by Bill Adair
Phillips got interested in accident investigation as an engineering student at the University of Evansville in Indiana. In 1977 the school’s basketball team had been killed in the crash of a DC-3 because of a problem with the plane’s rudder. Phillips read everything he could about the crash and decided he wanted to be an investigator. After he graduated, he designed airplane components for Cessna in Wichita, Kansas, and for Northrop in Los Angeles until he saw an ad for an NTSB job in Aviation Week and Space Technology in late 1987.
He had frizzy brown hair, button eyes, and a dimple on his chin. He always sounded happy, even when he was buried in work and about to be sent on a fifteen-hour flight to yet another crash. Like his friend Haueter, Phillips loved to put things together. He was building an RV-8, a lightning-fast experimental plane that could be used for aerobatics. He drove a Porsche 944 and a Harley-Davidson motorcycle and brewed his own beer. Phillips had the perfect personality for the safety board’s culture of caution. He was thorough and always skeptical. When other investigators were convinced about some kind of malfunction, Phillips was often the last holdout, constantly looking for more evidence.
Many investigators came across as dispassionate. They were like homicide detectives. They didn’t get wrapped up in the tragedies they saw because every day brought a new one. But Phillips wasn’t afraid to discuss his feelings about what he saw in the field. He said his job reminded him how fragile life is. “There’s no guarantee for the next day,” he said. “You can’t put off expressing emotions to people you love.” After an accident he often spent extra time with his wife, Debbie, and his two teenage boys, savoring each moment with them.
Phillips was blunt about the limits of technology. “You can’t have a perfect airplane,” he said. “I don’t think perfection exists.” The systems on modern jetliners were so complex that no human being could possibly account for the myriad ways in which things might fail. Yet he never worried about his own safety when he flew. He felt that the manufacturers built solid airplanes, that pilots were well trained, and that the government did a good job of regulating the airlines and the manufacturers. “The system works,” he said.
When he arrived on the hill in his rubber suit, the largest piece of wreckage he saw was the tail. On the wreckage map, the tail was piece No. 1, right at the center, like a big X on a treasure map. The rest of the plane had been squashed and shredded by the impact, but the tail was in surprisingly good shape. That was good news to Phillips because two of the primary flight controls were on the tail: the rudder (the movable vertical panel that made the plane’s nose go right or left) and the elevators (the horizontal panels that made the nose go up or down). Fire had devoured an eleven-by-four-foot section of the vertical stabilizer, which was the big fin area in front of the rudder, and there was a chunk missing from the top portion of the rudder panel itself. But inside the tail, the hydraulic devices that moved the rudder were in good condition.
Phillips’s approach on the hill followed the classic NTSB method. He had his team examine each component from the flight control system, even if the component was unrelated to the rudder. It was far too early to jump to conclusions. They did not want to zero in on one theory too fast, the way the investigators of the Knute Rockne crash did, only to find later that they had neglected something important.
His team found the other mangled flight controls—the ailerons (the wing panels used for banking and turning), and the flaps and slats (the devices on the front and rear of the wings that provided extra lift during takeoffs and landings). For each one, they also found the actuators—the hydraulic or mechanical devices that moved the panels. Phillips’s team carefully measured and photographed everything, to record whether the panels were up or down, right or left at impact. By measuring where things broke or where they bent, they were able to put together a snapshot of what the plane was doing when it struck the road.
The group spent a lot of time examining the device that moved the rudder, which was called the power control unit. Amazingly, it had survived the crash with virtually no damage.
“This thing is in pretty good shape,” Cox said when he saw it.
Phillips was relieved. He knew that the rudder unit would be the subject of many tests. One of his biggest regrets from the Colorado Springs crash was that the rudder system had been badly damaged, making it difficult to test some of the theories. Phillips’s first goal with the USAir wreckage was preservation. He needed to make sure the evidence wasn’t altered when his team moved it from the hill to the hangar and then shipped it to a lab for a detailed inspection.
