The Mystery of Flight 427

by Bill Adair

  New simulations by NTSB computer whiz Dennis Crider showed that a reversal would explain all three incidents. Crider’s most important finding was this: The rates of rudder movement on the three planes were nearly identical. That was powerful evidence. It was highly unlikely that three pilots on three different days in three different airplanes would move the rudder at exactly the same speed.

  Loeb and Haueter now were focused on counting votes. Goglia had disqualified himself from voting, but they still needed three of the four remaining board members. If Francis voted against them, they needed everyone else.

  Haueter was nervous that they wouldn’t get the votes. He had heard rumors that the board members were split 2-2 on whether to blame the plane. They were said to be skeptical about the report and wanted to blame the crash on “undetermined reasons.”

  Undetermined reasons.

  Those were the big blue words on the cover of the Colorado Springs report, and Haueter winced at the thought of them on his report. If they appeared on the Flight 427 report, Haueter feared he would be remembered “as the guy who flubbed it.” Even worse, the 737 would continue flying with its elusive rudder problem, and the NTSB would have no leverage to force Boeing to fix it.

  Haueter was concerned that Crider’s findings were not having an impact with the board. Crider had to keep revising them as he made last-minute discoveries, but the numbers always added up to the same conclusion—a rudder reversal. Haueter was afraid it looked to the board members like the investigation was in chaos and was biased toward that conclusion.

  Haueter and Loeb were also frustrated by Phillips, who was reluctant to blame the crashes on rudder reversals. Phillips said the reversals were a plausible explanation, but that he didn’t have enough solid evidence to be definitive. “I can’t lay a part in front of you and say this is what broke,” he said. He felt the NTSB lost credibility if it pushed too hard for a cause when the evidence wasn’t solid.

  Haueter thought his friend was being wishy-washy. He said Phillips would go out to Boeing and come back sounding as though he had been brainwashed. He kept telling Phillips that they had a strong case and had some latitude because they only had to come up with a probable cause.

  As Crider scrambled to finish the new computer simulations, Boeing, which knew all the angles to work at the NTSB, was busy lobbying the board members. Boeing had taken each of the board members for a ride in the M-Cab simulator to show how easily Flight 427’s pilots could have saved the airplane. The company also had provided the board members with a video about the plane that discussed the extensive plans for safety improvements.

  USAir was doing its own lobbying, trying to persuade the board members that the plane was at fault and thus putting the airline in the odd position of arguing that every 737—including the 200 that the airline itself operated—had a safety problem. A USAir official came to the NTSB offices and showed the board members a horrifying video that combined a computer animation of the crash with the actual cockpit voice tape. It gave board members a chance to see and hear two pilots fight and scream and then die, to emphasize USAir’s position that the pilots had no idea what the plane was doing.

  Haueter and Loeb figured they had a good chance with two of the board members—Chairman Jim Hall and John Hammerschmidt. Hall did not have the technical background to understand the intricacies of the valve, but he had been suspicious of Boeing and seemed to have high regard for the staff recommendations. Hammerschmidt, a shy man who was virtually invisible as a board member, was also expected to go with the staff recommendation.

  Bob Francis was shaping up to be just as difficult as Haueter had feared. He said the report was too absolute about the rudder reversal and that it was overly critical of the FAA. In his view, the NTSB had “fairly shaky evidence” and should not be so critical. He met with Hall in the chairman’s office and said he would vote against the report unless it was toned down.

  The other key board member was George Black, a brainy highway engineer from suburban Atlanta. He spent far more time studying the evidence than the other board members and filled a spare office with engineering reports, maps of the plane’s radar track, and a small plastic model of a 737 that he used to demonstrate the crash. He scrawled a sign for the door that said, THE WAR ROOM.

  No one was sure how Black would vote. He liked to play devil’s advocate, throwing out new ideas that often contradicted each other. He liked to tease Boeing executives about the possibly dismal future for the 737, their best-selling product, if the rudder was blamed for the crash. “Boeing has no corporate sense of humor,” he grumbled.

