Flying is safe in any rational sense, maybe safer than ever. More than three billion people flew on U.S. airlines from 2007 through 2011—that’s 10 times the entire population of the United States—and only 50 died, all in a single regional airline turboprop crash. Statistically, the risk is almost inconsequential. How inconsequential? What people really want to know is their risk of dying (by accident) on a random flight. MIT statistician Arnold Barnett calculated that risk in the United States in June 2009 as roughly one in 23 million. You would have to fly every single day for the next 63,000 years before you would be likely to die in a jet airliner crash, Barnett estimates. You’re more likely to freeze to death or drown in your own bathtub, about 10 to 40 times more likely to face injury or death in an automobile, by one estimate. Choose any American kid at random; he or she is more likely to be elected president of the United States than to die on any given jet flight, and 10 times more likely to win an Olympic gold medal.
The reality is that aviation safety is all about relative risk, and keeping that risk at an acceptably low level. In the commercial airline business, with lots of lives at stake, the tolerance for risk is—and should be—very, very low. But the notion that “one accident is one accident too many” is just an aero-political aspiration. The real-world goal, again, is just to keep the “margin”—an adequacy of extra protection, precaution, and redundancy that keeps risk tolerably low even when somebody makes a mistake or something fails. Getting beyond the dual unrealities of absolute safety and unwarranted fear, air travelers can take a mature look at what’s really worth worrying about, and what’s not.
So what is worth worrying about as you plan your vacation this summer? As the air-travel system gets more crowded and less error-tolerant, and the drive for efficiency and sustained profitability commands center stage, there are still weak links that need attention before somebody gets killed.
1. Naptime in the Cockpit ... and the Tower
Human error causes most fatal airline accidents. When it comes to safety, you don’t want the human sitting on the left side of the cockpit to be fatigued. Up to a point, planes can fly themselves—flight management computers do a great job flying jets when cruising at altitude. Beyond that point, those moments when a pilot’s split-second response is critical, fatigue can mean the difference between a bumpy ride and a catastrophe.
Speeding down the runway for takeoff, pilots can have only one, maybe two, seconds to react to a key warning light or alarm and make the decision whether it is safer to raise the nose and lift off or to hit the brakes and abort. React too slowly and there’s no option. The plane has to fly or you run out of runway if you try to stop. It’s similar landing in heavy crosswinds on an icy runway, or making the right call about when to pull back or push down the yoke (the control column that orders the plane’s altitude and direction) in a sudden blast of wind shear six miles above the remotest regions of the South Atlantic when the autopilot shuts down. These are what some pilots call the “full alert” moments—times that call for skilled, trained, split-second decisions not slowed or dulled by exhaustion.
By their own admission, though, pilots do fly fatigued. A National Sleep Foundation poll of commercial pilots in March 2012 (more than two-thirds flying for airlines) found that one-fifth of the 202 pilot respondents admitted to having made a serious safety error due to sleepiness. The same year, a British pilots’ union survey of 500 members found 43 percent admitted falling asleep at the controls at some time in their careers. In 2011, half of Norway’s airline pilots told a public-broadcasting survey that they dozed off behind the yoke at one time or another; 2 percent admitted doing so “often,” sometimes without alerting their copilot. In a 1999 NASA survey of more than 1,400 pilots at 26 regional airlines, 80 percent admitted to “nodding off” in the cockpit at some time in their careers. Body-clock confusion isn’t limited to international long-haul pilots crossing a dozen time zones in a day; domestic pilots can face even more frequent circadian rhythm disruptions and heavier daily workloads. So if you’re wondering, the FAA allows pilots to take the popular sedative Ambien (generic name Zolpidem)—as long as it’s more than 24 hours before flying and no more often than twice a week.
