Although Evolution Measurement which, they confidently assert, is logged and recorded. testing too, with the wind tunnel is itself a new company, managing the smallest, lightest such pressure Less apparent and no less models equipped with surface director Paul Crowhurst and export sensor currently available – no surprise important is the measurement pressure ports, and measurements sales manager Iain Gordon have then that it has been generating of surface pressures on the cars also being made around the wind been involved in the distribution and interest among F1, DTM, LMP and 14 www.racecar-engineering.com FORMULA 1 · WINTER 2018
Renault RS18 with pressure sensor array. This is highly visible but there might also be small sensors measuring air pressures that are fitted to the car itself The EvoScann P Series miniature pressure scanner is so small (36mm x 33m x 8mm) it can be located in areas of a racecar that have previously been impossible to access MotoGP teams. This compactness – the 8-channel launch version is just applications. Traditionally pressure sensor array and surface pressure other controlled environments’. It is a 16-channel device typically used in These enable 36mm (1.42in) x 33mm (1.30in) x 8mm (0.31in) and weighs only 16g (0.56oz) – measurement has often been done with one of the Scanivalve multi- wind tunnels where installation space is not an issue, overall length being aerodynamicists to gather will enable the sensors to be located in port scanners such as the ZOC (Zero, around 200mm (8in). areas previously difficult or impossible Operate Calibrate) range of analogue The newest and most advanced pressure data to access, such as within front wing devices in 32- or 64-channel form, and Scanivalve product is the MPS elements or other important small these measure 105mm (4.1in) x 36mm (miniature pressure scanner) range, aerodynamic parts. 1.4in) x 14mm (0.55in), finding use in a 64-channel high-end device wind tunnel models and other space- measuring roughly 89mm (3.52in) around and Pressure change limited applications. x 40mm (1.56in) x 22mm (0.87in), To see where and how this new compact pressure scanner fits into Another oft-used device is the popular DSA range, described by several of which might typically be seen on a wind tunnel model, on the car’s the overall scheme of things, let’s first take a brief look at some of the Crowhurst as ‘a good workhorse’ and by Scanivalve themselves as ‘intended and which is also used in on-car applications, too. ‘It’s a fantastic surfaces complimentary products and their for most laboratory, educational or product,’ says Crowhurst, ‘and the FORMULA 1 · WINTER 2018 www.racecar-engineering.com 15
TECHNOLOGY – PRESSURE SENSORS Scanivalve MPS Scanivalve DSA3217 4264 pressure pressure scanner scanner, with pen for scale Figure 1: Schematic layout of tube runs in a conventional pressure scanning system Tubing lengths should be as short as possible to avoid frequency response issues The schematic in Figure 1 might not look so complicated but in reality on a wind tunnel model (or a racecar) a conventional pressure scanning set-up can look more like this latest version is a big technical so forth has therefore inevitably usual on-vehicle communication on accuracy, performance, sensitivity advance over its predecessor as well as been highly compromised by the protocols (CANBus). The scanners and noise insensitivity was great.’ being the most technically advanced, long tubing runs that have been pick up their power supply from the The P8 launch model offers a fully digital instrument available.’ intrinsically required. communication cable, and feature choice of pressure ranges; at the low However, on the strength of integral signal conditioning. end of the scale +/-20kPa range is Tube lengths feedback from customers, Evolution offered with an accuracy of 0.1per Among the criteria for obtaining Measurements’ concept for the Small and light cent full scale claimed, providing +/- accurate and repeatable pressure EvoScann P Series was to reduce Gordon says: ‘Our Evoscann P series 20Pa resolution. However, a number measurements is that tubing lengths the size (and channel count) of the is certainly the smallest and lightest of developments are underway, should be as short as possible to scanner so that they could be located pressure scanner available and it Crowhurst says: ‘The launch version avoid frequency response issues, and much closer to where the pressure can be fitted [to measure pressures] was primarily aimed at absolute should ideally all be of equal length measurements needed to be taken. in places that were previously not surface pressure measurements. to maintain consistency between The concept of a ‘distributed system’, measurable. It’s been interesting But we will shortly be releasing a channels. The diagram in Figure 1 shown in Figure 2, was to switch from that the door has been held open differential pressure version which generically illustrates this challenge. a small number of centrally located wherever we have introduced it, so will measure the difference between With tightly packaged racecars, multi-channel scanners with complex it’s clearly meeting an un-met need. atmospheric pressure and the [local] available space for pressure scanners, tubing runs to more scanners with We have presented it to the F1 teams surface pressures creating lift or associated hardware including power lower channel counts, and to locate and a few others outside F1 and downforce. This will have a choice of supplies and signal conditioners, the scanners around the car close many have either bought it or asked ranges from +/-7kPa upwards.’ not to mention bundles of tubing, is to areas of interest. for customisations, or want to have it With a similar accuracy to the clearly at a premium, and options are This vastly reduces the length when we reach the next stages in our launch model this would enable +/- severely limited within the chassis of a and complexity of the tubing runs roadmap development.’ 7Pa resolution, which would be well racecar and perhaps other areas within and associated installation, with Gordon also revealed that able to resolve the small variations the body of a scale wind tunnel model. just a single cable emerging from EvoScann sensors were run on three in surface pressures seen on many So getting reliable data from ‘outposts’ each scanner to connect with cars in free practice at the last F1 race areas of racecars. However, Crowhurst such as front wings, endplates and data acquisition systems using the of 2017, and reported that ‘feedback added that ‘sensitivity and resolution 16 www.racecar-engineering.com FORMULA 1 · WINTER 2018
TECHNOLOGY – PRESSURE SENSORS will also improve, we will likely increase resolution by a factor of 10.’ Further variants with more channels will also be available; a 16-channel version that fits the same dimensional envelope will come, and a 32-channel version that will only be slightly longer will follow that. Embedded device One of the advantages of this slim pressure scanner is that in some instances it can be inset into the surface of the device it is intended to measure. For example, one F1 Figure 2: A distributed pressure sensor system vastly reduces the length and complexity of the tubing and installation. It puts local client wanted to attach one to a scanners at the point of measurement, with the tubing installation then connecting via a single CANBus cable to the nearest node bargeboard. The slender dimensions of the EvoScann would enable it to scanner inside the small chord front ‘We can even supply the EvoScann and mount the sensor nearby where it be embedded in the surface of the wing flap elements on an F1 car. And without its carbon composite casing, can be accommodated.’ bargeboard itself and cause negligible in some instances wind tunnel teams which reduces the thickness to just One F1 insider who is familiar interference with the flows and have used additive manufacturing 4.5mm (0.18in) if the customer wanted with these new sensors says: ‘The test pressures it was there to measure. methods (3D printing) to provide not an even more compact installation,’ teams are interested in mapping ever Clearly, the small size of the device only a snug location for the pressure Gordon says. ‘But if there really is an more areas of the car with these small will enable it to be mounted inside scanner within the test part but also inaccessible location that is too tight sensors. Getting a feel for downforce, many aerodynamic components to print the tubing runs that connect even for the EvoScann P8 then it is especially at low speed, is better done to allow the collection and local to the surface pressure tapping ports. possible to run the tubing through it with pressure mapping than with the conversion of pressure data into four [suspension] pushrod load cells. electronic signals to be communicated to the data acquisition system Its slender dimensions would The pushrods have to measure car weight and take impacts, whereas enable it to be embedded in the through just a single cable, rather than pressure sensors can be scaled for the through many metres of delicate and pressures they have to measure.’ surface of the bargeboard itself vulnerable small bore plastic tubing. A further interesting feature is It is even feasible to fit the P Series that because the EvoScann P Series Pressure scanners S imply put, a pressure scanner is data points could be obtained. Nobody Corporation in 1955, and the products including connections, small bore a device that converts data from else was doing this at the time and it were quickly adopted by the wind tubing, steel tubulations and so pressure tappings, for example became a huge advantage. tunnel industry. The company forth. Subsequently the company over a car’s surfaces or from a sensor Pemberton subsequently left developed a large line of products had to adapt to – and exploit – some array, into electrical signals that can be Boeing and founded Scanivalve to support the use of the scanners, world-changing technological logged by a data acquisition system. The concept was devised by one JC Pemberton, who worked at Boeing In Seattle, Washington, in the 1950s and was endeavouring to measure pressures over aircraft surfaces using lots of liquid- filled U-tube manometers. Scanning valve Not surprisingly it was incredibly difficult to zero and stabilise all these devices and to obtain synchronous data. So Pemberton invented and developed a motorised, mechanical scanning valve that multiplexed many pressure signals into one transducer. The device was called a scanning valve because, in essence, the motor drove the transducer to sequentially scan the 48 ports connected to the valve. It did this in about 90 seconds, which vastly The original Scanivalve improved the rate at which pressure 18 www.racecar-engineering.com FORMULA 1 · WINTER 2018
