Features - Technical
MAY 31, 2000
BY PETER WRIGHT
Reliability has become a science, practiced successfully by many industries. In the aircraft industry, where failures tend to have such catastrophic consequences, it has been understood and applied for many years. The production automobile industry has learnt how to apply it to mass production in the last 30 years. Perhaps the most remarkable application has been to space exploration, where there is only one chance, no testing under real conditions, and scant knowledge about the environmental conditions under which systems must operate. Formula1 is slightly different, as is any racing situation: the quest to go faster, and the pressure to fit development parts for a race must be tempered by their potentially detrimental effect on reliability. Ross Brawn, having learned the lessons of reliability at Le Mans with Jaguar (how often has Le Mans been won by the fastest car?), has managed to curb Ferrari's tendency to always fit the latest, "faster" parts, and impose a discipline of only racing what has been proven in testing. In doing so successfully, he has provided his drivers with a formidable reliability record over the last few years.
(Author's addendum: Just as the ink dries on these last sentences, Schumacher has suffered a rare Ferrari failure at Monaco. The heat induced breakage of the rear suspension, caused by a leaking exhaust, is just the sort of problem that jumps up and bites a team, however much testing and meticulous preparation they carry out.)
Failures can be classed as either due to design, or manufacture and assembly. All boil down to a physical property being exceeded in one or more components. Design failures are caused by an error in the estimation of the loads and environmental conditions the components are subjected to, or an error in the design to meet those requirements. Manufacturing and assembly failures cause the margins and tolerances, designed in to cater for uncertainties, to be used up. Whether the strength is inadequate for the actual loads imposed, or the number of cycles at given loads are too great (fatigue failures), or physical properties are changed by excessively high or low temperatures, or excessive wear takes place due to any of the latter, every stage of the concept, design, engineering, build, test and maintenance process has an effect on reliability. One error is all it takes to stop a Formula1 car.
Reliability issues are probably not at the forefront of the Chief Designer's mind while he is scheming out the concept of a new car. However, decisions he makes about the layout and the specifications for the designers to work to will affect reliability. The urge to improve the aerodynamic L/D figure has resulted in all components running at higher temperatures than before, in an effort to reduce cooling drag. Higher temperatures put pressure on material properties and mean that the engine cooling system runs at a higher pressure. Brakes wear faster if run hotter. Lubricants' properties deteriorate, and materials lose strength and stiffness at elevated temperatures. The arrangement of components, particularly the engine and its cooling, lubrication, fuel, and electric/electronic systems, and the transmission will have a major effect on serviceability. Ease of routine servicing and emergency replacement both have a big effect on reliability - the easier it is to access parts that must be worked on during assembly or in the Pits, the less likely is an error in tightening or locking a bolt, or in the connection of a hose or electrical connector.
The designers and detailers however, think about reliability all the time. Working to a functional specification for the component, assembly or system they are designing, they need to know the quasi-steady state and dynamic loads it will experience, the temperature range in which it will operate, rubbing speeds between moving parts, lubricant properties etc. This information comes from track testing, rig testing and simulation models, but is only as good as the data acquired. The development of sophisticated sensors and data systems, and better and better computer models has resulted in major improvements in estimating the loads, although the dynamic component which affects fatigue life, is still hard to measure. Supported by stress engineers using FEA software, designers choose materials and manufacturing processes to enable them to design the lightest possible components. Knowledge of the tolerances achievable at every stage of the manufacturing process is essential, as brochure figures do not always represent what is supplied. Raw material properties, heat treatment specifications, machining processes and surface finishes all have tolerances which are additive. The most successful Formula1 suppliers are those that have maintained consistent high quality, rather than low prices. Critical processes are more and more often being carried out in-house, especially by engine manufacturers.
Although computer controlled processes in all types of manufacturing have revolutionised the quality of manufacture and allowed tighter tolerances to be specified, the human element still has a big influence on quality. Composites, electronics, and component assembly are still highly labour intensive processes at the production quantities involved in Formula1. Clean rooms for all these processes are now standard, and CNC machines are used wherever possible: carbon-fibre pre-preg. cutting, autoclave cycle control, and soldering for example. However, laying up in the moulds, making wiring looms and assembling engine, transmission, suspension, brake, steering, electronics, hydraulics etc. have to be carried out by hand. It has been known for a long time that one cannot inspect quality into a product; it must be both designed in, and built in by the person(s) making it. Formula1 teams and suppliers employ state-of-the-art inspection techniques and equipment, often computer-controlled, but they only catch errors in what they check, and it is not possible to check everything. The majority of inspection techniques check the outside surface of a component - the dimensions, the surface finish and the surface material properties. X-rays for castings and ultrasonics for composites allow cavities and de-bonding to be identified respectively. Composites are amongst the hardest to inspect, as the properties cannot be fully tested without destroying the component. Bond integrity, in particular, depends on cleanliness of the joint surfaces, and correct mixing and curing of the adhesive - if the operator has had an argument with his wife that morning, he may not be fully concentrating on the job in hand. Several wing and suspension failures can be attributed to bond failures. The aircraft industry is working hard to develop new technologies for non-destructive testing of composites, but the allowable working stresses for composites in aircraft are still much lower than those used in racing cars, thereby maintaining safety factors throughout the 30+year life of the aircraft, in spite of deterioration in properties due to moisture absorption or invisible damage.