Asking airline pilots if they’re overworked may be a less-than-scientific way to assess the true extent of cockpit sleeping, but pilot fatigue has been, for more than two decades, high on the list of “most wanted” air-safety improvements published by the independent safety experts at the NTSB. Still, it’s not a sexy safety issue. There’s no new gadget that will correct the problem, and everyone knows that an effective solution—such as hiring more pilots—would cost airlines big money, maybe trigger new labor-relations issues with the pilots’ unions, and encounter fierce resistance from cost-focused managers. The gridlock means that, despite tinkering over the years, rules about how long pilots can fly, and how often and long they must rest, haven’t changed much since Ronald Reagan was leaving the White House. Until 2010, the last time the FAA even tried to update the rules comprehensively was in 1995; airlines and their supporters clogged the FAA mailroom with more than 2,000 written “comments” mostly opposing the changes.
Regulators turned to another approach that might avoid industry blowback, a cost-free way to fight cockpit fatigue: simply let pilots take naps in the cockpit on long flights. Sleep scientists at NASA were for it, and so were pilots. Experts largely agreed it would be better for pilots to nap in a controlled way on long flights than risk dozing at a critical moment or when the second pilot is out of the cockpit. But the FAA ultimately balked at changing its longstanding prohibition on cockpit sleeping.
There were real concerns that the remaining “awake pilot” could become sleepy or incapacitated, or that a groggy “resting” pilot wouldn’t function effectively in an emergency immediately on being awakened. But among the political folks in the regulatory agencies, there was also a less scientific worry—they call it “optics” in Washington-speak—about “controlled rest in the cockpit.” No matter what you called it, sleeping up front violated the "Tonight Show" Rule: if it’s grist for late-night comedians, it doesn’t happen. And sleeping pilots would make passengers nervous—almost as much as they would delight those comics and editorial cartoonists. Little has changed when it comes to the sensitive politics of the issue. When safety experts at the NTSB proposed in 2011 to let air traffic controllers nap briefly during late-night shifts when other controllers were present, the secretary of transportation was adamant: “We’re not going to pay controllers to nap.”
Pilot fatigue remained a nagging back-burner issue until a wintry February night in 2009 when Capt. Marvin Renslow, according to NTSB investigators, made a series of faulty piloting decisions that doomed his Colgan Air commuter flight and its 49 passengers in an ice storm near Buffalo, New York. Crash investigators cited fatigue as a factor; Renslow’s relatively inexperienced copilot, 24-year-old Rebecca Shaw, was hardly well rested. To get to her Newark, New Jersey, flight in time, she had hitchhiked by air all night cross-country from her Seattle home before sacking out on a couch in an airport crew lounge. Her long-haul commute to her $16,254 job wasn’t that unusual. A July 2011 National Research Council report found that more than 20 percent of pilots live more than 750 miles from their jobs. Of 137 Newark-based Colgan pilots, 93 commuted to work by air, sometimes for hours.
The gory details of the Colgan crash—fatigue, pilot error, hiring and training issues—reignited the battle over pilot rest rules, and the FAA in late 2010 came up with what it hoped was a modest proposal—albeit complex at 145 pages—that airlines would accept. As finalized, it would mean that pilots would get ten hours of rest instead of eight hours between flying shifts, with an opportunity for eight hours of uninterrupted sleep, and they could be scheduled to be at the controls no more than eight or nine hours a day, depending on when they started work. Their total workday, including time on the ground and commuting by air to their job, could last no more than nine to 14 hours a day, down from the grueling 16-hour workdays that the previous rules could allow.
The industry went through the roof. Some 8,000 public comments poured into DOT; most opposed the proposed new rule. American, for example, claimed the rule would require it to hire 2,325 more pilots and spend a half billion dollars more every year. The airlines’ trade association estimated that an initial draft rule proposal issued in late 2010 would cost carriers $19.6 billion over 10 years to comply—15 times more than what the FAA projected. Weathering the withering criticism, though, regulators issued a revised final rule—the first significant change in flight and duty time in almost 30 years—a few days before Christmas in 2011. Still, it won’t take effect until 2014, and aspects of the rule could be tied up in court for years.