than most people need’. Pressure ranges from +/-20kPa to +/-120kPa (200-1200mbar or 2.9-17.4psig) are currently available, with lower ranges set to become available. Homologation As this article was being written Evolution Measurement was notified that EvoScann had been homologated for use with the FIA standard ECU. This process included, among other things, being able to demonstrate that, as a microprocessor-equipped sensor, it was not possible to re-programme the device for ‘alternative purposes’, something it is demonstrably not possible to do with the single CANbus Figure 3: Screenshot of EvoScann GUI. Device has an onboard integral microprocessor so the output is in engineering units of pressure communication cable that provides the power to and data from the devices have an onboard integral fits most known requirements’ Gordon environmental conditions to be sensor. So prospective customers microprocessor, the output is directly says. ‘Our aim is to not only fit the handled, although for applications now have the added confidence in engineering units of pressure, as niche applications where the standard that don’t need it the outer casing can that this compact, innovative new Figure 3 illustrates. Here the pressure device is most suited but also, where be omitted. The size of the internal pressure sensor has FIA homologation scale is in mbar but this is configurable required, to customise for bespoke printed circuit board is a key factor for use on their racecars. via the GUI to the units of choice. requirements. And, for example, the in the dimensions of the sensor; This also means that it will be The design of the EvoScann price per channel is comparable to the however there is the capability to permissible in FIA-sanctioned events incorporates temperature correction Scanivalve MPS scanner [mentioned reduce the size and thickness still to run the sensors during qualifying for each pressure channel. It can be earlier] so it’s a good fit in terms of the further. The in-plane alignment of the and in the races, and not just in test configured to measure differential applications it can satisfy.’ output tubulations helps to maintain sessions or free practice. Before such pressures by setting one channel The carbon fibre outer casing, a compact installation. Scanning rates a compact sensor was available, to measure static pressure. And it combined with the resin potting up to 1kHz per channel are possible, though, this probably wasn’t even corrects for ambient pressure too. ‘It process enable a wide range of which Crowhurst says ‘is typically faster considered by the race teams. Evolution Measurement E advances. During the 1970s Scanivalve In 1982 Pemberton sold the volution Measurement The company offers sales of the developed a combination of valving and company to his sons Addison and was founded in July 2016 devices including spares supply, calibration that was applied to miniature Jim, and around this time PCs started and is located in Andover, consultancy, support, installation, silicon sensors and used computer landing on all our desks, so the Hampshire, UK. It is staffed by a team service, calibration, repair and correction of temperature errors. And company developed PC-based data of engineers highly experienced bespoke solutions, as well as now measurement of individual sensors was acquisition systems which allowed in measurement, instrumentation designing and manufacturing its now multiplexed electronically, enabling ever faster data sampling rates. and calibration. The team has own new products. much faster sampling rates. actually been in its current premises, Managing director Crowhurst Digital age in a previous guise, since 2006 says: ‘we work in various sectors, In the 1990s Ethernet-based and is now fully focussed on the niche areas especially where, as communications boosted things, and high-end fluid temperature and a small, responsive company we in the early 2000s UBS connectivity pressure measurement market. The can provide turn-key solutions for speeded everything up further, and relationship with Scanivalve in fact multiple applications in building all the while miniaturisation was goes back to 2001, via the companies design, wind engineering, aircraft continuing. Originally signal outputs that the current staff used to work at. design, automotive, wind tunnel were analogue so the voltage signals assessment and models, and of required conditioning, but now Highly evolved course in motorsport.’ The latter everything is digital and processed Evolution Measurement now handles includes MotoGP, from where interest to output in pressure units, and northern Europe-wide factory- is emerging in what is an increasingly it is possible to take thousands of level calibration, repair and service aerodynamics savvy sport. readings per second per channel from support for Scanivalve while special There are applications in cycling, increasingly compact devices. relationships also exist with Guildline too. And the company is also working Such has been the impact of (as its exclusive UK distributor for on a package for Formula Student. Scanivalve Corporation that it has precision measurement solutions) Please form an orderly queue … become the Hoover of the pressure as well as with Meggitt (as its UK Contact: Evolution Measurement Ltd scanning industry, the generic name application-specific distributors for Tel +44 (0)1264 316470, web: An advert for the first Scanivalve device most folk reach for when a need arises. dynamic pressure sensors). www.evolutionmeasurement.com FORMULA 1 · WINTER 2018 www.racecar-engineering.com 19
