Race from CAD to reality

Stewart-Ford’s grand prix challenger was designed from start to finish on computer screens. David Fowler reports on fast-track teamwork

Designing a grand prix car from scratch in nine months is no mean feat. Doubly impressive perhaps is the fact that the Stewart-Ford team of 18 engineers and designers was created at the same time as deciding on the most appropriate technology to use.

Starting with a clean sheet had its advantages. One was deciding conciously not to carry over traditional practices which would be `inappropriate’. One of these was producing drawings, says technical director Alan Jenkins, formerly of the Formula One Footwork team.

The whole SF-1 car was designed on screen using CAD – drawings were produced only when absolutely necessary, says Stewart Grand Prix IT systems manager Steve Nevey. And the team set up an almost ideal concurrent engineering approach.

Central to this were technology partners EDS Unigraphics and Hewlett Packard, both introduced through their relationships with former Formula One champion Jackie Stewart, who founded the team with his son Paul.

Stewart’s approach was to bring in partners keen to get involved in the work, with the payoff of creating possible spinoff benefits for both parties.

HP, for example, supplied the team’s Unix workstations. Meanwhile HP Laboratories’ precision measurement specialists in California are investigating ways of measuring parameters the team want to monitor but cannot do so in arduous race conditions.

EDS Unigraphics’ software, in particular its Assembly Navigation Tool, allowed data to be stored in a single database shared by all the designers.

`A racing car is defined mainly by its suspension and aerodynamics. Then it’s mainly a packaging exercise,’ says Nevey. The assembly module had several key advantages: it prevented incompatible changes being made by different people, while encouraging the sharing of information.

Nevey describes the design strategy: `We tried to define the assembly sequence early on. We started with a logical sequence and then populated it with components.’

Some dimensions, such as wing height, are set by regulations. A preferred wheelbase was decided at an early stage. The dimensions of the Ford V-10 Vetec engine were already established. From such information a crude model of the chassis could be created, with spaces blocked out roughly for components. `That way everyone can see where the cooling system is going, for example,’ says Nevey. Then any potential clashes become apparent immediately, and the car develops `like a picture coming into focus’.

The ability to check for clashes in the CAD model allowed the team to work to high tolerances. Some of the clearances between components and the carbon fibre composite body tub are as little as 60 thou.

Mechanical Dynamics’ Adams kinematic package was used to run animations of the suspension and check for conflicts. This made it possible to design to close tolerances with as little as 40 thou between the wheel hub and the wishbone on full lock.

`This would be impossible on a drawing board,’ says Nevey. `In CAD you just put the wheel on full lock and optimise the suspension as close as you can make it. If it fits in computer aided design, it will fit in reality.’

Before the days of CAD, engineers would not be surprised if, when the car was assembled, a wheel hit the suspension wishbone on full lock, or the engine cover would not fit.

Stress analysis was performed using MSC’s Nastran finite element analysis software and Laminate Modeller for modelling composite materials. To reduce the overall weight, computer-aided engineering analyst Iain Bomphray worked on stressing components not normally thought of as structural.

Aerodynamic design was supervised by Dr Eghbal Hamidy, recruited from the rival Williams team. A combination of wind tunnel testing in the Swift wind tunnel in California, plus computational fluid dynamics by Ford engineers in Dearborn, was used. There was a network connection from the Stewart-Ford headquarters in Milton Keynes to allow the team’s modelmakers to download modifications from the CAD system and incorporate them into the half-size model.

Another innovation was how the shape of the car was created. This is traditionally done in a similar way to the hull of a ship. A pattern maker creates the shape by interpolation between a series of two-dimensional cross sections. For the SF-1, the shape was built up from simple planes and conic sections. Flat surfaces were created around the engine, radiator intakes and other critical areas, such as where dimensions were dictated by regulations; then the CAD package was used to blend them together.

This produced a cleaner shape and if a modification was made, such as adding a larger radiator, the relevant surface could simply be moved and the blending procedure repeated.

Network of suppliers

Developing the car digitally meant it was important that major suppliers were set up to receive files. Apart from the composite bodyshell, most manufacturing was contracted out. ISDN links were set up to suppliers such as MGA, which machined the metal patterns for the moulds for the carbon fibre parts, and gear maker Xtrac.

The team’s way of working put unusual demands on suppliers, but Stewart-Ford is building up a network of firms it can rely on. Typically, when a machined component was needed, the subcontractor received a preliminary file before the part was finalised, allowing a start to be made on tooling up. Even when the final design was ready, it could still be modified part-way through a batch. This, combined with the fine tolerances needed and small batch sizes, made finding suitable partners difficult. The team is now installing its own CNC machining capability.

Nevey says the result of all this CAD activity – the SF-1 car – is not a radical departure from Formula One practice, but he believes it will be competitive when the season starts at Melbourne, Australia in March.

Testing began last week. Until then the SF-1 had only been assembled for its press launch and driven once, by driver Rubens Barrichello, who took it up to 180mph on a straight test track.

Whatever the success of the SF-1 on the grand prix circuit, the team’s biggest achievement could be to show how concurrent engineering can work in the right circumstances.

Comparisons with large car makers such as Ford are valid, says Nevey. `We have very similar development processes to a normal car. Because a race team is quite small we all know what each other is doing. In a larger company, designers may be geographically separated, so communications between the different departments have to be set up consciously.

`We form a compact model of a typical manufacturing process and show how the techniques of concurrent engineering work.’