In the cockpit of a Boeing 747 jumbo jet approaching Madrid in the fall of 1983, a recorded voice sounded urgently from the flight console: "Terrain! Terrain! Pull up! Pull up!" Capt. Tulio Hernandez ignored the warning, trusting his instincts over what he apparently perceived as a faulty ground proximity alarm.
Instead, to the English-speaking robotic voice, Hernandez sighed: "Bueno, Bueno. " There was no urgency. He disconnected the autopilot and casually said again: "Bueno. "
His last words were recorded on cockpit tapes. Moments later, the Avianca jetliner scraped one hilltop and began disintegrating as it slammed into a second and then a third, killing 181 people.
Official cause of the accident: human error.
It is a common story. "People are the more slowly improving side of the safety system," said Delmar Fadden, a systems engineering executive with Boeing Co.
Last month in Los Angeles, an air traffic controller mistakenly directed a jetliner to land on the same runway where a commuter plane was awaiting takeoff clearance. In Detroit last December, a jetliner crew lost in the fog taxied into the path of another plane that was accelerating for takeoff. In New York in 1989, a "confused and inexperienced" crew mishandled an aborted takeoff and the plane plunged off the end of a runway into the East River. Three years ago in Denver, a poorly skilled young co-pilot botched a takeoff and plowed into the runway just seconds after lifting off in a snowstorm.
Such accidents have been tragic exceptions to an otherwise improving safety record in aviation. The accident rate reached an all-time low in the 1980s--about 2.3 serious crashes per million commercial jetliner flights worldwide, according to Boeing. In the United States, the rate was one in nearly 2 million flights.
But even as technology and engineering improvements make the machinery of air travel safer, one thing remains constant: People make mistakes.
Boeing research into 30 years of commercial airline accidents found that pilot error was the primary cause of more than 72% of them (10.8% were blamed on aircraft failure and 5% on weather).
Particularly alarming to the aviation industry is a Boeing prediction that, in the next decade, the number of major commercial jetliner crashes is likely to rise more than 40%--to a yearly average of 20 to 25 worldwide. The reason: The sheer number of jetliner flights is increasing faster than the rate of airline accidents is dropping. That could mean a major plane crash every two to three weeks by the year 2005.
"We can't stand that many accidents on into the future," said Boeing's Earl F. Weener, chief engineer for airworthiness, reliability, maintainability and safety. He warned that such statistics could undermine public confidence in air travel.
"The public doesn't see the accident rate improving. They see headlines about accidents," Weener said.
Both Boeing and the Federal Aviation Administration have launched campaigns to reduce the human factor in aviation accidents.
Boeing officials are advocating use of improved ground-proximity warning systems in all commercial cockpits, and of instrument-landing systems (which send out electronic guidance beams) on all runways used by commercial jetliners, and routine review of flight-recorder data--the so-called "black box"--to analyze crew members' job performance.
The FAA recently released a 10-year master plan for coordinating research into what it calls the "human factors" of aviation safety. H. Clayton Foushee, the FAA's chief scientific adviser for human factors research, calls exploring human error "the last frontier of aviation safety."
"Because our mistakes are more catastrophic, the aviation world has to be a leader in studying human performance," Foushee said.
The Aviation Safety Reporting System database, maintained for the FAA by the National Aeronautics and Space Administration, is a veritable gold mine of human-error examples. It is a collection of tens of thousands of reports filed voluntarily by pilots and air traffic controllers about incidents in which they committed a vast range of procedural blunders. By reporting on themselves, they can gain immunity from federal sanctions against their licenses.
A Times analysis of those reports, obtained through a Freedom of Information Act lawsuit with the names of pilots and airlines removed, found that potentially hazardous operational errors are made many times every day--commonly due to such conditions as fatigue, miscommunication, inexperience, distractions and carelessness.
--A co-pilot reported that during a flight between Kansas City and Los Angeles he had "dozed off" while steering a gentle 3-degree bank for a course correction. He was still turning, substantially off course, when he was awakened by a controller who radioed to ask where he was going. The call also woke up the captain, who had been asleep in the left seat.
--A captain wrote that as he approached the Los Angeles Basin one night, he thought he heard the controller clear him to descend to 5,000 feet. He misheard. The clearance was to an altitude of 7,000 feet. As the jetliner got down to 5,700 feet, the controller ordered an immediate climb. Simultaneously, the cockpit ground-proximity alarm sounded. The plane pulled up and cleared a ridge of mountains.
--An inexperienced jetliner pilot, flying into Burbank Airport for the third time in his career, descended to 500 feet over the San Fernando Valley, landing gear down, before he realized he was about to land at a small Pacoima airfield suitable only for light private planes.
