Just reading Professor Petroski's post reminded me of watching those heart-wrenching images of oil gushing into the gulf and I'm glad it did. Truth is, once disasters like the BP oil spill or Japan's Fukushima are behind us (or at least out of sight in the media), the general public tends to forget and move on, which lets the corporate conglomerates get away with the human failure that Petroski's describes--the finger pointing and internal jockeying for where to place blame. Seems to me that dollars could have been well spent solving the mechanical problem--that is, redesigning or reengineering the blow-out preventor to operate more effectively no matter that it was a complex piece of machinery. Probably would have been far less painful to the bottom line then the PR and environmental recovery effort that befell them after the disaster.
Excellent analysis, and the Challenger example spotlights the psychological aspect of the "normalization of deviance" culture which works its way into the engineering mindset in situations where the failure rate has previously been so low that it's easy(easier) to coerce the engineers responsible for ensuring safety that things have been OK for so long, why should this time be any different. In any life situation, there's pressure to conform to the group, and that's exploited in situations such as those described here. That's why when the disastrous consequences come, they seem to be outliers, but in reality they're not and are to be expected.
It is interesting to draw parallels between the Space Shuttle and oil drilling. While deep water drilling is much more complex than most other drilling, the Shuttle is something altogether different and more complex. In the early days of rocket development, there were many failures. Then, expendables became very reliable, although there are still occasional failures. The thing that differentiates the Shuttle Program is that it invoives manned flight and that it was an attempt to present space flight as a routine, repeatable activity like airline travel. It most decidely is not. Between the high cost and high visibility of the program, failures are magnified. We accept far more danger when we drvie a car.
More people died in the Deep Horizon accident than in the Challenger accident. In addition, there was significant environmental damage in the oil rig disaster than in the Shuttle accident.
Another excellent article by Professor Petroski. In a couple of other recent threads on this site there has been some discussion of groupthink, and the kind of treatment which engineers who challenge it can expect.
When I worked in quality, I often encountered the argument, "We've accepted this out-of-spec condition before and everything worked out ok, so we might as well accept it now." My response was always, "If you're playing Russian roulette and you pull the trigger and no bullet comes out, does that mean no bullet will come out the next time you pull the trigger?"
Excellent point, Dave. I should note that I spoke with Roger Boisjoly after the Challenger disaster. (He was the one engineer who resisted going ahead with the launch, and lost his job as a result.) I also attended the first Washington, D.C. hearing of the Rogers Committee. That's the group where the late physicist Richard Feynman famously dipped an o-ring in ice water to show how brittle it became. I could go on; it was a fascinating experience.
Nice article. Seems to me that if the blowout perventer's actual performance included a real-world 45 percent failure rate -- even while tests indicated an 0.07 percent failure rate, this would be grounds to call a foul and look into whether the blowout preventer system was adequate protection against catastrophe. Is this an example of regulators asleep at the wheel?
Thanks for a great article. I agree with Rob, you'd think that it's the scarier real-world numbers that would be paid attention to, not what is supposedly the norm based on a few tests.
But the numbers also need to be related to actual people and actual harm, not thought about abstractly. If the statistical likelihood of something occurring is greater than zero and that occurrence has fatal results, then that risk is too high. For example, I once took a prescription medical for allergies that started getting bad press for fatal heart attacks. When discussing this with my doctor he said "but the risk is only 2%." Uh, right, but what if I'm in that 2%? No thanks.
The Columbia 'accident' may have been preventable; I think it was the book "Comm Check". Several engineers' / groups' concerns, if acted on, could have detected the damage.
The Challenger ' incident' was preventable. I think that was the book "The Challenger Launch Decision". The Shuttle operational limits were something like 40F to 99F. So when ice was observed on the vehicle, the engineers' recommendations against launch were well founded.
Before that was Apollo 1, when engineers argued against a 100% oxygen test, on top of many poor design features.
In each case, the advice of the engineers (experts) was ignored or over-ruled. I had much more respect for NASA before reading these books.
In an age of globalization and rapid changes through scientific progress, two of our societies' (and economies') main concerns are to satisfy the needs and wishes of the individual and to save precious resources. Cloud computing caters to both of these.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.