2.
Twenty years before the crash of KAL 801, a Korean Air Boeing 707 wandered into Russian airspace and was shot down by a Soviet military jet over the Barents Sea. It was an accident, meaning the kind of rare and catastrophic event that, but for the grace of God, could happen to any airline. It was investigated and analyzed. Lessons were learned. Reports were filed.
Then, two years later, a Korean Air Boeing 747 crashed in Seoul. Two accidents in two years is not a good sign. Three years after that, the airline lost another 747 near Sakhalin Island, in Russia, followed by a Boeing 707 that went down over the Andaman Sea in 1987, two more crashes in 1989 in Tripoli and Seoul, and then another in 1994 in Cheju, South Korea.*
To put that record in perspective, the “loss” rate for an airline like the American carrier United Airlines in the period 1988 to 1998 was .27 per million departures, which means that they lost a plane in an accident about once in every four million flights. The loss rate for Korean Air, in the same period, was 4.79 per million departures—more than seventeen times higher.
Korean Air’s planes were crashing so often that when the National Transportation Safety Board (NTSB)—the US agency responsible for investigating plane crashes within American jurisdiction—did its report on the Guam crash, it was forced to include an addendum listing all the new Korean Air accidents that had happened just since its investigation began: the Korean Air 747 that crash-landed at Kimpo in Seoul, almost a year to the day after Guam; the jetliner that overran a runway at Korea’s Ulsan Airport eight weeks after that; the Korean Air McDonnell Douglas 83 that rammed into an embankment at Pohang Airport the following March; and then, a month after that, the Korean Air passenger jet that crashed in a residential area of Shanghai. Had the NTSB waited just a few more months, it could have added another: the Korean Air cargo plane that crashed just after takeoff from London’s Stansted airport, despite the fact that a warning bell went off in the cockpit no fewer than fourteen times.
In April 1999, Delta Air Lines and Air France suspended their flying partnership with Korean Air. In short order, the US Army, which maintains thousands of troops in South Korea, forbade its personnel from flying with the airline. South Korea’s safety rating was downgraded by the US Federal Aviation Authority, and Canadian officials informed Korean Air’s management that they were considering revoking the company’s overflight and landing privileges in Canadian airspace.
In the midst of the controversy, an outside audit of Korean Air’s operations was leaked to the public. The forty-page report was quickly denounced by Korean Air officials as sensationalized and unrepresentative, but by that point, it was too late to save the company’s reputation. The audit detailed instances of flight crews smoking cigarettes on the tarmac during refueling and in the freight area; and when the plane was in the air. “Crew read newspapers throughout the flight,” the audit stated, “often with newspapers held up in such a way that if a warning light came on, it would not be noticed.” The report detailed bad morale, numerous procedural violations, and the alarming conclusion that training standards for the 747 “classic” were so poor that “there is some concern as to whether First Officers on the Classic fleet could land the aircraft if the Captain became totally incapacitated.”
By the time of the Shanghai crash, the Korean president, Kim Dae-jung, felt compelled to speak up. “The issue of Korean Air is not a matter of an individual company but a matter of the whole country,” he said. “Our country’s credibility is at stake.” Dae-jung then switched the presidential plane from Korean Air to its newer rival, Asiana.
But then a small miracle happened. Korean Air turned itself around. Today, the airline is a member in good standing of the prestigious SkyTeam alliance. Its safety record since 1999 is spotless. In 2006, Korean Air was given the Phoenix Award by Air Transport World in recognition of its transformation. Aviation experts will tell you that Korean Air is now as safe as any airline in the world.
In this chapter, we’re going to conduct a crash investigation: listen to the “black box” cockpit recorder; examine the flight records; look at the weather and the terrain and the airport conditions; and compare the Guam crash with other very similar plane crashes, all in an attempt to understand precisely how the company transformed itself from the worst kind of outlier into one of the world’s best airlines. It is a complex and sometimes strange story. But it turns on a very simple fact, the same fact that runs through the tangled history of Harlan and the Michigan students. Korean Air did not succeed—it did not right itself—until it acknowledged the importance of its cultural legacy.