Air traffic controllers nod off too, especially during overnight “mid-shifts” when flights are sparse and they’re alone in the tower, surrounded only by softly glowing instruments and quietly buzzing fans. (The controllers’ union, the National Air Traffic Controllers Association, or NATCA, prefers the term “restorative break” to “nod off.”) Major hubs like O’Hare, Dulles, Boston, and the New York airports have at least two controllers on duty all night, but dozens of other busy airports with few overnight flights may have just one controller. One of them used to be Washington’s Reagan National Airport, just three miles and less than a minute’s flight time from the White House. That changed on March 23, 2011, when the sole controller, pulling his fourth consecutive graveyard shift, took a half hour’s nap while on duty. Two inbound flights heard only silence when they tried to radio the tower. Both landed safely on their own while in contact with regional traffic controllers, a common procedure at smaller, “uncontrolled” airports, but the “nobody home at Reagan” scenario caused a media firestorm.
It wasn’t the first time. In 2006, a Comair commuter jet had crashed while trying to take off from the wrong runway in Lexington, Kentucky, while the lone tower controller—working on two hours of sleep—was distracted. So when the National Airport incident hit the Washington Post front page the next morning (“Tower at Reagan National Goes Silent as Planes Attempt to Land”), all hell broke loose. By day’s end, congressional inquiries were brewing, the secretary of transportation—the executive department that includes the FAA—had launched an investigation and ordered a second controller to man the tower, and the head of the FAA had suspended the hapless napper. By late afternoon, a “nationwide review” of the air traffic control system was under way, and an “interim plan” was in the works. New reports of napping controllers slowed only after the FAA announced a plan that still barred sleeping but let late-night controllers take a “break” as long as they “conduct[ed] themselves professionally” and were “available for recall.” And used an alarm clock?
2. Commuter Airline Safety—A Different Level?
It may be extreme, but some serious people with serious safety concerns—including one former DOT Inspector General—have gone out of their way to avoid commuter airlines. I don’t and, since these airlines account for more than half of all commercial flights in the United States every day, I probably couldn’t. The fact remains, though, that, all told, commuter carriers just don’t have the same safety record as the major airlines that typically contract with them to fly their passengers.
“Regional airlines,” a term often used interchangeably with “commuter airlines,” such as Delta Connection or US Airways Express or United Express look like the major airlines they serve, but most are really subcontractors, completely separate businesses flying under the big airline brands. Forget the names and logos painted on their planes; these carriers—including SkyWest, Mesa, Colgan, ExpressJet, Republic, and Pinnacle—are on their own, with their own pilots and training and maintenance. The big airlines call them “partners,” but that’s just marketing lingo. Most (though not all) sell seats on their commuter-sized planes (generally under 100 seats) to the major airlines for a negotiated per-flight fee.
Over the last decade, nearly all of the U.S. commercial airline fatalities involved regional airline crashes, and, as the Wall Street Journal reported in 2009, commuter airlines from 2003 until then had ten times the serious accident rate per flight than the major airlines. The real risk to any individual flier is still tiny, but it is higher. This isn’t news to regulators. After a series of commuter-airline crashes in the mid-1990s, the FAA announced grand plans for “one level of safety”—translation: trying to make commuter flying as safe as the major airlines. “Safety summits” were held and speeches given, but some concrete proposals met powerful resistance, including from flight schools and aviation colleges that turn out new pilots as quickly as possible for entry-level industry jobs.
Nearly 20 years later, “one level of safety” for all airlines, big and small, remains largely an aspiration. Commuter captains still must meet the same minimum FAA experience standards as their large-airline brethren to earn an air transport license, but those are minimums. In the real world, there can be a chasm in training and experience. A full-fledged “air transport pilot” needs a minimum 1,500 hours of flying time—equivalent to flying five to six hours every workday for an entire year—but virtually every captain at the big airlines has thousands more hours of actual flying experience than the minimum required.
The gulf has been even wider when it comes to regional airline first officers. Until new Congressionally mandated FAA pilot training rules required all airline pilots and copilots to earn full “air transport” ratings, copilots needed only a “commercial” license, the same as crop-dusters and beach ad banner-towers. For that you must be 18 years old, have a private pilot’s license, and have flown for 250 hours (essentially several months). Though regional airlines typically insist on more experience than that, some are scrambling for pilots to meet the tougher requirements expected to take effect in 2013.
