LEARNING FROM FAILURE

In 1357, King Charles IV had his chief architect build a bridge across the river in my native city of Prague. The bridge has survived many floods and even wars; it is unlikely to be destroyed by anything short of high explosives.

The reason for this structural strength is simple: parts of it have been torn down by floods and ice floes so often over the past 600 years that it emerged stronger after each replacement of the failed part, and after six centuries of this treatment the bridge is so strong that not even a Communist economy can do it in.

This type of engineering by trial and error was typical of the Middle Ages. It was not quite as wasteful as the design of the Egyptian pyramids, whose life expectancy is for all practical purposes infinite, but at the expense of a gigantic waste of resources. Contemporary civil engineers would produce the same result by making them hollow, and would use computers to optimize the design for minimum material and labor to achieve structural integrity, augmented by a hefty safety factor, of course.

But engineers are human and sometimes botch it as badly as their medieval collagues who had little mathematics and no computers to guide them. Such failures are the subject of a very interesting and readable book by Henry Petroski, a civil engineering professor at Duke U., To Engineer is Human: the role of failure in successful design (St. Martins's Press, 1985, 247pp., hdbd., $16.95). For example, the worst accident in US history due to structural failure took place in Kansas City in 1981, when 114 people were killed and almost 200 injured in the collapse of two crowded walkways suspended from the ceiling of the Hyatt Regency Hotel.

The original design, shown on the right, called for the end of the beam supporting the walkway to rest on a nut and washer on a single vertical rod which extended from the ceiling through the top walkway some 15 feet below it, and continued for another 30 feet to the lower walkway.

The 114 victims did not die in vain, for no architect will ever repeat such an inept design, and no construction company will ever overlook the stress on that type of "improvement." So with all our computers, we sometimes still have to learn the hard way, as did the medieval engineers.

With hindsight, such mistakes should never have happened; but they do and they will, and at least we can learn from them. Would anyone stop, not the errors, but our learning their lessons?

Read on.