FORMULA 1 2018 – TESTING The reigning champion Mercedes team has taken a conservative approach to Halo, where design and implementation regulations allow some freedom 20 www.racecar-engineering.com FORMULA 1 · WINTER 2018
The Halo effect Formula 1 2018 hit the ground running in Barcelona with all new Pirelli compounds, aero and the controversial head protection Halo system By GEMMA HATTON and SAM COLLINS F ‘In testing we will ormula 1 has ushered in a host of is confirmed we’re confident that the parts we’ll changes for the 2018 season. The new bring to the car will sort out those losses.’ head protection system, known as Halo, is the most obvious from a visual It is something being worked on right up and down the pitlane with lots of airflow make sure we understand point of view, and has already attracted a lot of negative feedback from the teams. It has also sensors fitted to cars around the Halo structure and downstream of it. ‘Aerodynamically that the losses coming off Halo are where we had a significant effect on the rest of the car, speaking, Halo is certainly not penalty free in terms of weight and design thanks to a late and I think there is a challenge there to either introduction of the regulation leading in some cope with it in the first instance, let’s call cases to an all-new chassis design. With new it damage limitation, and thereafter think think they are’ tyres from Pirelli, offering teams a new challenge about opportunity and exploitation,’ Peter of working them at different circuits, and longer Prodromou, McLaren’s Chief Technical Officer ‘It has effects on the cockpit because it is local life power units for this season, teams have had for aerodynamics adds. ‘It does open up some to that opening. You have got the driver in anything but an easy preparation for the season. avenues which are possibly interesting to look there and so you’ve got to make sure you don’t The Additional Frontal Protection-Halo (AFP- at. I am sure there will be a variety of different have the negative effects there,’ Toro Rosso Halo, or just Halo) is without doubt the biggest solutions out there but the scope is quite Technical Director James Key adds. ‘You’ve visual change between the 2018 Grand Prix cars limited to the allowance around the basic got effects on the engine air intake and effects and those used in 2017. In design terms the shape, but there is opportunity.’ after that towards the back, so there are a Halo is governed by its own specific appendix to number of different things you have to think the FIA technical regulations. Everything from Aesthetic gain about. None of them are massive effects but the shape and dimensions of the device to the The rules allow a 20mm area of freedom around they all require some level of attention.’ material it is made from (titanium alloy Ti6Al4V the titanium structure, introduced partly for Fitting the Halo is no easy challenge either; Grade 5) is defined. However there is still scope aesthetic reasons but predictably these fairings not only does the Halo have to be homologated for different manufacturers to supply their are being used for aerodynamic gain, as some independently, it also has to pass crash tests as own products into the category, though each teams have added winglets and in one case part of the chassis homologation procedure. must be homologated independently at the airliner style vortex generators to their Halos. This has proved to be a major issue for teams. Cranfield Impact Centre. At the time of writing three companies had homologated Halos; CP Autosport of Germany, SS Tube Technology in England and a third company, V System, from which each team must purchase their Halos. Airflow impact As can be imagined for such a visually obvious addition to the car, the aerodynamic impact of Halo is noteworthy, and the teams are doing what they can to deal with its impact, particularly on the airflow over the whole car. ‘It has a significant downstream effect, especially round the rear wing area,’ highlights Andy Green, Technical Director of the team known as Force India at time of writing (the team name is likely to change by the first race in Australia in March). ‘It is not designed to be an aerodynamic device, so it doesn’t do us any favours in that department and it requires a lot of work to mitigate the issues that it causes. In testing we will make sure we understand that the losses coming off the halo are where we think they are from our modelling tools. If that Toro Rosso is one of several teams to try to increase aero efficiency with its Halo design, one of many to choose this option FORMULA 1 · WINTER 2018 www.racecar-engineering.com 21