A yellow construction crane was brought in to lift the big tail section onto a flatbed truck. Workers tied nylon straps around it and stood back as the diesel engine strained to pick it up. When Cox heard the engine groaning, he worried for a moment that the crane might drop the huge piece. But the tail lifted slowly into the air, and the workers used ropes to direct it onto the truck. Once it was secured, the truck drove to the decontamination station so it could be sprayed with the Clorox solution. The process reminded Cox of his planes’ being de-iced. The wreckage was covered with a tarp to hide it from the snooping eyes of the media, and then it was driven to the hangar.
Phillips and his group decided to send the key pieces to two labs on the West Coast. Most devices from the rudder and aileron systems would go to the Boeing Equipment Quality Analysis lab outside Seattle, which was regarded as one of the best places in the world for forensic analysis of a plane crash. The rudder power unit would go to a Parker Hannifin plant in Irvine, California, where the units were manufactured.
The rudder unit had been carefully cut from the tail of Ship 513. Its hydraulic lines were capped to keep the fluid inside. It was packed in a padded trunk and sealed with several layers of shipping tape. Phillips and other members of the group signed their names on the tape so it would act like police evidence tape. When Phillips got the box on the West Coast, he could check the tape to be sure that no one had opened the trunk and tampered with the evidence.
9. PIPSQUEAK
Hidden in the tail of every 737 was a hydraulic device that acted as the muscle to move the plane’s big rudder. It was about the size and shape of an upright vacuum cleaner and was known as the power control unit, or PCU.
Inside the PCU was an ingenious valve. It was shaped like a soda can and was strong enough to move the rudder when the plane was going 500 miles per hour. The gadget had an impossibly dull name—the dual concentric servo valve—but engineers used an amusing hand gesture to show how it worked. They curled the fingers of one hand to create a hole and then stuck their index finger in and out, as if they were demonstrating sex.
That was how it worked. When a pilot pushed on a rudder pedal, he moved a tube the size of a pencil in and out of the soda can. Holes in the tube allowed hydraulic fluid to squirt against a piston, which pushed the rudder to the right or left.
The 737 rudder valve was unique because it had two tubes, one inside the other. Both moved back and forth. That design was revolutionary in the arcane world of hydraulics. It allowed Boeing to use one power control unit instead of two, which helped it fit in the tight confines of the tail and trimmed at least fifty pounds off the final weight, saving airlines thousands of dollars in fuel over the life of a plane.
The valve-within-a-valve also had an important safety feature. If one tube jammed, the other one could still move and neutralize the rudder.
At least, that was how it was supposed to work.
Investigators had suspected a problem with the rudder valve in the Colorado Springs crash, but the evidence had been too sketchy. The valve was damaged and difficult to test. This one, however, was in perfect condition. So Phillips’s trip to the West Coast had the potential to wrap up the investigation quickly and neatly. All he needed was to find a tiny flaw, and the mystery would be solved.
But as he flew to Seattle, he wasn’t fantasizing about solving the case. He was being his usual cautious self, fretting about the
test plan for all of the flight controls—not just the rudder—to make sure he didn’t forget anything.
“We can’t just stay focused on the rudder,” he told his group when they gathered in a lab at Boeing’s big plant in Renton, Washington. There was mounting evidence pointing to the rudder, but Phillips did not want to exclude other possibilities and then discover he had missed something crucial. He said he would be open to any idea, any test. “No question is too trivial.”
The lab, in a giant gray building beside Interstate 405, about ten miles southeast of Seattle, resembled a high school science classroom, with drab yellow walls and lots of counter space for tests. The technicians who worked there were the unsung heroes at Boeing. They were a notch lower on the food chain than the engineers who got most of the glory, but they were just as important. They tested hundreds of airplane parts every year to make sure they were safe and durable. They could figure out why a window cracked, why an actuator leaked, how to improve a toilet seat.
They had lots of gadgets to help them examine broken parts—microscopes that let them see tiny imperfections in plastic and borescopes that could peek inside a device. The technicians were like a band of renegade nerds. For kicks, they once used the borescope to check out a guy’s ingrown nose hair.
Phillips’s group planned to test anything that could have played a role in the crash—the main rudder power unit and its backup, the standby PCU; the rudder trim actuator, which adjusted the rudder constantly to keep the plane flying straight; and the hydraulic devices that moved the ailerons. The wreckage arrived in big wooden crates, all sealed with tape, just as the rudder crate had been, to make sure no one tampered with the evidence. Everything reeked of Clorox.
Phillips, Cox, and the other team members met in a conference room to decide how each piece would be tested. Phillips insisted on a consensus for every step. They would not proceed until all team members—from the NTSB, the FAA, Boeing, ALPA, and the machinists union—agreed.
They methodically inspected and tested each item. Some parts were taken to another Boeing building and X-rayed to search for internal cracks. The team was looking for anything out of the ordinary—scrapes, leaks, dents, electrical fluctuations—that might explain what had gone wrong on Ship 513.
One theory about the crash was the possibility of “runaway trim” of the rudder. Normally, pilots could turn a dial to trim the rudder to adjust for crosswinds and keep the plane flying straight. But an electrical malfunction might have made the trim “run away,” sending the rudder swinging to one side. The trim actuator, about the size of a box of animal crackers, was badly dented from the crash. But one of the Boeing technicians figured out a way to hook a testing device to it. When he measured the amount of electricity that went in and out of the device, the electrical readings showed the trim was centered. Runaway trim had not caused the crash.
They examined dozens of other pieces the same way, taking photographs and carefully documenting the condition of each piece. On the hydraulic pressure gauge, a tiny indentation known as a “witness mark” provided valuable information. When Cox found the gauge at the crash site, it had one needle pointing to 3,000 pounds per square inch (psi), which showed that the “B” hydraulic system was operating normally when the plane crashed. Unfortunately, the “A” needle had broken off. But when the technicians put the gauge under a microscope, they could see a distinct mark indicating that the A needle had also been pointing to 3,000 psi. That meant both systems were working properly and they could rule out a hydraulic failure.
After scores of tests, they came up empty. Everything they tested was working fine when the plane struck the road. It was time to go to Irvine, California, and see what they could learn from the soda can valve.
At the time of the Hopewell crash, Boeing was number one in a dwindling group of airplane makers. Lockheed and Fokker were no longer building airliners, and McDonnell Douglas was struggling. Boeing’s primary competition was Airbus Industrie, a consortium of European companies based in Toulouse, France.
In Seattle, Boeing was the city’s largest private employer and a mighty economic force. You couldn’t drive far on a major highway without seeing the company’s familiar italic logo. The major plants were along East Marginal Way on the edge of Boeing Field; in Everett, a town north of Seattle where the wide-body jets were assembled; and south of Seattle in Renton, where the 737s and 757s were made.
Boeing sometimes lagged behind its competitors in developing new planes—a practice that earned the company the nickname the Lazy B—but it seemed to gamble at the right times. In that regard, the 737 was typical. When Boeing officials were debating whether to build the plane in the early 1960s, archrival Douglas was already several years ahead with the similar-sized DC-9. Boeing president William Allen was dubious about building the 737 because his chief competitor was so far ahead. He had struggled through the high costs and development problems of the three-engine 727 and was not sure the company could afford another painful birth so soon.
The 737 was designed for short-haul flights of five hundred to a thousand miles, serving airports that otherwise might not get jet service. Boeing planned to offer customers an optional gravel kit, so the plane could land on an unpaved runway. Early designs had the engines on the tail, as on the 727, but they were later moved to the wing because placing them there would allow seats for six more passengers.
By 1964 many Boeing executives felt that the 737 project was dead. But a handful of believers led by chief engineer Jack Steiner pushed to keep the project alive. They said costs could be reduced because the plane would be heavily based on the 727. The new plane would have many of the same parts as its sibling.