3.
Planes crashes rarely happen in real life the same way they happen in the movies. Some engine part does not explode in a fiery bang. The rudder doesn’t suddenly snap under the force of takeoff. The captain doesn’t gasp, “Dear God,” as he’s thrown back against his seat. The typical commercial jetliner—at this point in its stage of development—is about as dependable as a toaster. Plane crashes are much more likely to be the result of an accumulation of minor difficulties and seemingly trivial malfunctions.*
In a typical crash, for example, the weather is poor— not terrible, necessarily, but bad enough that the pilot feels a little bit more stressed than usual. In an overwhelming number of crashes, the plane is behind schedule, so the pilots are hurrying. In 52 percent of crashes, the pilot at the time of the accident has been awake for twelve hours or more, meaning that he is tired and not thinking sharply. And 44 percent of the time, the two pilots have never flown together before, so they’re not comfortable with each other. Then the errors start—and it’s not just one error. The typical accident involves seven consecutive human errors. One of the pilots does something wrong that by itself is not a problem. Then one of them makes another error on top of that, which combined with the first error still does not amount to catastrophe. But then they make a third error on top of that, and then another and another and another and another, and it is the combination of all those errors that leads to disaster.
These seven errors, furthermore, are rarely problems of knowledge or flying skill. It’s not that the pilot has to negotiate some critical technical maneuver and fails. The kinds of errors that cause plane crashes are invariably errors of teamwork and communication. One pilot knows something important and somehow doesn’t tell the other pilot. One pilot does something wrong, and the other pilot doesn’t catch the error. A tricky situation needs to be resolved through a complex series of steps—and somehow the pilots fail to coordinate and miss one of them.
“The whole flight-deck design is intended to be operated by two people, and that operation works best when you have one person checking the other, or both people willing to participate,” says Earl Weener, who was for many years chief engineer for safety at Boeing. “Airplanes are very unforgiving if you don’t do things right. And for a long time it’s been clear that if you have two people operating the airplane cooperatively, you will have a safer operation than if you have a single pilot flying the plane and another person who is simply there to take over if the pilot is incapacitated.”
Consider, for example, the famous (in aviation circles, anyway) crash of the Colombian airliner Avianca flight 052 in January of 1990. The Avianca accident so perfectly illustrates the characteristics of the “modern” plane crash that it is studied in flight schools. In fact, what happened to that flight is so similar to what would happen seven years later in Guam that it’s a good place to start our investigation into the mystery of Korean Air’s plane crash problem.
The captain of the plane was Laureano Caviedes. His first officer was Mauricio Klotz. They were en route from Medellin, Colombia, to New York City’s Kennedy Airport. The weather that evening was poor. There was a nor’easter up and down the East Coast, bringing with it dense fog and high winds. Two hundred and three flights were delayed at Newark Airport. Two hundred flights were delayed at LaGuardia Airport, 161 at Philadelphia, 53 at Boston’s Logan Airport, and 99 at Kennedy. Because
of the weather, Avianca was held up by Air Traffic Control three times on its way to New York. The plane circled over Norfolk, Virginia, for nineteen minutes, above Atlantic City for twenty-nine minutes, and forty miles south of Kennedy Airport for another twenty-nine minutes.
After an hour and a quarter of delay, Avianca was cleared for landing. As the plane came in on its final approach, the pilots encountered severe wind shear. One moment they were flying into a strong headwind, forcing them to add extra power to maintain their momentum on the glide down. The next moment, without warning, the headwind dropped dramatically, and they were traveling much too fast to make the runway. Typically, the plane would have been flying on autopilot in that situation, reacting immediately and appropriately to wind shear. But the autopilot on the plane was malfunctioning, and it had been switched off. At the last moment, the pilot pulled up, and executed a “go-around.” The plane did a wide circle over Long Island, and reapproached Kennedy Airport. Suddenly, one of the plane’s engines failed. Seconds later, a second engine failed. “Show me the runway!” the pilot cried out, hoping desperately that he was close enough to Kennedy to somehow glide his crippled plane to a safe landing. But Kennedy was sixteen miles away.