--A co-pilot absent-mindedly wedged his ballpoint pen into a slot on a control panel, presumably so it would be within easy reach. He inadvertently activated a fuel shut-off valve, which caused the No. 3 engine to shut down. During an emergency descent the pen was removed and the engine was restarted.
Foushee said the FAA monitors such reports, filed at the rate of about 2,000 a month, "for ideas about where the problems are."
Clearly a leading complaint among the volumes of reports is fatigue, a subject that has been getting increased research attention recently.
Not only do pilots sometimes fall asleep at the controls, their physical performance and judgment also are hampered by fatigue. Among the NASA reports are numerous accounts of tired pilots flying at the wrong altitude or on the wrong course or approaching the wrong airport. Some of these mistakes have resulted in close calls.
In one case, a co-pilot on the fifth leg of a six-hop day that had started with an early wake-up call heard the controller approve a climb to 31,000 feet, and did so. But when the captain said, "Did you see that?" they realized the controller had been talking to another plane, which just missed them going by the other way.
Another jetliner with a weary crew, bound for Boston from Washington, D.C., was discovered to be 12 miles off course over New York. One crew that climbed to the wrong altitude departing Las Vegas blamed the failure on their hotel. "The cockpit crew did not get a good night's sleep due to other patrons of the hotel making loud noises," the report said.
Fatigue also was blamed for pilots, even the most experienced ones, landing on the wrong runways and without clearance from control towers at Los Angeles and other crowded airports.
"It is the first time it has happened to me in over 25 years of flying," an embarrassed captain wrote after he landed a jetliner at busy Hartsfield Airport in Atlanta without clearance.
In other fatigue-related incidents, a co-pilot accidentally shut down one engine on a trans-Atlantic flight to Spain, and another flight engineer miscalculated fuel requirements for a flight to Tokyo from Anchorage, forcing an unexpected refueling stop en route.
"After doing three trans-Pacific (flights) in seven days, I can't even add up a few simple numbers or spell my name," one crew member noted.
Noting the volume of "significant operational errors" related to fatigue that were showing up in the aviation safety reports, NASA and FAA researchers conducted studies leading to a recommendation that planned cockpit naps be approved for one crew member at a time on long flights.
R. Curtis Graeber, a pilot-fatigue expert with NASA before he joined Boeing this year, said rest must be scheduled in part because pilots, like most other people, "are very poor at predicting how tired they are."
Crew members of a flight from Chicago discovered how tired they were when they tried to land at Providence, R. I. They had begun their workday flying all night from Mexico City, then flying for two hours into a bright morning sun.
First, they landed hard and bounced back into the air. The tired captain, thinking he was on the runway, activated the reverse thrusters while still slightly airborne. After the second jarring bounce, "we knew we were on the ground," noted the co-pilot.
Then, taxiing to the gate, the crew "almost missed the turn." The co-pilot wrote: "These situations are almost humorous, but if you realize they were caused by crew fatigue, you realize the possibility for a more serious mishap is there."
Fatigue is common also among commuter pilots, according to experts and the NASA reports. Although they cannot be scheduled to fly more than eight hours in a 24-hour period, it is not unusual for crews to work 12- to 14-hour days, counting preflight briefings, weather delays and other non-flight duties, while making several takeoffs and landings during a continuous shift and working a number of consecutive days.
"That can be more tiring than a long-haul flight," Graeber said.
One commuter captain, who mistakenly taxied onto an active runway as another plane was about to land at West Palm Beach, Fla., a year ago, complained that he was tired and distracted dealing with an inexperienced co-pilot.
The pilot said his company "consistently schedules us for long duty hours, several days in a row. This leads us to flying fatigued and the consequent probability for unsafe operations." The captain warned: "With our company expanding into the Miami hub, our situation is an accident waiting to happen."
Federal safety officials acknowledge that fatigue is "one of the most recurrent, chronic gripes of airline pilots." Pilots represented by labor unions tend to have the most protection from harsh schedules, but the FAA in recent years has established more stringent rest rules for most commercial airlines--with the exception of commuter lines that have been less affected.
"There is a movement afoot to look at bringing the commuter operators on a par with the major carriers," said a National Transportation Safety Board official.
Graeber said that most accidents happen "because wrong decisions were made." Because fatigue can have a profound effect both on physical performance and judgment, Graeber advocates research toward improving flight crew "alertness management."
The Boeing accident survey also found that about 70% of serious accidents happened during takeoff or landing--a period of about six minutes in an average flight that is most demanding on pilots and controllers.
Some ways to improve safety in those critical phases are subtle. For example, in its new jetliners Boeing reduced the number of cockpit alarms that can go off during the first 400 feet of takeoff climb, a time when little short of a lost engine is worth distracting a pilot's attention from the immediate need to reach a safe altitude.
