To Engineer Is Human: The Role of Failure in Successful Design

by H. Petrosky

If you like the notes, go ahead and buy the book!

Main idea

Failures in engineering are more important knowledge than success stories.

Overview

• There are always constraints: time / money / knowledge / environment
• Engineering is “something” between art, science & design

Engineering as Hypothesis

• safe structure will be one whose weakest link is never overloaded by the greatest force to which the structure is subjected.. the need of limits for outside forces/interactions.
• Yet, no matter how many examples of agreement one may collect, they do not prove the truth of the hypothesis, for it may be argued that one has not tested it in the single case where the theory may fail to agree with reality => this might be one of Taleb’s inspirations for the induction problem (Black Swan).
• right answers can be gotten through wrong reasoning - those answers might lead us into wrong theories
• Revisions of design == importance of peer reviews for creative work

Success Is Foreseeing Failure

• Early engineering was mainly trial and error
• The most important is to avoid total failures under given conditions. In architecture this is mainly to save lives.
• Structural (design) failures in single building can invalidate design of many others, no mater how long they have last.

Computers

Automated aid in given field (computers for architects) can solve many problems but at the same time create new ones, which shouln’t be overlooked:

• Computers make possible calculations so complicated that it is beyond human to catch errors. We are increasing level of complexity at the cost of error detection
• idea of “failures databases”
• Faster calculations creates space for easier optimization (minimum weight of elements), which in turn can imply less space for errors. The most economical structure wins (please note that we’re talking about CAD solutions from 80s here).
• Specialized software is designed by a few for the whole profession(s). Tools are build by structural analytics and used by structural designers. It is not easy to include real experience into process of making software.
• Problem with data generated by computers: Because the computer analysis is available it will be used. Because the answers are so precise there is a tendency to believe them implicitly.

Other insights

• With “every part as strong as the rest,” then every part would “wear out” at the same time. We expect failures of given parts of the system at different times (pace) but we can’t tell this time exactly.
• Cyclic occurrences of failures: when structural failures occur, a larger factor of safety is used in subsequent structures of a similar kind. Conversely, when groups of structures become very familiar and do not suffer unexplained failures, there is a tendency to believe that those structures are overdesigned, that is to say, they have associated with them an unnecessarily high factor of safety. Confidence mounts among designers that there is no need for such a high factor of ignorance in structures they feel they know so well, and a consensus develops among designers and code writers that the factor of safety for similar designs should in the future be lowered. The dynamics of raising the factor of safety in the wake of accidents and lowering it in the absence of accidents clearly can lead to cyclic occurrences of structural failures
• fail-safe design
• The paradox of engineering design is that successful structural concepts devolve into failures, while the colossal failures contribute to the evolution of innovative and inspiring structures
• Good judgment is usually the result of experience. And experience is frequently the result of bad judgment. But to learn from the experience of others requires those who have the experience to share the knowledge with those who follow.
• design process itself, that transcend all eras and cultures. Thus every case study— no matter how obsolete its technology or how fresh its trauma, whether in a book or in tomorrow’s news — is potentially a paradigm for understanding how human error and false reasoning can thwart the best laid plans.