3. When Airlines Don’t Obey
When a yard-long tear in the top of the fuselage of a Southwest 737-300 flying over Yuma, Arizona, in 2011 forced an emergency landing, the FAA quickly ordered every airline flying that “classic” version of the venerable 737—dozens in U.S. fleets—to complete detailed metallurgical inspection of the heavily flown planes within five days—too bad about the havoc played on schedules and stranded passengers. The FAA order was broader and more urgent than most, but hardly unique. Almost every weekday, on average, regulators issue airworthiness directives (ADs) telling U.S. airlines to check for or correct unsafe conditions. Often the fix is relatively simple—for instance, replacing a $240 metal pin on a landing-gear assembly for one type of 777. In rare cases, though, an AD can ground scores of planes for urgent and costly inspections that must occur before their next flight.
The FAA’s air safety directives—about 250 are issued every year—are not optional. They have the force of law. They indicate that there is a known safety problem that must be corrected or checked within a specified time in order to make or keep a plane safe to fly. That doesn’t mean, though, that airlines always do precisely what they’re told to do when they’re supposed to do it. Often the failure is inadvertent—a sloppy repair that doesn’t completely comply or genuine confusion about exactly what the FAA wants (certain complex technical directives can run hundreds of pages long). Sometimes, it’s a failure of record keeping-—the check or repair is made but not properly documented. But in the real world, there’s also pressure to keep planes moving on time and to keep costs down. What if postponing the FAA-required inspection for a few days would avoid the cost and schedule grief of yanking a large jet out of the flight rotation just before it’s due to be down for its next four-to-six-month regular maintenance anyway?
How serious is the risk that airlines won’t do what safety inspectors and regulators tell them to? The FAA inspectors may be “tough as nails,” as a former agency head told a 2008 annual “forecast conference,” but in reality, much, if not most, safety regulation is based on trust. It practically has to be. Even with some 4,500 FAA safety inspectors and technical specialists combined, the government can’t closely oversee everything that happens daily in an aviation system as vast as America’s. We’re talking about well over 25,000 airline flights every single day, plus 5,000 aircraft repair stations and 600,000 U.S. pilots. “FAA is unlikely to ever have enough safety inspectors to oversee every aspect of aviation,” DOT’s Inspector General acknowledged in 2012. Fortunately, U.S. airlines normally comply with safety edicts, and so earn that trust. A large-scale FAA audit in 2008 of 5,600 government safety directives found 98 percent compliance by airlines. In most industries, that’s a great result. In aviation, though, anything short of 100 percent isn’t.
4. Jet Geriatrics
Older aircraft are a bit like older people—they keep going, but it takes more and more to keep them healthy. And by the end of the difficult decade after 9/11, U.S. airlines were flying some of the oldest jets in the developed world. According to websites that track aircraft fleet ages, like Airsafe.com and Airfleets.net, some MD-80 and -90 jets flown by American and Delta averaged 20 years old; so did a third of Southwest’s fleet of Boeing 737s. US Airways’ Boeing 767s averaged 22 years old. While big U.S. airline fleets today overall average some 12 to 16 years old (several low-cost carrier fleets are half that age), as of 2008, half of the world’s 4,400 aging aircraft (those at least 21 years old) were flying in the United States, according to an analysis reported by ABC News that year.
Buying pricey new airplanes was simply not a priority during the difficult years soon after September 11. Those shiny new wide-bodies taking off from Seattle’s Boeing Field were headed for Dubai and Beijing, not Dallas or Chicago. For years, the official line was that new planes were better for fuel efficiency, but that no plane, properly maintained, is too old to fly safely. In one sense that may be true. Chronological age matters less when it comes to an aging airframe than the number of flights the airplane has made, specifically how many “cycles” of pressurization and depressurization stress it has undergone. Every time a plane is pressurized, forced air pushes on the airtight fuselage from inside out. (Think of blowing up the same balloon repeatedly.) A 14-hour New York-Tokyo flight pressurizes the aircraft only once—much less than the stress on the structure of a short-haul narrow-body jet that flies that same day to four or five cities. The long-range jumbo jet may be fine for two or three decades, but even a much newer 737 on a workhorse schedule can have problems in half the time. The Southwest 737 that ripped open in Yuma was a middle-aged 15 years old.