FORMULA 1 2018 – TESTING The 20mm area at the top of the Halo has been exploited differently by the teams. The Haas team has adopted this toothy solution while others have mounted a wing everything that moves around – which it is designed to do – it is a bit scary.’ During the chassis homologation tests the Halo has to withstand various loads without it or the monocoque failing. The biggest load applied to the structure is 116kN from above, which has to be endured for five seconds. Longitudinal forces of 46kN and 83kN are applied from the front as well as a lateral load of 93kN from the side. For comparison, the roll structure on top of the car has to withstand 50kN laterally, 60kN longitudinally and 90kN from above. Weighty issue To survive these severe loads, the Halo itself has become quite a substantial structure, weighing by regulation 7kg (+0.05kg, -0.15kg). In addition, the monocoque has also had to increase in strength significantly to cope with these tests. This has further increased the weight of the New rubber from Pirelli is designed to help drivers and teams at ‘We always knew it was going to be a chassis by approximately 12-13kg. Whereas, the particular tracks. Pressure sensors were all the rage in Barcelona challenge so have invested time and money 2018 technical regulations have only allowed as teams completed their aero maps during pre-season testing up front to do a lot of test pieces,’ McLaren a minimum weight increase of 5kg to 733kg, Chief Technical Officer Matt Morris admits. forcing teams to save weight in other areas of ‘It takes the weight of a ‘Obviously, you don’t want to build a complete chassis but we built a few test pieces with the car. Interestingly, now at the start of a race a 2018 Formula 1 car will weigh roughly the same London bus and when dummy Halos and parts of Halos to test how the interfaces would behave and we found as a non-hybrid LMP1 in qualifying trim. ‘From a design perspective, weight is a big you see that test going some issues. It was close, we didn’t breeze through and there were some heart-stopping part of it. The weight limit did go up, but not by nearly as much as the installation weight of the with that amount of load, moments with particular static tests coming halo so it put additional stress on all the other in from an oblique angle. It takes the weight parts of the car,’ Green continues. ‘We had to try it is a bit scary’ of a London bus and when you see that to optimise the weight in those areas to try and test going on with that amount of load and keep the weight limit below the minimum so 22 www.racecar-engineering.com FORMULA 1 · WINTER 2018
Short sidepod concept I n 2017 Ferrari introduced a new short sidepod concept, relocating the upper side impact structure (a single specification design shared by all teams) and moving the main cooling aperture rearward. A set of box shape aerodynamic elements forward of the duct ensure rules compliance. Ferrari took this approach for aerodynamic reasons rather than those of cooling. In 2018, half the grid featured the same solution, but not all teams agree that it is the right route, with Mercedes, Renault, Force India and others all opting against adopting the concept. Conservative approach Sidepod design seems to be led by Ferrari, with impact structure relocation for efficient aerodynamic effect ‘Everything you do in aerodynamics has an opportunity cost; there is much more opportunity to make the car worse than better,’ claims Mercedes Technical Director, James Allison. ‘If you want to pursue a new and different concept, you will expect to find a fair amount of loss before you get back into positive territory. We looked at that concept and felt it would spend too much time being in negative territory before it would perhaps offer any gain at all. If you are a [team] that is a long way down the grid the situation is different it is worth taking that gamble, as you have less to lose and you know that the path you are on is not right.’ It is likely that the relocation of the side impact structure would require a substantial change to the monocoque design, while getting adequate airflow into the cooling system with such a complex arrangement of aerodynamic elements around the leading edge of the sidepod duct is also likely to be a major challenge. Mercedes has not adopted this same approach, believing that too much time would be lost in development that we can run ballast because the other area that we have to bear in mind is we have to hit a weight distribution target as well.’ Although it was originally introduced as a temporary measure to help Pirelli develop tyres when it became the sole tyre supplier in 2011, the technical regulations still limit every car in terms of weight distribution, with just a 7kg window of freedom. This means that while some teams may be able to build a car under the minimum weight, they cannot get it fully within the distribution window. Halo kitty ‘You only have a very small window of weight distribution so the actual architecture of the car needs to be correct to start with, otherwise you’re adding ballast to a car that doesn’t need ballast just to get the weight distribution right,’ Green says. ‘We would have loved to have added a huge safety margin to the whole The loss of the T-wing is not total; some teams are trying to recover some of the effect with lower mounted winglets design so that we would happily sail through the crash and load tests without any issues but that wasn’t possible because the weight limit of rear end of its car as a result, abandoning its cast titanium gearbox casing (something it ‘You have a small window the car didn’t go up enough. We couldn’t afford to increase the base weight of the car more has evolved over many seasons) in favour of a composite transmission. of weight distribution so the architecture of the car than a few kg because we knew we only had a While the price of the Halo itself is relatively few kg that wecould take out of the car. It was, modest, the cost of developing a chassis to fit it needs to be correct’ structurally, incredibly challenging.’ is higher than some of the smaller teams would This weight challenge has seen at least like. This cost was worsened by the late decision one team, Renault, substantially rework the to adopt the Halo as the 2018 AFP solution, with FORMULA 1 · WINTER 2018 www.racecar-engineering.com 23
