Petroski on Engineering: The Normalization of Deviance

Last spring and summer, while oil gushed into the Gulf of Mexico, much of the news coverage following the fatal explosion on the drilling rig Deepwater Horizon focused on the blowout preventer located a mile below the surface. As its name denotes, the device's function was to prevent exactly the kind of blowout that did occur. It did not work properly because some pipe from the runaway well was forced upwards into the preventer and jammed the mechanism.

Over a 25-year period, a pre-accident survey had found blowout preventers on about 15,000 other wells had to be activated in an emergency only 11 times. Unfortunately, in five of those cases, the preventer failed, as it did in the Gulf. This 45 percent historical failure rate did not jibe with the 0.07 failure rate claimed during the government-mandated testing of blowout preventers.

Even as lax oversight and testing procedures were being called into question, the oil industry was using this low failure rate to argue for less frequent testing of the complex system of valves and rams that were the last line of defense against a blowout. It was estimated that reducing testing requirements could save oil companies almost $200 million per year.

A blowout preventer is also an expensive piece of equipment to maintain, with an estimated cost of $700 per minute incurred during the time that drilling had to be stopped while the device was disconnected, hauled to the surface, repaired, lowered back down, and reattached to the wellhead. The economics of the situation clearly argued against a conservative maintenance regimen and promoted a culture of risk-taking.

In the case of the oil company BP, whose Gulf operations were directed out of Houston, the culture that developed around deepwater drilling operations was not unlike that of another Houston-based technology. At the outset of the space shuttle program, the total-failure rate of shuttles was estimated by engineers to be 1 percent and by managers to be 0.001 percent. The Challenger accident proved the actual failure rate then to date to be 4 percent, and after the Columbia accident, it still stood at close to 2 percent. Repeated negative experiences with eroding O-rings and shedding insulation were not heeded as warnings. They were taken as signs of the robustness of the space vehicle and promoted a fault-tolerant culture that allowed for what has been called a "normalization of deviance."

Normalized deviance has also plagued the oil drilling industry, where at least some companies have allegedly let the financial bottom line dominate decision-making. Just as NASA managers were emboldened by two dozen successful shuttle flights before the accident with Challenger and, after the hiatus, another 87 successful missions before the disintegration of Columbia, so the low incidence of needing to call upon the blowout preventer in an emergency promoted a sense of bravado in the operation of offshore oil rigs.