Boeing officials also advocate routine use of flight data recorders to check on the performances of flight crews. Usually, the "black box" is reviewed only after an accident. But since it records every technical detail of the flight, from the power of each engine to the angle of the flaps from takeoff to landing, Boeing believes the tapes should be used as a tool to tailor pilot training.
Members of U.S. pilot unions are generally suspicious that their companies might use the data against them, but the practice has long been common in Europe.
Other safety improvements involve technology--such as advanced radar technology that puts in the cockpit ground-proximity warning systems, traffic collision avoidance systems, weather monitors, computer terminals and detailed maps of the terrain below.
Human experience shows that technology isn't magic, however.
The first ground warning systems issued so many false alarms that crews routinely ignored them, as did the Avianca crew in Madrid, or regarded them as such a distraction that they disabled them. Nonetheless, the FAA required all U.S.-flag jetliners to have ground warning devices.
"Within the first few months that the initial generation was installed, the pilot community thought it was a piece of junk," said Fadden, in charge of Boeing's flight deck systems engineering. "Pilots have to believe that the warning is real or it's useless--or worse."
According to the Boeing study, in four of five instances during 1989, when "perfectly good airplanes were flown into the ground inadvertently," the planes had the unreliable first-generation devices on board.
Boeing's Weener said today's systems are substantially more reliable and can help prevent one of the most common types of accidents. He said that since 1975, there have been 66 accidents in which planes were flown into the ground, and that about 2,300 of 5,700 fatalities worldwide occurred in such crashes.
"This is the ultimate crew-caused accident, because most of these can be prevented," Weener said. He maintains that reliance on ground-proximity warning devices and improved pilot training could virtually eliminate one type of major accident.
But, he said: "To our consternation, not all airlines train their pilots to pull up first and ask questions later."
The history of false alarms with ground warning systems makes some safety experts skeptical of on-board collision avoidance radar, which some regard as "the latest technological gadget."
An advanced traffic collision avoidance system (TCAS) can cost about $150,000 to install (contrasted with about $30,000 for a ground warning device). And in busy airspace like Southern California's, a device may issue almost continual alerts. "You don't want people getting inured to warnings," a Boeing official said.
However, some of the earliest enthusiasm for the system comes from pilots flying into the congested skies of the Los Angeles area--what one pilot told NASA was "like dropping into a bucket of honey alongside a swarm of bees."
In one case, a jetliner pilot said TCAS helped him avoid colliding with a small plane on final approach to John Wayne Airport in Orange County. Air traffic controllers failed to report the other traffic until it had passed the inbound jet. The pilot called TCAS "extremely reliable and (not prone) to give false warnings."
Alaska Airlines, which flies into all five commercial airports in Southern California, is installing the system on its MD-80s, half of which already are equipped. "We believe in it," said Tom Cufley, assistant chief pilot. "It certainly enhances safety."
But Earl Wiener of the University of Miami, a consultant to NASA and airline carriers on human factors in aviation, questions whether TCAS is worth the expense. "It's a way to prevent something that almost never happens," he said, noting the rarity of midair collisions.
"Pilots think their biggest hazard is a midair collision. It is not. Other things are more dangerous," the Miami professor of management science said.
Some safety experts worry about future over-reliance on technology, about turning pilots into what the FAA's Foushee calls "systems monitors." He sees a philosophical dilemma looming in the future of aviation as technology takes more and more of the flying tasks away from the "stick and rudder people." He questions whether human beings make efficient backups to computers.
"It's not so much a question today, with the current generation of aircraft still primarily controlled by humans, but it is clear we're going somewhere down an evolutionary path," Foushee said. "We're moving in the direction of increasing automation and we don't understand enough about what the implications are."
There could be clues already in the aviation safety reports. A flight engineer who relied on a faulty computer read-out to determine fuel load for a trans-Pacific flight wrote that tired crew members are more inclined to accept computer data without double-checking.
"With computers doing more and more for us, I foresee more dependence on computers and less on the human factor," the engineer said in his report. "We have to keep the computer and human factors equal or we are going to lose the ability to even check the computers.
"When accident reports read 'Computer Error,' you (will) know we went too far."
This story was reported by Times staff writers William C. Rempel, Sam Fulwood III and Richard O'Reilly. It was written by Rempel.
Times statistical analyst Maureen Lyons contributed to this story.
AIRCRAFT ACCIDENTS: WHERE AND WHEN
Worldwide commercial jet fleet, 1980 to 1989
Percent of accidents
Load, taxi and unload--3.3%
Percent of flight time
Percent of accidents:
Takeoff & initial climb--28.7%
Final approach & landing--41.1%
Percent of flight time
Takeoff & initial climb--2%
Final approach & landing--4%
Percentage based on average flight duration of 100 minutes.
Source: Boeing Aircraft Co.