For four hours and 30 minutes, the redeye that took off from Thailand with 181 people aboard made its unremarkable way some 2,000 miles over land and sea.
But in the last four and a half minutes, with the destination in sight, it all came apart.
“Bird!” warned the first officer. “There are many below.”
“Hey, hey!” the captain declared seconds later. “It’s not going to work.”
It was the start of one of the deadliest air disasters in years. Within minutes, 179 people would lose their lives.
Many things went wrong for the crew of Jeju Air Flight 2216, including regulatory failures and bad decisions by others. But for all their bad luck, a New York Times investigation has found, the pilots appear to have worsened the crisis by acting too quickly, rushing into trouble that may well have been avoidable.
After consulting with pilots and aviation experts, we modeled the view inside the Boeing 737 cockpit to show what the pilots were likely to have been experiencing in the final moments of flight.
The chaos would have tested anybody’s nerves. Instinct screams: Do something. Fast.
But pilots are taught not to react to crisis the way others might. “Wind your watch,” was what they were once told. “Light a cigarette.”
“We were trained to wait,” said Ludo Gysels, a retired pilot and trainer. “First get the pieces together before you decide on a course of action.”
Though the wording of the advice may have gone the way of windable watches and cockpit smokes, the message has not changed: Don’t act too quickly.
That is easier said than done.
Much of the explanation may lie in a quirk of human psychology that aviation experts call the startle effect. When things go seriously wrong, and safety is at stake, the initial shock can tempt pilots to make ill-advised decisions.
No pilot is completely immune, however experienced.
Even Chesley B. Sullenberger III, the pilot lionized for saving all aboard a crippled jetliner by landing it on the Hudson River in 2009, felt the startle effect, at least at first.
“Oh yes, absolutely,” he said in an interview with The Times. “It is a very natural, normal human physiological stress response to an extreme, life-threatening challenge.”
The Go-Around
Just 9 seconds after spotting the birds, the pilots, who were speaking in a combination of Korean and English, both said: “Go around! Go around!” Then they launched into a high-stakes maneuver that involved aborting the landing, throttling up the engines and climbing. It may have been their first big mistake.
In the years after Captain Sullenberger’s feat, often called the Miracle on the Hudson, many pilots made a close study of how he managed to safely land an Airbus after a bird strike shortly after takeoff left both its engines virtually lifeless.
Rare as that scenario is, the captain of the Jeju flight, Han Gwang-seop, had used a simulator to practice landing a plane with two dead or severely damaged engines. He was not required to do so; Jeju Air required training only for single-engine loss.
But it is unclear if one of Captain Sullenberger’s most important lessons — wait and analyze the situation before acting — had any effect in the Jeju cockpit the day of the disaster.
In 2008, the crew of a jetliner landing in Rome faced a similar situation as the Jeju crew and made a similar decision, conducting a go-around to try to avoid a flock of starlings. They ended up crash-landing.
The investigation report that followed cited a warning from two Boeing safety officials. The added thrust needed for a go-around, they said, increased the risk of engine damage from birds, because the faster the turbines spin, the more violent the collisions with the equipment inside.
“Consider landing through birds versus a missed approach to avoid birds,” the Boeing officials advised.
Following the Rome crash, Boeing also issued new guidance to all airlines and updated its training manuals, advising pilots on how to handle birds.
Some pilots said the Jeju pilots’ last-second swerve might be understandable if they believed they could not fly through so many birds and land safely. Jeju Air would not comment on whether its pilots should have tried to land through the flock, citing the ongoing investigation. It did say that its flight crews were aware of Boeing’s guidance.
Still, it’s possible the pilots should not have had to face the situation at all.
Muan International Airport failed to meet bird strike prevention guidelines. The airport sits near bird habitats yet lacks thermal imaging cameras and bird detection radar used to alert air traffic controllers and pilots to the presence of birds, according to the government. And on the day of the crash, only one bird patroller was on duty, instead of the minimum of two that government rules require, lawmakers said.
Experts also say that the airport should have given the crew an earlier heads-up about the potential danger.
Michael McCormick, a former air traffic controller and Federal Aviation Administration official, said he was struck by “how large that flock was.” In his experience, a flock that large would not simply appear out of nowhere.
“If the controllers had been functioning as they should have,” he said, “they should have observed the flock, and they should have provided notification to the pilots.”
Had controllers warned the pilots about the Baikal teals — a kind of duck — as the flight was being cleared to land, the crew would have had an extra three minutes to assess the situation and devise a plan.
Instead, the plane was hit by the birds just 36 seconds after the controllers gave warning.
The airport corporation declined to comment on the air traffic controllers’ actions that day.
Engine Trouble
Much of what happened in the Jeju cockpit that morning remains shrouded in uncertainty. In no small part that is because the cockpit voice recorder and the flight data recorder — often called the black boxes — failed to preserve some critical data, including the final discussions of the pilots.
More than anything the instrument panels may have been showing, the 737’s disorienting vibrations would have made chillingly clear to Captain Han and his first officer, Kim Gyeong-hwan, that something had gone terribly wrong.
Both engines were damaged, at least one badly.
