Safety by numbers

A hot new piece of software is set to revolutionise the aircraft building industry.

Investigation reports into the spate of plane crashes over the last two years have cast a question mark over how truly safe the skies are. But a new piece of software developed by a team from the University of Wales, Aberystwyth, could bring a sigh of relief to troubled airline passengers by helping consign to history crashes caused by equipment failure.

It has caused such a ripple of excitement among plane manufacturers that it is also being investigated by Nasa, which sees it as a way of preventing another space shuttle disaster.

AutoSteve carries out what is known as Failure Mode Effects Analysis, a test of all the electrical circuits in a product. It predicts each possible failure for every component in the circuit and the knock-on effect of that failure. Until now, these tests have been carried out by hand, a laborious, time-consuming process which is open to human error brought about by boredom or lack of concentration. The implications are obvious for complicated pieces of technology such as aeroplanes and space shuttles, where the failure of a single circuit can have myriad repercussions which fan out throughout the structure.

But its potential was also recognised early on by car makers Ford and Jaguar, which helped finance the prototype. It has been tested by both companies for almost a year, checking circuits of their various models with remarkable results. And those results will benefit just about everyone - cars will be safer and more reliable. They will also be cheaper to make, but whether that saving will be passed on to the buyer remains to be seen.

"It will make cars safer. Absolutely. It will also make costs cheaper," explains Richard Shipman, AutoSteve systems engineer. "You can take individual components and test them and if it proves it doesn't make any difference to the safety of the car it can be cut out. And they will be more reliable because there will be fewer recalls.

"In the past, there has been a lot of lip service paid to this kind of fault analysis, but often the most basic kind of analysis slips off the engineer's desk. Their attitude, honed through experience, has been, `This is so trivial, I don't have to do it.' But AutoSteve often picks up on faults they wouldn't have expected, even with all their experience," Shipman says. "The program always looks at the most detailed level whereas an engineer might say, `My experience tells me I can gloss over this.'

"Only about two or three years ago, the complexity of the average car electrically was such that an engineer could sit and think about the fault analysis and then write it up," Shipman says. "Now the average Ford has several tons of computers in there. Lots of wiring. Engineers don't have the ability to conceptualise that, and that's where AutoSteve comes in."

It has taken nine years' hard research and development, but it seems like the time has been well spent. When the prototype was first developed, it was too slow. Its laborious checking process might have saved wear and tear on the engineer's concentration, but the time-saving was minimal. Ford and Jaguar made suggestions and the university launched a follow- up project which cleared up bugs and managed to get it moving at lightning speed, far faster than any human process.

It didn't take long for feedback. Two of the core team members, Dr Chris Price and Dr Neal Snooke, were invited to Ford's US headquarters in Dearborn, where they gave executives a crash-course in the wide use of AutoSteve. The company responded with an instant recommendation. Jaguar also gave it an unequivocal thumbs-up.

This overwhelming response prompted the research team to look into the project's wider applications. Once it became apparent it could be adapted to any hi-tech system, it was decided to launch it commercially under a new company, First Earth.

At the Reliability and Maintainability Symposium conference in Washington DC, a big gathering for the top companies' techno-whizzes, Dr Price's presentation got an enthusiastic response from a range of companies, including Siemens, Raytheon and GEC-Marconi.

"We're now so pleased with what we've got we're taking the project fully commercial," Shipman says. "We're selling to several auto-manufacturers and we're also in serious discussions with Boeing and Nasa. The aerospace industry, I must say, is very interested now. The great strength here is that it will free up engineers' time to concentrate on what I'd call the more interesting failures - the ones that have more catastrophic effects and higher risks."

One of the aspects of AutoSteve which is winning over buyers is that it not only highlights potential faults, but also provides enough back- up analysis for a redesign of any failing system thanks to a powerful simulation tool.

Shipman is enthusiastic about the software's numerous strengths. "AutoSteve also allows circuit testing much earlier in the design cycle - in fact, very early on when the designers are working out the schematics for the new car models. They can tell in advance which gauge of wire uses which resistors, say, instead of waiting five months down the line when the basic design doesn't work."

But then there's the name. AutoSteve doesn't exactly conjure up what may be a major leap forward in safety and reliability for our most worrying forms of transport, but it was christened in true techie humour early in the development process and now it's too late to change it. Two of the engineers advising the project were called Steve - Leedham from Ford and Twitchett from Jaguar - and it would automate part of their job, so ...

"It's a twee name," Shipman says, "but the initial research proposal called it Flame, which was a technical acronym. Ford, unsurprisingly, didn't like the connection with cars bursting into flame so in a moment of madness we settled for AutoSteve. But it doesn't matter what the name is. This software is a tremendous advance, and that's what's gaining the recognition."