FORMULA 1 2018 – TESTING With only a 7kg weight distribution, teams have struggled to get the weight down and remain in the window; Renault adopted a composite gearbox casing to reduce weight teams only informed of this final decision in as it is much lighter than Halo with lower The significant aerodynamic changes of September, 2017 after a long discussion process. requirements on the chassis structure. the 2017 regulations resulted in an increase in ‘Expense-wise it’s huge because we had to The weight increase as a result of the Halo loads of over 20%, demanding the tyres to be do a new chassis. We wouldn’t have anticipated also places an additional demand on the four extremely robust, leading Pirelli to ramp up the doing a new chassis this year given the number power unit suppliers, which have also had to stiffness of their entire compound range. Pirelli of changes we made last year. For a team like increase the life of their power units. Teams can also had to develop tyres with little knowledge us we would look to try and get two years out now only use three combustion engines (ICE), of the potential performance that the teams of the chassis if possible. So in that respect three MGU-H’s and three turbochargers (TC) could achieve in 2017. Despite 12,000km of it cost us a huge amount to redevelop and during the season, compared to four last year. testing, the 2014 adapted ‘mule’ cars that Pirelli redesign the new chassis. It is in the hundreds That’s 2,100km of racing mileage not including used to develop the 2017 compounds only of thousands, if not million dollar mark, to put practice sessions or qualifying. Whereas the achieved a 10% increase in downforce and the Halo on the car for us,’ says Green. energy stores (ES), control electronics (CE) and therefore the results were unrepresentative MGU-K’s are all limited to two per season, or and inconclusive. To cope with this, Pirelli went Screening process 3,150km of racing. This demand for increased for a conservative approach last year, and The Halo has had a largely negative reception reliability will no doubt have forced the having tried and tested their designs for an from drivers, teams, the media and fans. This suppliers to manufacture more robust units, yet entire season, the 2018 range is a slightly more has lead to work continuing on alternative they have had to minimise weight to help teams aggressive evolution of 2017. additional frontal protection systems. In comply with the minimum weight regulations 2017 a brief test run was conducted with which have been challenging to achieve with Compounding issues a clear windscreen fitted to a Ferrari, but the consequences of Halo. It remains to be seen ‘The 2018 compounds are from the same family while this solution solved the frontal impact how successful they have been. of compounds as 2017,’ explains Mario Isola, requirements, the driver complained of Sporting Director of Pirelli. ‘The reason why visual distortion. However, Indycar is now Tyre dilemma degradation was so low last year was because experimenting with a similar aeroscreen The other major changes for this year come these compounds have less surface overheating solution (see p16). Teams prefer the windscreen from the tyres. To encourage overtaking and pit and in general behave in a different way. In option not only for aesthetic reasons but also stops, Pirelli have added two more colours, and particular we used the 2017 Soft as a baseline therefore compounds, to their tyre compound [for 2018] because last year the Soft had a ‘We used the 2017 Soft as rainbow, the Superhard and the Hypersoft, as well as making the entire range a step softer, wider working range compared to the other compounds. Last year’s Soft is now the Medium.’ a baseline because last and introducing new allocation rules. The Superhard is now the hardest compound, From there, the 2017 Soft ‘baseline’ was then developed and used to create this year’s year the Soft had a wider adopting the conventional orange colour of softer compounds (Soft, Supersoft, Ultrasoft the Hard, which has now become the light and Hypersoft), each decreasing in stiffness in blue, and the Hypersoft is the softest compound relatively consecutive steps. Although Pirelli, working range compared and is light pink in colour. However, to gain a along with some drivers, have commented full understanding of these additional that the softer compounds of the 2018 range, with the other compounds’ compounds we need to reflect on 2017. tested at Abu Dhabi last year were ‘much closer 24 www.racecar-engineering.com FORMULA 1 · WINTER 2018
You can also read