The Limits of Simulator Training
Back when I was editing IFR magazine, I got myself invited for a two-day sea tour on the U.S.S. Roosevelt, a Nimitz class carrier then doing workups to return to the Persian Gulf. After the fact, the Navy invited me to fly an F-14 simulator at NAS Oceana, for a follow-up story on how landing signal officers are trained.
The plan was to take off from Oceana at night and go land the sim on a simulated carrier steaming offshore. Once I was strapped in, the sim op gave me a takeoff clearance and a heading, after which I advanced the throttles and promptly ran the airplane off the runway into a ditch. This elicited gales of laughter from the control booth. "Sorry," came the explanation over my headset, "we set a 100-knot crosswind for all first takeoffs. It's kind of a tradition."
Maybe the Navy's lucky that its sims will do that kind of silliness. Evidently, airline simulators don't—or at least they didn't used to—and that shortcoming was cited by the NTSB as a factor in the December 20, 2008 crash of Continental Boeing 737 in Denver. The airplane careened of a runway and caught fire, an accident eventually blamed on strong, gusty crosswinds that weren't reported to the pilots and simulator training that didn't address the reality of conditions that challenging.
Last week on AVweb, we quoted a USA Today report whose story hook was that shortcomings in simulator training were linked to nearly half of the 522 airline fatalities in the U.S. during the past 10 years. Could this really be? To me, it just doesn't pass the smell test. Further digging into the story revealed that while the NTSB cited simulator training shortcomings as a factor in the Denver crash, it seemed to concede a certain far-outishness to the finding. It gets down to this: To what degree can you expect to train pilots in weird abnormals? Where do you draw the line and simply decide to take your chances?
The Continental crew probably encountered an unexpected 45-knot direct crosswind on their takeoff roll, an extreme condition the tower didn't advise them of. Had the pilot been properly informed, he would probably have delayed the takeoff or insisted on another runway. To put the need for training to handle those conditions in realistic perspective, the NTSB report said Continental's operations records showed that during 940,000 takeoffs, only 250 (.03 percent) encountered crosswinds of 25 knots or greater and 62 (.01 percent) had crosswinds of 30 knots. Wonder how many had more than 40 knots? (It's above the company limits, by the way.) The probability on that is less than one in 15,000. For that, you're going to re-jigger a sim and introduce new training?
And how about the poor homo the sap who has to digest, process and store response and muscle memory for a one in 15,000 event? I realize we train for engine outs with a lower probability than that, but engine outs aren't predictable or at least observable. Crosswinds are. Aren't there more likely hazards on which to spend limited training time and dollars?
A more recent accident cited in the USA Today article was the Colgan Air crash in Buffalo in February of 2009. Once again, the NTSB found that flaws in simulator training were a factor. What they can't say, however, is that even if sim procedures were focused tightly on exactly the two rarified eventualities described above that the outcomes would have been any different. In one case, the pilot was dealt a bad hand, in the other, maybe the pilot's skills just weren't up to the task no matter how much he trained.
In any case, shouldn't a pilot bring to the table certain basic skills—like knowing a stall when he sees one or noticing that his knuckles are being rapped by a stick shaker? Sometimes there's just not much you can do about ten-to-the-ninth accidents and I'm wondering if these two don't qualify. Which is another way of saying USA Today has again overstated the case and I feel for the hapless reading public unversed in the finer points of aviation risk.
Speaking of overstated cases, even when I knew the 100-knot crosswind was coming, I couldn't keep the Tomcat on the runway. Besides, the Navy has a solution for this. That's why they point the ship into the wind.