  Black was convinced that the plane had a problem in its rudder system. He had discussed the crash with pilots and engineers and decided that Emmett and Germano would not stomp on a rudder pedal and hold it to their death. He was especially impressed with Crider’s simulations and Brenner’s work matching the grunts to the precise moments at which the rudder appeared to reverse. He was not persuaded by Boeing’s M-Cab ride and thought it was understandable that the pilots—unaware of the dangers of the crossover point—could lose control of the plane.

  Yet he thought the report was too strong. “Tom and Bernie come in here thumping the report like it were the Torah,” he said, but they still had no proof. He said the board members had some latitude because they were naming the probable cause, but he wondered, “At what point does this rise to the level ‘probable’?” He also was concerned that Crider’s simulations were a little too perfect, matching the reversal scenario every time, even when Crider had to make adjustments. Haueter assured Black that the reversal scenario was just one of several that could match, but Black was still wary. He worried that Loeb and his deputies were too adamant about the plane.

  Suddenly, right in the midst of the deliberations, there was another 737 scare: a USAir MetroJet plane had a strange rudder incident.

  Shortly before noon on February 23, 1999, MetroJet Flight 2710 was cruising at 33,000 feet over Maryland when the pilots noticed the wheel suddenly turn to the left. They quickly realized the autopilot was turning the wheel to compensate for rudder movement. But the pilots had not touched the rudder pedals. The copilot then put his feet on the pedals and discovered that they were displaced to the right. He turned off the autopilot and pushed on the left pedal to return the rudder to the center position, but the pedals seemed to be jammed.

  The pilots quickly followed an emergency procedure that had been developed by Cox and another USAir pilot. They turned off the yaw damper, the device that makes small adjustments to the rudder, and activated the standby rudder system, which uses a backup hydraulic valve instead of the main one. The rudder pedals moved back to the center.

  The pilots announced to passengers that they had a flight control problem and then landed at Baltimore-Washington International Airport. After they parked at the gate and everyone got off, a platoon of investigators from the safety board, the FAA, ALPA, and USAir arrived and began removing evidence. They took the two flight recorders, the yaw damper coupler, the rudder PCU and its valve-within-a-valve, and one liter of hydraulic fluid.

  The flight recorder showed that the incident was very different from the Pittsburgh crash. Instead of the fast-moving hardover, the MetroJet rudder had moved slowly to its limit. One FAA official called it “a slow-over.” The incident was even more curious because the plane had one of the new, improved rudder valves that prevented a reversal, and the valve showed no signs of a jam.

  The incident left Haueter and Loeb scratching their heads. It added urgency to their work in the final weeks before the board meeting, but it also raised some troubling questions. Had they found all the problems that might lurk inside the rudder system? Did the valve have a different flaw that they had not uncovered? Had there been a jam somewhere else in the PCU? Cox, who had blamed the USAir crash on a “gremlin” in the plane, said the MetroJet incident showed there was “another gremlin in the tail of the 737.”

  It also revealed a major shortcoming
of all their work. Haueter’s team had some of the smartest engineers in aviation, supplemented by the Greatest Minds in Hydraulics. Boeing had deployed its best and brightest and called in people with decades of experience. But for all that brainpower, they still had not come up with a definitive answer about what had happened to Flight 427. That was remarkable because they were not dealing with a million lines of computer code or a newfangled electronic gadget. They were dealing with an old mechanical device—an ordinary valve designed when John F. Kennedy was president. And now, more than thirty years later, they were struggling to understand the gadget and the myriad ways it could work.

  “I don’t even think the inventors understand it,” said Steve O’Neal, the FAA flight test engineer. He said the MetroJet incident suggested that more changes were needed in the rudder system. “We just hope another 737 doesn’t come screaming out of the sky in the meantime.”

