Title: Design Paradigms: Case Histories of Error and Judgement in Engineering
Author: Henry Petroski
ISBN: 0-521-466649-0
Reviewed by: Richard Chambers
Design Paradigms is an interesting book under 200 pages though the writing style at times lends to a touch of incomprehension in places. The author wants to make sure that the reader understands the point of design with tunnel vision leads to design problems so he tends to belabor the issue at every opportunity.
Though the book should have had some further editing, the examples from engineering both ancient and modern are good ones. Just scan through the conclusion at the end of the chapters and continue on. The book also has some fine photographs and diagrams that provide good illustration to the text. The only mathematics is a couple of linear equations during a discussion of where Galileo went wrong in some assumptions about material failure when he wrote the first materials engineering book in the 1600's.
The author shows through a review of civil engineering (mostly the construction of bridges in England and the United States though the first few examples come from ancient Greece) case studies the impact of pushing the design envelope when the engineer attempts to build on previous successes without reviewing the differences between the new project and the previous success. For instance, the Tacoma Narrows Bridge built in 1940 lasted only a few months before breaking apart during high winds. The bridge, a suspension type bridge, while perfectly capable of handling heavy loads on the roadbed, was too light and flexible during high, variable winds due to the span length and the lack of cross bracing in the roadbed. This inherent design problem was ignored by the designers who didn't consider the effect of high, variable winds on the bridge. The book contains a couple of photographs of the bridge during its death throes. I would have never believed that a bridge could be that flexible as it looks like string vibrating.
There is also an amusing story about how an engineer named Diognetus saved the city of Rhodes during a siege as well as a discussion about storing and moving columns in ancient Rome.
The author believes that case studies of design failures should be a standard part of any engineering curriculum because failure allows the student to learn something new while success merely affirms that something is possible. The difference is that through the study of failure, similar failures can be averted. A successful project may be successful despite flaws that under other circumstances would have caused failure. Petroski points out that all failed bridges that he examines were derivatives of successful bridges. The failures occurred because a variable that didn't impact the successful bridges caused a failure because the variable leading to failure became important enough to be a failure mode for the bridge.
An example would be an earlier suspension bridge that due to the span length was less susceptible to wind flexing than the longer span of a new suspension bridge. If an engineer indulges in the fatal error of believing that the new bridge is just a linear extrapolation of the earlier bridge (the same bridge just longer) and designs around the increased span support without considering the effects of wind on the bridge, the new bridge, if the span causes the wind effect variable to become more important than static load, can fail.
Petroski also looks at the Britannia Tubular Bride, a railroad bridge built in England in the early 1800's, as an example of a bridge that met all of its requirements for robustness and stability yet was not successful with train riders. The design, a cast iron tube large enough for a train to run through, caused high internal temperatures due to solar gain (estimated as approximately 120F) and the soot and fumes from the train engine were also a problem since the only openings for the soot and fumes to escape were at each end of the 1000 feet of tube. The design also lead to high construction costs.
All in all, an interesting book for someone with an interest in mechanical design. The major point of the book concerning system reviews that back away from sub-optimizations so as to optimize the entire system are applicable to any engineering activity including software.