Engineering is not normally a field that produces fine writers, as I’m sure anyone who’s had to slog through engineering reports can attest. Henry Petroski, however, is a fine exception. He’s been writing great books since the eighties, and has been a regular columnist on engineering matters for much longer. I just finished reading his book: To Engineer is Human: The Role of Failure in Successful Design. Originally published in 1982, and updated with a brief forward about the Challenger disaster in 1986, it explains the role of failure in improving engineering design. His writing is not only concise and accurate, but is really quite literate. He has some more recent works, one of which I’ve read, Remaking The World: Adventures in Engineering that was also completely readable. I hesitate to use the world “enthralling” on these two books only because engineering, like a good martini, is best served a bit dry. Nevertheless, I continued to turn the pages and was never disappointed by what I found on them.
But, this blog is not about book reviews, so onward (at this point you may be scratching your head thinking, “There’s something this blog is not about?”). The aspect I found most interesting in “To Engineer is Human” was Petroski’s general view of the positive nature of catastrophic failure of engineered structures. He consistently repeats the mantra that, while certainly the objective, failure-free designs teach us nothing about engineering other than how to make exact replicas that are safe. Failure, on the other hand, teaches us what went wrong so that we can fix it, learn from it, and then move forward. But, if humans are ever to truly move into space, they will be dependent on engineered systems, and catastrophic failure would mean sudden death.
So, does this mean that if we are to avoid risk altogether in the construction of future space stations we must stick to the tried-and-true blueprints of earlier eras? Of course not; we will seek to build larger structures that can house more people in more of an increasingly accurate semblance of Earth-bound life. Space engineering will always be looking for ways to extend what has been done so that human life can be improved. Failures, then, must be expected to occur.
When our lives become increasingly dependent on technology, and doing so requires engineering in a realm seldom before attempted, inevitable failure must be expected and designed for. Instead of looking for ways to completely obviate failure, although that is certainly the goal, space engineering will require failure tolerant systems in a way that we simply do not have great experience with here on Earth.
Redundancy, the traditional term for this, does not describe what I mean either, for that indicates simply layered backup systems. Instead, I mean designing for the failure of each part rather than simply choosing a part that is not expected to fail. For our expectations and probabilities of failure rely on having accurately predicted the conditions to which that part will be subjected. Inevitably, we will rate a part safe that is not so. Or, it may be subjected to loads from completely unanticipated events. Thus, designing structures and systems to fail, and then continue working (or at least to save the lives of the inhabitants or users), will be of prime concern. For if we are to move off of this planet, our engineering becomes no longer a tool for accomplishing tasks, but the very means of our survival.
