To Engineer Is Human

I finished up To Engineer Is Human a few weeks ago, but have been considering how to blog about it in the back of my head. It is a book from 1982 that is supposed to be about engineering in generally, but is almost entirely focused on structural engineering (and at the very end has a short chapter about using computer aided design). While it does have some general engineering lessons, it really isn't very relevant to pretty much anyone I know reading this blog. So while I'm tempted to dig into a bunch of examples in the book I'm going to try to just skim some highlights.

First off is he mentions The Deacon's Masterpiece which is one of the few engineering poems I've ever read. The poem itself is rather long, but the idea is that someone decides that for any wagon there's always one thing that breaks first. So he's going to design each piece of the wagon so it won't be the first to go. And the thing works great for years and years and eventually every part fails at once and the whole thing falls apart. It's pointing out that in any design, no matter how good there has to always be a weak point.

One of the main themes of the book is that the reliability of structures kind of goes in cycles. Once a type of bridge has been design properly with a certain set of materials so it doesn't fall one of a few things happen:
1. The same design is used in a different set of conditions and fails. Then the engineers figure out what is required for that set of conditions.
2. Lessons of the past are forgotten and repeated. Once the new bridge fails the lessons are remembered and the next set of bridges are good again.
3. Since the current bridges are good it is time for progress and new designs or new materials or the same idea but with less safety margin are used. Then if there are failures the issues are worked out and then some better reliable bridges are built.

Along similar lines he suggests that every design is a hypothesis. The hypothesis is the design will work without failure. When the hypothesis fails it can be very unfortunate, but it is also the most informative. When the structure doesn't fail it doesn't say much because who knows what slight change in conditions would cause it to fail.

The two themes above suggest the problem with using structural engineering to draw general lessons for all engineering. In many areas of engineering failure is not as catastrophic and as long as failure rates are within acceptable limits meeting performance requirements can be far more important.

A lot of the book is going through various structural failures and talking about why they happened. One of particular interest to me is the Kansas City Hyatt Walkway Failure. Less than a month after I was born the Kansas City Hyatt was hosting a large party, which some of my family members almost attended, and the walkways over the lobby where the party was held collapsed and killed 114 people. Some of the initial thoughts were that the walkways simply weren't designed for people dancing on them. What actually happened is that the original design used beams that stretched from the ceiling and each beam supported multiple levels of walkways. But getting bolts to the middle of a long beam was practically very difficult so someone modified the design to have one beam from ceiling to the top level and then another beam from the top walkway to the bottom walkway. The problem is that now the bolt holding the top walk way had to support the weight of two levels instead of just one. Besides the personal relevance of the story I think this is interesting because it is tempting to blame whoever made the change at the last minute. But it is also the fault of the original designer for creating a design that couldn't actually be built or as someone else put it:

A detail that begs a change cannot be completely without blame when the change is made.

Few other random items:
-Kind of cool to see a book from 1982 mention a congressman working on improving safety regulations and it's Al Gore.
-Excalibur and other sword myths have a lot to do with engineering (why when a bunch of swords are built one is especially reliable). So is Icarus (wings melting as he tried to fly to the sun) - this one is really about a good design being used improperly (for the SGU fans - get it, the Icarus planet, get it? - oh and for the SG1/SGA fans, Daedalus is the guy that built the wings). And the liberty bell is an engineering failure.
-In this book written in 1982:

Computer models that predict the behavior of the economy have come increasingly to be relied upon to justify major economic decisions, and yet these models are not necessarily any more infallible than the ones that predict the fatigue life of a bus frame.

-This one is actually an important lesson in many aspects of life - random failures can be mitigated by having two of something, but to backup against systematic failure the backup must be of a different design (a friend has a story about a bunch of people backpacking who all brought water pumps so they'd be ok if a few broke, but then the temperature dropped and all of the pumps froze - luckily some of them had iodine tablets).

Oh crap, so much for not writing too much. Sorry.