Throttling one back could reduce the vibrations. And if the damage was severe enough, an engine could be turned off completely. But which one posed the greatest danger?
This is where the Jeju pilots appear to have made their second big mistake.
It may have been the wrong engine.
Though both engines were damaged, video taken from the ground captures the right engine having the most visible trouble. It can be seen spewing a cloud of black smoke, and perhaps bird parts, from its back right after the bird strike.
The total electrical failure right after the left fire switch was pulled further suggests that the right engine was the bigger problem: Its generator was already offline, and shutting down the left engine disconnected the plane’s remaining source of power. Why the right generator was off remains under investigation.
Boeing’s emergency checklists warn pilots not to rush through a formal call-and-response between the captain and the first officer before shutting down an engine. Once an engine is shut down in an emergency near the ground, there is not enough time to turn it back on again.
“It’s like a reading from scripture,” said Dennis Tajer, a 737 pilot who is a spokesman for the American Airlines pilot union. “Everything has to be read exactly as directed and executed exactly as directed.”
With the left engine shut down, the power it generated for everything from the hydraulics to the black boxes themselves was gone. Only the sputtering right engine remained to provide sporadic thrust and to power some flight controls.
Im Jung Hoon, the chairman of Jeju Air’s pilots union, argued that the pilots could have initially misspoken when they indicated they were shutting down the right engine. They may have always intended to shut down the left engine, he said. He conceded that by not verbally confirming the engine at each step, the pilots skipped parts of the emergency procedures.
Some pilots say that — once again — waiting, and keeping both engines running, might have been the most prudent move.
It is still not known precisely what motivated the Jeju pilots to shut down either engine. Until more flight data is released by Korean investigators, Captain Tajer noted, it is impossible to say exactly what readings the pilots were seeing on their control panels or why they moved as quickly as they did.
But 19 seconds, he said, is “at the very edges of human ability to execute without flaw.”
Final Approach
As the plane’s electrical systems failed, the pilots lost still more control. In yet another questionable decision, the pilots never activated an emergency generator called the A.P.U. It would have restored electrical power to some critical flight functions, and kept the flight data recorder and cockpit voice recorder running.
The failure of the cockpit voice recorder to record the last four minutes of the flight has made it much harder for investigators to reconstruct what the pilots were thinking in those final moments — but there are some clues.
With the Yellow Sea spread out before them, they may have considered taking a leaf from Captain Sullenberger’s book. Photos from the wreckage show, among other pages from the emergency manual, instructions for a water landing lying loose at the crash site.
Captain Han instead told controllers that he was heading in for a landing. But rather than going forward with the full circle usually done in a go-around, he decided to change direction, making a quicker 180-degree turn and landing from the opposite direction.
Captain Han’s training on dual-engine failures in the simulator appears to have served him well at that point. He managed to bank the plane around toward the runway.
But another sheet of emergency instructions found at the crash site suggests that the pilots had wrestled with another challenge. The old-school style page with charts and text explained how to calculate the safety of lowering landing gear when a plane is moving slowly at low altitudes, as the Jeju flight was.
As the pilots were struggling to negotiate the go-around, they had to decide when — or if — to drop the landing gear. And once again, they were paying a heavy price for the loss of the engines.
In the complex circulatory system of a Boeing 737, the landing gear is controlled mainly by hydraulics driven by the left engine. With that option off the table, the Jeju pilots would have had to consider dropping the gear manually.
Belly Landing
It was, in many respects, an extraordinary accomplishment. Captain Han and First Officer Kim piloted a stricken jetliner through a series of turns and landed dead-center on a runway.
But for reasons that may never be known, the pilots landed their craft with the landing gear undeployed.
They may have feared that lowering the landing gear might create so much drag they would not make it to the runway. But had the landing gear been deployed, it would have helped slow the plane down and enabled braking.
Instead, the plane skidded on its belly, overran the runway and hit a concrete wall experts said should never have been there. That wall sealed the flight’s fate. The plane burst into a fireball within a split second.
“Had it not been for that concrete barrier, they probably would have survived,” Captain Sullenberger said. “Circumstances were with us. Circumstances were not with them.”
After Captain Sullenberger’s plane came to a rest on the Hudson, all 150 passengers and five crew members walked off alive. After the Jeju crash, only two people, flight attendants in the back of the plane, survived. One hundred and seventy-nine people were killed.
Jeff Guzzetti, a former F.A.A. and National Transportation Safety Board accident investigator, ruefully noted the parallels between the Sullenberger and the Jeju fights.
“It certainly seems like a Miracle on the Hudson,” he said. “Without the miracle.”
About the cockpit model and flight path
The cockpit interior model is based on a standard Boeing 737-800. The exact selection and layout of instruments may vary by airline. Engine vibration and power indications on the center screen reflect flight data recorder data points; other values are estimated based on the phase of flight. Indications on other screens are not shown for clarity. How the engine vibrations are felt in the cockpit is an estimate based on pilot interviews. The engine throttle angles shown in the animation are approximate and based on electronic signals.
Because the plane’s data record was cut off when its electrical systems failed, we reconstructed the flight path by geolocating CCTV footage and witness videos. We then traveled to some of those locations to measure landmarks on site to help triangulate the plane’s path.