The 707 slammed into the estate owned by the father of the tennis champion John McEnroe, in the posh Long Island town of Oyster Bay. Seventy-three of the 158 passengers aboard died. It took less than a day for the cause of the crash to be determined: “fuel exhaustion.” There was nothing wrong with the aircraft. There was nothing wrong with the airport. The pilots weren’t drunk or high. The plane had run out of gas.
4.
“It’s a classic case,” said Suren Ratwatte, a veteran pilot who has been involved for years in “human factors” research, which is the analysis of how human beings interact with complex systems like nuclear power plants and airplanes. Ratwatte is Sri Lankan, a lively man in his forties who has been flying commercial jets his entire adult life. We were sitting in the lobby of the Sheraton Hotel in Manhattan. He’d just landed a jumbo jet at Kennedy Airport after a long flight from Dubai. Ratwatte knew the Avianca case well. He began to tick off the typical crash preconditions. The nor’easter. The delayed flight. The minor technical malfunction with the auto-pilot. The three long holding patterns—which meant not only eighty minutes of extra flying time but extra flying at low altitudes, where a plane burns far more fuel than it does in the thin air high above the clouds.
“They were flying a seven-oh-seven, which is an older airplane and is very challenging to fly,” Ratwatte said. “That thing is a lot of work. The flight controls are not hydraulically powered. They are connected by a series of pulleys and pull rods to the physical metal surfaces of the airplane. You have to be quite strong to fly that airplane. You heave it around the sky. It’s as much physical effort as rowing a boat. My current airplane I fly with my fingertips. I use a joystick. My instruments are huge. Theirs were the size of coffee cups. And his autopilot was gone. So the captain had to keep looking around these nine instruments, each the size of a coffee cup, while his right hand was controlling the speed, and his left hand was flying the airplane. He was maxed out. He had no resources left to do anything else. That’s what happens when you’re tired. Your decision-making skills erode. You start missing things—things that you would pick up on any other day.”
In the black box recovered from the crash site, Captain Caviedes in the final hour of the flight is heard to repeatedly ask for the directions from ATC to be translated into Spanish, as if he no longer had the energy to make use of his English. On nine occasions, he also asked for directions to be repeated. “Tell me things louder,” he said right near the end. “I’m not hearing them.” When the plane was circling for forty minutes just southeast of Kennedy—when everyone on the flight deck clearly knew they were running out of fuel—the pilot could easily have asked to land at Philadelphia, which was just sixty-five miles away. But he didn’t: it was as if he had locked in on New York. On the aborted landing, the plane’s Ground Proximity Warning System went off no fewer than fifteen times, telling the captain that he was bringing in the plane too low. He seemed oblivious. When he aborted the landing, he should have circled back around immediately, and he didn’t. He was exhausted.
Through it all, the cockpit was filled with a heavy silence. Sitting next to Caviedes was his first officer, Mauricio Klotz, and in the flight recorder, there are long stretches of nothing but rustling and engine noise. It was Klotz’s responsibility to conduct all communication with ATC, which meant that his role that night was absolutely critical. But his behavior was oddly passive. It wasn’t until the third holding pattern southwest of Kennedy Airport that Klotz told ATC that he didn’t think the plane had enough fuel to reach an alternative airport. The next thing the crew heard from ATC was “Just stand by” and, following that, “Cleared to the Kennedy airport.” Investigators later surmised that the Avianca pilots must have assumed that ATC was jumping them to the head of the queue, in front of the dozens of other planes circling Kennedy. In fact, they weren’t. They were just being added to the end of the line. It was a crucial misunderstanding, upon which the fate of the plane would ultimately rest. But did the pilots raise the issue again, looking for clarification? No. Nor did they bring up the issue of fuel again for another thirty-eight minutes.
5.