  George Black, the NTSB board member, was worried that the MetroJet incident revealed that the investigators had been too fixated on the valve and had missed a problem elsewhere in the plane. He knew they had ruled out hundreds of possibilities—everything from bird strikes to the fat guy theory—but he was still worried that they were missing the real problem. He was concerned that they would approve the report and then end up having to revise it later. “Are we premature?” he asked. “We want to make sure we don’t start off down some path and decide it was unnecessary.”

  As they debated the wording of the report, Black and Francis were particularly concerned about one of its findings, which lambasted the FAA for certifying the 737 rudder system. The finding said that the 737 would not have been certified if the FAA had insisted on “a high level of safety.”

  If that was true, the board members said, the NTSB should be calling for the entire 737 fleet to be grounded—which would be viewed as a ridiculous request. The plane had 92 million flight hours and one of the lowest crash rates of any transport jet.

  Black and Francis also complained that the report’s recommendations for the rudder system were too specific, reading like a mandate for Boeing to split the system into two separate valves. They said the NTSB should not tell Boeing or the FAA how to design airplanes. The board should recommend broad principles and leave the details to the experts. Black kept invoking a philosophy used by physicians: “Do no harm.”

  Haueter and Loeb insisted that they did not want to ground the plane, which would put entire airlines out of business and wreak havoc with the world’s economy. But they argued that the rudder system needed to be redesigned because a single failure could cripple the plane. Indeed, two years earlier, the five board members had unanimously approved a safety recommendation letter that said the 737 was not as safe as other planes.

  While the other board members debated the wording, Hall began to strategize how he could get a unanimous vote. He believed a 4-0 decision was critical in order to maintain the board’s credibility on such a touchy issue. Hall was comfortable with the strong wording in Loeb and Haueter’s report, but he was perfectly willing to tone it down to get a 4-0 vote.

  He was in a bind. If he kept Loeb’s strong language but ended up with a 2-2 deadlock, the whole four-year investigation would go down the drain. The FAA and Boeing might end up doing nothing to fix the plane. On the other hand, if the report was weakened too much, there was a risk that Loeb would get angry and publicly criticize the board. Loeb had never explicitly made such a threat, but he didn’t have to. He was about to retire and had nothing to lose. He was well regarded among aviation reporters and a complaint from him that the board wasn’t tough enough would surely make front-page headlines.

  The deliberations showed the NTSB as a dysfunctional family. Francis rarely spoke with other board members or with the investigators, yet he complained that Loeb wouldn’t permit his underlings to speak freely. But Loeb and the staff complained that Francis was aloof and had not spent much time studying the accident. Likewise, relations between Black and Loeb were strained. Black felt that Loeb was too abrasive and that he stifled discussion of ideas.

  Hall appointed his assistant, Deb Smith, to act as a peace negotiator. She shuttled back and forth from Francis to Black to Loeb, trying to broker a compromise. Black and Francis had tremendous leverage in getting the changes they wanted. Francis, using his special counsel, Denise Daniels, as his own peace negotiator, sent a lengthy memo that detailed his concerns. He said he would vote against the entire report unless the changes were made. His complaint was primarily about tone. He wanted the report softened so it did not sound so absolute that the rudder had reversed.

  There were lengthy debates about a single word. The draft report had said the 737 needed a rudder system that was “truly redundant.” Loeb and Haueter had added “truly” to strengthen the sentence so McSweeny, the FAA safety official, would not have any wiggle room. They were afraid if they simply said the valve was not redundant, McSweeny would retort that it was redundant. By adding “truly,” they gave the sentence more impact.

  Black, who was often annoyed by Loeb’s aggressive style, was not about to accept Loeb’s word. Black suggested “reliably redundant,” which ultimately was adopted as the final wording.

  The group also debated whether to ask the FAA to establish an independent panel to assess the 737’s rudder system. The panel, which would have representatives from Boeing, the FAA, the NTSB, and academia, would conduct a year-long examination and make recommendations on how the plane could be improved. Loeb regarded that as a waste of time. The NTSB was supposed to be the independent safety agency, he said. There was no need to call in another independent group to validate their work. That just undermined the NTSB’s authority. But Black and Francis liked the idea of an independent panel that would take a broad look at the 737’s problems and make an impartial recommendation. The suggestion for the panel got added to the report.

