Simanaitis Says

On cars, old, new and future; science & technology; vintage airplanes, computer flight simulation of them; Sherlockiana; our English language; travel; and other stuff


ENGINEERING HAS been defined as “doing a lot with a little.” But, of course, too little is not a good idea. This is the theme of To Forgive Design: Understanding Failure, its review by Philip Nobel appearing in the London Review of Books, 21 FEBRUARY 2013. The magazine recently resurfaced, and the book has joined my to-read list. Maybe you’d be interested too?


To Forgive Design: Understanding Failure, by Henry Petroski, Harvard University Press, 2012. Both and list it.

Reviewer Nobel cites his own architectural professor, Peter Galdi, for several tales. One is the story of the Manhattan Bridge, crossing the East River to connect lower Manhattan and Brooklyn.


The Manhattan Bridge, under construction, March 1909.

The Manhattan Bridge is a suspension design, one in which the “dead loads” of the structure itself and the “transient loads” of its traffic are fed through a network of steel cables to its high towers and anchorages on each bank. Though it has been standing for more than 100 years, the exact paths of these loads aren’t known—the necessary engineering mathematics didn’t exist when it was designed.

This bridge was engineered by Leon Moisseiff, another of whose designs was decidedly less successful: the Tacoma Narrows Bridge. Its failure in November 1940, just five months after its completion, was traced to a destructive vibrational mode brought about by 40-mph winds. This is one of the more documented engineering failures of all time (

Another reason for failure cited in the London Review of Books article was rather less complex, though no less destructive. Lockheed Martin was the prime contractor with NASA’s Mars Climate Orbiter in 1999.


The Mars Climate Orbiter suffered from a trivial—but costly—mistake. Image from NASA.

After a 10-month journey, the craft burned up entering the Martian atmosphere. The problem: NASA had described operational data in SI (Système International metric) units; Lockheed Martin had used English units.

Author Petroski lists several challenges confronting engineers. Among them are marketing (for example, the highly touted, overly quick—and deadly—luge track at the 2010 Vancouver Winter Olympics). Tight time frames can be another (one leading to an incorrect epoxy in Boston’s Big Dig 2006 ceiling failure). And, of course, there are plain old unscrupulous suppliers.


The Brooklyn Bridge, its opening celebrated in 1883. Fortunately, John Roebling designed it with a sufficient safety factor. Image from the Brooklyn Museum Collection.

John Roebling engineered the Brooklyn Bridge with a six-fold safety factor. Years later, after Roebling’s death, his son discovered that substandard wire had been substituted in the cables. Analyses proved this only diminished the safety factor to five-fold.

I’m reminded of another talented engineer, Colin Chapman, and his purported view on safety factors. Chapman would execute a new race car design, then have his drivers test it.


Innes Ireland winning the 1961 U.S. Grand Prix in a Lotus 18 (the first GP win for the team). Image from Bonhams.

If the race car worked okay, it’s said he’d remove something. If it broke, he’d put the piece back. If it didn’t break, he’d remove something else…. ds

© Dennis Simanaitis,, 2013

One comment on “FAILED DESIGNS

  1. Andrew Johnson
    June 30, 2013

    Structural design is pretty much based on failure. Basically designs were based on applying a factor of safety to material yield stresses to arrive at an allowable stress and comparing the computed stresses in a structure to this allowable. Now most design is based on limit states where we use the yield stress with a small reduction factor to account for material variation and fabrication/construction quality, and then multiply the applied loads with load factors. Dead loads (because they are generally well defined) get smaller load factors than live loads like occupancy loads, traffic loads, wind and snow, which are based on statistical analysis. Interestingly seismic loads get a load factor of one, because the design earthquake as defined in the codes is already an extreme event.

    Structures when they hit the yield point of course generally don’t just collapse, but technically they are considered to have failed because of moving into the yield zone and would need to be remediated or demolished. Shear failures are much more catastrophic than flexural (bending) yielding, because they don’t show the telltale signs of sagging that a yielding beam will show, and can actually be quite sudden, so the material capacity reduction factor for shear loads is smaller.

    Coming back to the concept of failure, the failure that most design is based on comes from lab testing, but also gains from the real life laboratory, actual failures around the world. Codes and design standards are constantly evolving as a result of real failures as well as the results of Phd Theses. And in fact structural behaviour and failure mechanisms are a good deal more complex than the relatively simple explanation that I have just given.

    While on a day-to-day basis the actual factor of safety can be quite large, mainly due to the fact that structures rarely see their design load cases, at the design load case the effective factor of safety for most structures is about 1.67, much lower than most people assume (including most of the contractors and clients that I have to deal with).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: