Monday, 20 September 2010

KM failure stories number 6

Here's a story from New Scientist magazine, 08 July 2000, which is a classic example of a KM failure, through lost knowledge.

"How could the designers of a revolutionary bridge miss something so obvious?
ENGINEERS who designed London's wobbly Millennium Bridge admitted last week that the computer simulations they used to model its behaviour couldn't cope with the effect of people walking across it. The blunder, say observers, resulted from an out-of-date standard for testing bridges and poor dissemination of information in the industry about previous incidents.

The 320-metre footbridge over the Thames, which cost £18 million, was closed on its opening weekend last month, when it was found to wobble more than expected. Since then Arup, the engineering firm responsible for the bridge, has been analysing its response to vibrations.

The engineers used shaking machines to send vibrations through the bridge. They found that horizontal vibrations at 1 hertz (one complete cycle per second) sent the bridge into the kind of S-shaped lateral wobble that was seen when it opened.
This was a clue to the source of the problem, says Pat Dallard of Arup, structural adviser on the bridge project. "Normal walking pace is about two strides a second, so you produce a vertical force at around 2 hertz," he says. But the horizontal frequency is half that. "As we walk, one foot pushes left, then the other pushes right, so you have a 1-hertz horizontal force," he says.

To make the bridge wobble, a lot of people would have to be walking in step, says Tony Fitzpatrick, Arup's head of engineering. And when engineers checked TV news footage, this is exactly what they found. "What we saw was unintentional synchrony of walking," says Fitzpatrick. People on the bridge were adjusting their balance as the bridge moved slightly--probably because of high winds on the day, Fitzpatrick says."When what you're walking on moves, you start to steady yourself. And here, everybody is doing the same thing." Synchronised walking made the bridge wobble even more.


So why didn't Arup engineers know small movements of the bridge could cause "lock-in", as synchronous walking is called? One reason might be industry secrecy, says Bill Harvey, a civil engineer at the University of Exeter. "Engineers are very, very sensitive about these things," he says. "The only reason anybody knows about the Millennium Bridge is because it's so public. If it had been somewhere small, then I imagine it would have been kept quiet." Even when problems are aired in civil engineering journals, they rarely identify the projects under discussion.

Pavic, too, isn't surprised that Arup did not know about the effect. He says it took a Sheffield PhD student four years to track down three cases of horizontal wobbling on footbridges. "What chance have practising engineers got, when they're working on several projects at the same time?" Pavic asks. Arup has now called in Yozo Fujino, who worked on one of the previous cases in 1992.

Arup is now working on a new computer simulation that will account for lock-in--which should be cured with dampers. "I'm embarrassed for my industry," says Fitzpatrick. "But if you're not going to do new things, you might as well put a bin liner over your head and sit in the corner."
Interesting, eh?

It had happened before, but a combination of secrecy and buried knowledge meant that it happened again, in a very emberassing high-profile way, resulting in 2 years delay and costing £5 million.

No comments:

Blog Archive