Even if industry lore holds that planes need never get “too old,” there’s still a caveat: They need ever-more-careful inspection and TLC the older they do get. After thousands of takeoffs and landings, tiny fatigue cracks in the metal can develop and spread, then link with other tiny cracks to undermine the aircraft structure—especially around fastener holes where rivets hold together the overlapping aluminum plates of the fuselage. This type of metal fatigue was blamed when an 18-foot slab of fuselage ripped off a 19-year-old Aloha Airlines 737 “island-hopper” at 24,000 feet between Hilo and Honolulu in 1988, blowing a flight attendant out of the aircraft after explosive decompression. The plane had recorded nearly 90,000 pressurization cycles—more than a dozen flights every day for its not-so-long lifetime.
There’s plenty of debate about the quality of this outsourced airplane maintenance and also about how well the FAA monitors it—an issue that grows in safety importance as planes age. As aircraft technology gets ever more sophisticated, so does the task of maintaining it. A typical passenger jet today has more than four million parts, Boeing estimates, and the newest A380 jumbo jet contains more than 300 miles of wiring alone. Traditional maintenance work—fixing loose airframe rivets, tightening fasteners, and checking hydraulic lines—is just the start. Every five years or so, each airliner is virtually disassembled, then checked inch by inch—a month-long maintenance visit known as a “D check.” Walls, floors, ceiling panels, lavatories, and galleys are removed to look for cracks and corrosion. Landing gears and hydraulics are largely replaced. Engines are separately inspected and tested. Fluids are sent for laboratory analysis. Even fuel-tank interiors are checked.
Only 100 or so of the FAA’s several thousand safety inspectors are assigned to oversee the 700 foreign repair stations that perform this work, and industry critics worry that outsourcing today’s complex maintenance is exceeding the FAA’s limited ability to oversee it, at home and abroad. The DOT’s Inspector General, for instance, found it “imperative” in 2012 that the FAA “provide more vigorous oversight of this [repair] industry,” both foreign and domestic, and the FAA has assured concerned members of Congress that it will tighten up monitoring. Still, with thousands of aircraft-maintenance jobs in the United States on the line, you’ve got to wonder if some of the brouhaha isn’t really as much about economics as safety. The president of the Teamsters, which represents about 20,000 airline mechanics, says outsourcing maintenance abroad is “a betrayal of passengers’ trust.” On the other hand, the pilots who fly the planes haven’t made it a priority issue. Meanwhile, apparently the only maintenance-related U.S. commercial airline crash in the last decade involved an Air Midwest turboprop flight from Charlotte, North Carolina, in 2003 that lost control on takeoff. NTSB investigators faulted in part lousy maintenance work—performed in Huntington, West Virginia.
5. Something’s Gotta Give
Nobody dresses up to fly anymore; air travel has become routine, commonplace, accessible—so much that it’s stressing the system. We’re cramming more planes into more congested airspace, flying them longer and farther, turning them faster, and overtaxing busy runways and airports and air traffic control systems that have expanded little in the last 20 years at least. The last new big U.S. airport opened in Denver almost two decades ago, planes still navigate with World War II-era radar-based technology, and the pressure on controllers to move the airborne aluminum only accelerates. The gnawing worry is just that something’s gonna give. What keeps aviation-safety folks up at night?
More flying means more workload for air traffic controllers—and a greater chance of error. In the year between October 2009 and October 2010, FAA “operational errors” jumped more than 50 percent. The most serious type of errors—the kind that really could end in midair disasters—climbed nearly 50 percent in just two years. The error frequency has since leveled off but remains almost double the error level of 2007. While there’s debate about whether the jump was due at least partly to more accurate reporting of errors, the disquieting bottom line is that planes are coming closer together than they’re supposed to nearly a half dozen times every day, on average.