Virtually all assembly of sub-systems is carried out by hand, and so is even more dependent on human factors. A philosophy of self-inspection against checklists is the normal practice, as inspection is only possible in many cases by disassembly. Recording torque-wrenches and other techniques for maintaining a build record are being employed in critical areas. Some sub-systems can be tested e.g. running an engine on a test bed or burning-in a new electronics assembly, but most cannot.
It has been said, somewhat tongue-in-cheek, that an optimally designed racing car would win a race and every component would fail simultaneously as it crossed the line. The engine would suffer a massive blow up, as pistons, rods, valves and crank all failed, the wheels would fall off, the brakes would be completely worn out and every other part would suffer a fatigue failure! Not a very practical or affordable way of going racing! However, it does highlight the critical aspect of a reliability policy. Because very few parts on a racing car are stressed to have an infinite life, the actual life of assemblies under racing conditions must be established i.e. for how long the component with the shortest life in an assembly will run. A replacement schedule for assemblies can then be drawn up, based on this information. Teams have computerised databases that record the build and running history of every part and assembly, based on their predicted safe life. A very large part of track testing and engine dynamometer testing is allocated to establishing that the life of components is as predicted. "Completing a race distance", as quoted so often towards the end of pre-season testing, is establishing that the life of the whole package is adequate.
The team that has so effectively applied this philosophy is Ferrari. Benefiting enormously from having their own test track at Fiorano, they are able to roll out a car fitted with a development engine or other system, straight onto the track, and their test driver will put the required mileage on it at representative speeds. They must have the track fully calibrated by now, and know exactly how many laps driven by Badoer represents a race in Schumacher's hands. Ross Brawn has imposed the discipline of not racing a development part, however promising its performance, until it has a successful race distance proof test has been completed. The days of Ferrari fitting the latest development set of pistons to their engines, at the track, are past. This approach has paid off well, allowing Schumacher to challenge for the Championship in a slightly slower car, by accumulating points when the McLarens run into reliability problems.
Ignoring a failure and dismissing it as a one off incident, is rash. Every failure has to be correctly diagnosed for the cause - a sometimes difficult task when confronted by a puddle of oil, some scrap metal and a lot of metal swarf that has accumulated in the bottom of an engine or gearbox. Only then can a cure be established and proof tested. A senior Boeing engineer once said: "There is no such thing as an isolated failure; it is always the start of an epidemic". Some reliability problems are fundamental to the concept e.g. Jaguar/Cosworth's decision to share the lubrication oil between engine and gearbox. The recurring problem significantly hindered their pre-season development, as all the quick fixes were not totally successful and it required a fundamental re-design to cure the problem.
The final link in the reliability chain is servicing and maintenance - the responsibility of the race engineers and mechanics. While it used to be left up to individual mechanics to build cars how they thought best, there is not much that is not specified closely by the design department nowadays. However the tiniest part, such as a tie wrap, can lose a race. The technical issues are no different from those governing the reliability of the most complex part: if the temperature exceeds the properties of the plastic from which it is made, the hydraulic hose it is locating will come loose and touch the hot exhaust, inevitably bursting. Brake bleed nipples, as Schumacher found out when he crashed at Silverstone in 1999, can come loose if they are unable to withstand the vibration of the latest lightweight, high-revving engines. In this case it was not the fault of a mechanic, but rather that the vibration spectrum had changed, overcoming a proven method of locking the nipple. Such are the pitfalls of progress in engineering.
Hydraulics and electrics/electronics can be made extremely reliable, but they do not like being disassembled. The chances of a problem at a hose connection or multi-pin connector are greatly increased following their being disconnected and reconnected. Formula 1 cars are stripped down after every 350km or thereabouts. The attention to detail and meticulous checking that are needed to ensure that they run correctly and reliably for the next 350km, all too often carried out at night by exhausted mechanics, requires an organised approach, cool headedness under pressure and real dedication. I worked with Nigel Stepney when he was Ayrton Senna's mechanic at Lotus, and it was an eye opener to observe him at work. His car was always prepared in advance of the others, and Nigel spent the time gained checking, checking, checking. If the car broke down in the race, Nigel became a bear with a sore tooth for days afterwards and made the designers' lives hell until he considered a solution to the problem had been found. It is no wonder that Ferrari has made so much progress in its reliability record with Nigel as their Chief Mechanic. The fact that he has motivated Ferrari's large pool of mechanics to think the way he does, shows how far he has come from being a mechanic on a car.
The greatest fear of any individual who has an influence on the reliability of a Formula1 car, or any racing car, is that a failure of "their" part causes an accident, particularly one that injures the driver. Wings, suspension, steering, tyres and brakes are all safety critical, but the failure of any other part can be equally dangerous. Outside suppliers manufacture some of the most critical components, and they too must operate the same reliability philosophy. Many are suppliers of similar components to the motor industry, and so the sight of their brand failing on TV does nothing for their marketing strategy.
To achieve the reliability record necessary to win a World Championship, particularly while developing the performance of the car and engine during the season, not only requires the technology and engineering procedures, but also the ethic, throughout the organisation and its partners, that a failure during a race event is not acceptable. Everyone must sign up to that and commit to the effort and time that it takes to realise a 100% finishing record. Every individual must believe in and take some responsibility for ensuring that any test or race failure, however insignificant and inconsequential, is considered important, the cause found and steps taken to ensure no repeats. To instil this ethic takes training and motivation, which can only come from the top of the engineering hierarchy, supported by the budget setters. Introducing new equipment takes only a month or so, but changing the mind-set of 200 people takes a lot longer. It requires stability, belief in the management, giving people responsibility, and discipline. Some teams have it, and some probably never will.