Designers tried to simplify every aspect of the 737 to save money and increase reliability. By combining two parts into one—such as the rudder’s unique valve-within-a-valve—the designers would have fewer things that could break. That was crucial if they were to achieve the ambitious goal of 99 percent reliability. They also simplified the cockpit and pilot workload so the plane could have a two-person crew instead of the costly three that were needed for the 727.
Steiner, an intense workaholic who wore Buddy Holly glasses, decided to sneak behind his boss’s back to convince the board of directors to support the 737 launch. It was easy for Steiner to reach board member Ned Skinner, because he lived nearby; Steiner simply walked over to Skinner’s house and gave him the sales pitch. Then Steiner tracked down three other board members and gave them the same message. The lobbying campaign worked. The board decided in 1965 to build the plane.
When the first 737 came off the assembly line in Renton in January 1967, it was christened by seventeen stewardesses with seventeen bottles of champagne, representing the number of airlines that had ordered the plane. When Allen spoke at the dedication, he noted that the small plane had been overshadowed by bigger jets being developed—the 747 and the supersonic transport. The 737 might be small, he said, but it would fill an important niche for shorter flights. “We expect the 737 to serve long and well on the air routes of the world.”
It took nearly twenty years for the plane to make a profit, but the 737 is now the best-selling jetliner in the history of commercial aviation. Boeing boasts that its 99.2 percent reliability is the best in the world. There are more than 3,800 of the planes in use, with 700 in the air at any given moment. The 737 is so successful that Boeing now makes the “Next Generation” models, with longer ranges, more-efficient wings, quieter engines, and more-modern cockpits.
The plane that crashed in Hopewell was a 737–300, a model that some pilots called the Quichewagon. The 300-series was the first to have a flight-management computer, which enabled pilots to enter complicated routes into the computer so the plane could virtually fly itself. The computer helped them find the most fuel-efficient altitudes, saving up to 7 percent in operating costs. Pilots who relied on the high-tech device were said to be less macho. They were the quiche eaters.
The 737 was dull but efficient,
with all the sex appeal of a four-door sedan. Other planes had mean-sounding nicknames like Mad Dog and Mega-Dog, but the 737’s monikers sounded downright wimpy: Fat Albert, Guppy, and Fat Little Ugly Fellow (FLUF). Boeing was famous around the world for its gigantic 747, the premier long-haul airplane. When a 747 took off from a runway, people stopped and watched the 400-ton machine climb into the sky. But the smaller 737 rarely stopped traffic, since so many of them were making so many flights. Watching a 737 take off was about as exciting as watching a minivan pull out of a garage.
The lackluster image of the 737 was largely a function of its plain-vanilla use. The big 747s flew twelve-hour missions on long, glamorous routes such as New York to Paris or Los Angeles to Honolulu. The pipsqueak 737s flew short hops like Baltimore to Buffalo or Detroit to Little Rock. The 737 was a relatively basic machine. New planes were “fly-by-wire,” with computers sending electrical signals to move the flight controls. But the 737 still relied on old-fashioned cables that ran beneath the cabin floor.
The plane had an excellent safety record, but so did virtually every other modern plane. The 737–300, -400, and -500 planes had nine “hull losses” (the industry euphemism for a crash), for an accident rate of 0.43 for every million flights. The 737’s rivals, the DC-9 and the Airbus A320, had higher rates, but the differences were insignificant. Crashes were rarely caused by the same mechanical problem. About 70 percent of crashes were blamed on pilots, and only 10 percent on malfunctions.
Yet two or more crashes caused by the same problem in a plane could destroy the plane’s image and make airlines stop buying it. A design flaw that led to crashes of the de Havilland Comet in the 1950s nearly doomed the plane. The same thing happened to the Lockheed Electra in the early 1960s after a series of accidents were blamed on engine vibration. The DC-10’s reputation deteriorated so much after two spectacular crashes in the 1970s that airlines began canceling orders.