To Ratwatte, the silence in the cockpit made no sense. And as a way of explaining why, Ratwatte began to talk about what had happened to him that morning on the way over from Dubai. “We had this lady in the back,” he said. “We reckon she was having a stroke. Seizing. Vomiting. In bad shape. She was an Indian lady whose daughter lives in the States. Her husband spoke no English, no Hindi, only Punjabi. No one could communicate with him. He looked like he had just walked off a village in the Punjab, and they had absolutely no money. I was actually over Moscow when it happened, but I knew we couldn’t go to Moscow. I didn’t know what would happen to these people if we did. I said to the first officer, ‘You fly the plane. We have to go to Helsinki.’”
The immediate problem Ratwatte faced was that they were less than halfway through a very long flight, which meant that they had far more fuel in their tanks than they usually do when it comes time to land. “We were sixty tons over maximum landing weight,” he said. “So now I had to make a choice. I could dump the fuel. But countries hate it when you dump fuel. It’s messy stuff and they would have routed me somewhere over the Baltic Sea, and it would have taken me forty minutes and the lady probably would have died. So I decided to land anyway. My choice.”
That meant the plane was “landing heavy.” They couldn’t use the automated landing system because it wasn’t set up to handle a plane with that much weight.
“At that stage, I took over the controls,” he went on. “I had to ensure that the airplane touched down very softly; otherwise, there would have been the risk of structural damage. It could have been a real mess. There are also performance issues with being heavy. If you clear the runway and have to go around, you may not have enough thrust to climb back up.
“It was a lot of work. You’re juggling a lot of balls. You’ve got to get it right. Because it was a long flight, there were two other pilots. So I got them up, and they got involved in doing everything as well. We had four people up there, which really helped in coordinating everything. I’d never been to Helsinki before. I had no idea how the airport was, no idea whether the runways were long enough. I had to find an approach, figure out if we could land there, figure out the performance parameters, and tell the company what we were doing. At one point I was talking to three different people—talking to Dubai, talking to MedLink, which is a service in Arizona where they put a doctor on call, and I was talking to the two doctors who were attending to the lady in the back. It was nonstop for forty minutes.
“We were lucky the weather was very good in Helsinki,” he said. “Trying to do an approach in bad weather, plus a heavy plane, plus an unfamiliar airport, that’s not good. Because i
t was Finland, a first-world country, they were well set up, very flexible. I said to them, ‘I’m heavy. I would like to land into the wind.’ You want to slow yourself down in that situation. They said, No problem. They landed us in the opposite direction than they normally use. We came in over the city, which they usually avoid for noise reasons.”
Think about what was required of Ratwatte. He had to be a good pilot. That much goes without saying: he had to have the technical skill to land heavy. But almost everything else Ratwatte did that made that emergency landing a success fell outside the strict definition of piloting skills.
He had to weigh the risk of damaging his plane against the risk to the woman’s life, and then, once that choice was made, he had to think through the implications of Helsinki versus Moscow for the sick passenger in the back. He had to educate himself, quickly, on the parameters of an airport he had never seen before: could it handle one of the biggest jets in the sky, at sixty tons over its normal landing weight? But most of all, he had to talk—to the passengers, to the doctors, to his copilot, to the second crew he woke up from their nap, to his superiors back home in Dubai, to ATC at Helsinki. It is safe to say that in the forty minutes that passed between the passenger’s stroke and the landing in Helsinki, there were no more than a handful of seconds of silence in the cockpit. What was required of Ratwatte was that he communicate, and communicate not just in the sense of issuing commands but also in the sense of encouraging and cajoling and calming and negotiating and sharing information in the clearest and most transparent manner possible.
6.
Here, by contrast, is the transcript from Avianca 052, as the plane is going in for its abortive first landing. The issue is the weather. The fog is so thick that Klotz and Caviedes cannot figure out where they are. Pay close attention, though, not to the content of their conversation but to the form. In particular, note the length of the silences between utterances and to the tone of Klotz’s remarks.