Most often, that doesn’t mean really dangerously close. Even nearly five miles apart can constitute what the FAA calls a “proximity event” that transgresses the agency’s complex and conservative separation standards when a small plane is landing behind a much larger jet. Even 50 nautical miles’ separation can be the bare minimum between jets cruising over the ocean beyond radar range. But the most serious mistakes are growing too. The most “severe” or “category A” incidents—those that pose an imminent threat of collision and require evasive action—more than doubled from mid-2008 to mid-2011 and now occur almost weekly on average.
These errors, too, have leveled off, but there’s another lurking issue behind the numbers: Lots of today’s air traffic controllers are relatively green, hired in the last few years to fill the spots of experienced controllers who, as twentysomethings, were recruited en masse to replace the nearly 13,000 PATCO air traffic controllers fired by President Reagan for staging a walkout in 1981. In recent years, those 1980s new hires have been hitting the mandatory retirement age of 56 (it’s a tough job) in droves. That means “trainees” (and what the FAA calls “developmentals”) manned about 26 percent of the radar scopes nationwide in 2010, albeit under experienced supervision. The number rises to near 40 percent of the controller workforce at some of the busiest facilities, such as Denver and the Southern California terminal radar control area, or even the tower at ultrabusy LaGuardia.
Forget about midair disasters. The deadliest aviation accident in history occurred on the tarmac—in the Canary Islands in March 1977—when two fully loaded jumbo jets collided almost head-on approaching takeoff speed on a foggy runway, killing 583 people. That was a quarter century ago, but at busy, crowded airports with multiple taxiways, high-speed turnoffs, and low visibility, pilots still get confused, tower radio communications still get garbled, and planes still make dangerous wrong turns at night. Busy controllers hustling to get planes in the air can have little time to play tarmac traffic cops.
New airport guidance technology and sustained FAA focus seem to have kept the worst problems (the “bare-miss” category especially) in check when it comes to commercial airliners. The most dangerous kinds of incursions dropped from 53 in 2001 to only 7 in 2011, even though total runway incursions held constant at around 1,000 every year. Together with runway excursions—where the plane undershoots or overruns the tarmac—runway screwups still make NTSB’s “most wanted” list of safety improvements. And they have since the list’s inception in 1990.
Horizontal Man-Made Mini-Tornadoes
The airspace around airports can hold only so many planes safely at any one time. Congestion ups the risk they could smack together, but it also amplifies another serious, if less obvious, risk to safe flight: “wake vortex.” Jet aircraft trail behind them potentially deadly wakes of air turbulence that swirl off their wingtips. In some atmospheric conditions, you can even see these whorls coming off landing jets in super-humid air or low clouds near airports. As airline pilot and writer Patrick Smith vividly recounted in his Salon “Ask the Pilot” column, these normally invisible horizontal mini-tornadoes can drive to the ground a plane that’s following behind the wake of another. It’s especially dangerous when the following plane is flying near stall speed at low altitude, on takeoff or near landing. Pilots say the effect of flying into a vortex can be like “hitting a wall.” Just two months after the 9/11 attacks, 260 passengers and five on the ground died when American Flight 587 leaving JFK apparently hit the powerful wake of a loaded Japan Airlines 747 taking off moments before; the pilot overcorrected and the rudder failed as the plane’s tail (vertical stabilizer) tore off.
Tower controllers take care to warn pilots cleared to land behind a larger jet: “Caution, wake turbulence.” But how far planes need to stay separated to avoid wake turbulence depends on the type and size of the aircraft involved, and their speed. Large, heavy wide-body jets generate more turbulence, so most single-aisle jets must stay four to six nautical miles behind them on landing, for example. The superjumbo A380 creates such a wake that small jets were originally required to stay back 10 nautical miles, though that distance has since been reduced to eight nautical miles. All of the required extra spacing between jets at busy hubs in order to avoid wake turbulence further clogs the system. As aircraft grow larger, it takes more time to safely “fit” them on the same flight paths in the same airspace.
Excerpted from "Full Upright and Locked Position: Not-So-Comfortable Truths About Air Travel Today" by Mark Gerchick. Copyright © 2013 by Mark Gerchick. With permission of the publisher, W.W. Norton & Company, Inc.