  After exchanging drafts by E-mail, the board members also decided to tone down the probable cause statement. Instead of Loeb and Haueter’s definitive assertion that a rudder reversal had caused the crash, the board members softened it to say the sudden movement by the rudder was “most likely” caused by a jam and reversal.

  As the final meeting neared, it appeared that Smith and Daniels had brokered a compromise. But as they walked into the hotel ballroom, they were nervous that the deal might unravel.

  The final meeting on Flight 427 was held at the Springfield, Virginia, Hilton on March 23 and 24, 1999. The big room looked like a movie set. Blue lights illuminated the curtains. The board members sat behind a wooden desk like a jury deciding if the 737 was guilty of murder.

  Chairman Hall opened the meeting by reading a statement that said the event was part of the Government in the Sunshine Act, which required federal agencies to do their work “in open session.” But in fact, all of the real debate had occurred behind closed doors long before the meeting.

  It was one of the biggest days in Haueter’s career, and his wife, Trisha Dedik, had come to watch. Haueter hadn’t slept well the night before the meeting, waking up several times worrying about the outcome. But as he tossed in bed, he couldn’t pin down any single thing that was likely to go wrong.

  The ballroom was packed with about three hundred people. Relatives of Flight 427’s passengers sat in a special section, many wearing photos of the victims on buttons or chains around their necks. Brett Van Bortel took a seat in the last row of the family section. He didn’t realize it, but sitting in the row directly behind him were the people he blamed for Joan’s death—top officials from Boeing and USAir.

  Brett was looking healthy and confident. It had been four and a half years since the crash, and he had healed as much as anyone could. He was dating again and engrossed in his job at a mutual fund company. He listened attentively as Haueter began his presentation.

  “Today, the investigative staff is pleased to present the report on the crash of USAir Flight 427,” Haueter said. He noted that it had been the longest investigation
in the safety board’s history.

  Over the next several hours, Haueter and his team explained the 737 rudder system and why they believed it had reversed on the USAir, United, and Eastwind planes. They spoke in absolutes, as if there was no doubt about what happened.

  “The pilot is surprised and pushes harder—as hard as he can,” Malcolm Brenner said about Flight 427. “But instead, the controls reverse and move the rudder all the way to the left.”

  Brenner talked about the risks of the crossover point. He said Emmett, Flight 427’s first officer, discovered “for the first time in his career” that he could not turn the wheel to bring the wings level and stop the plane from rolling out of control. The veteran pilot was helpless as the big 737 began to plummet toward the ground.

  Crider, a goateed former McDonnell Douglas engineer who was considered a genius with flight data, used so much jargon in his presentation that many people in the audience had no idea what he was saying. But they understood when he summarized the 737 incidents by saying, “A rudder reversal scenario will match all three events.” In the audience, Boeing engineers could only listen and bite their lips. They had no opportunity for rebuttal. This was the NTSB’s show.

  Smith and Daniels sat behind the board members wondering if their compromise would fall apart. But as the board questioned Haueter and his staff, there were no hints of any disagreements. As the session stretched into late afternoon, Hall decided to adjourn and resume the following day, although many NTSB staff members wanted to wrap it up that evening.

  The next morning, Haueter was feeling confident as he drove his Toyota 4-Runner to the hotel. “Everything we wanted to say is now out there,” he said. “I feel pretty good about that.” But as he pulled into the hotel parking lot, Haueter could see that his message about the 737 was not getting the attention he had hoped it would. There were only two TV trucks in the lot. Many network crews had been diverted to the Pentagon because U.S. planes had begun bombing Yugoslavia a day earlier. Hall’s decision to extend the meeting to a second day also diluted the impact, because reporters had to write two incremental stories instead of a single, more powerful account. In many newspapers the 737 decision would be relegated to a short story buried inside.