Ask the pilot

When airplanes collide, who is responsible? Are we doing enough to prevent such disasters?

Published August 9, 2002 7:30PM (EDT)

What happened when those two planes collided in midair in Europe last month? Were the pilots or the system at fault?

On July 1, 1986, as a young private pilot, I was nearly killed in a midair collision over Nantucket Sound. I simply looked up, and there in front of me was the other plane. It was a twin-engine propeller plane, a four-seater not much bigger than mine, coming head-on. I did not react. There wasn't enough time; my brain never processed any left, right, up or down resolution. Only luck -- the slightest difference in our respective altitudes and alignment -- prevented our two machines from hitting.

The whole event, from when I saw the airplane until it was safely behind me, lasted, probably, two seconds. Our closure rate was about 300 miles per hour. And this marks the only time in more than 15 years of flying that I've been close enough to another aircraft to clearly make out the shape and colors of a human being inside -- in this case the lone pilot sitting in his seat, buzzing past in a blur of blue-and-white aluminum.

What's ironic about distinguishing the other pilot is how often I hear airline passengers claiming a similar feat. "We were landing at O'Hare," somebody might recall excitedly. "Another plane was right next to us, and it was so close you could see the people inside!"

While landing on parallel runways or crossing perpendicularly at high altitudes, airliners often pass within close proximity. But close proximity, in the meticulously orchestrated doings of air traffic control, is measured in miles, or thousands of feet. To an anxious flier with emotions revved by adrenaline, distances and sensations are distorted. Trust me, you have never been anywhere near close enough to another plane to see faces through those small, oval windows.

The same cannot be said, however, for the 71 people aboard two jets that collided on July 1 over the Swiss-German border. On this evening -- the 16-year anniversary of my own flirtation with mortality -- A Boeing 757 freighter flying under contract for DHL Worldwide Express, and a Russian-built passenger jet operated Bashkirian Airlines, smashed into each other at 35,000 feet.

There are various frightening buzzwords commonly invoked by passengers, most of whom know very little of what these situations truly entail. Wind shear. Turbulence. Another one is "near-miss." The idea of airplanes coming dangerously close to each other, even hitting each other in flight, is shocking -- a virtual rape of the sanctity of the sky. How does it happen? How often? To address these concerns, both the philosophy of safety and the technical workings of the airways must be examined.

For starters, we should learn to accept the reality of error. Passengers are uncomfortable with the inherent perils of flying, and tend to assume a delusional "zero tolerance" notion of air safety. In fact, no flight is ever a perfect one, and minor, ultimately harmless errors occur with a fair amount of routine. It has always been this way, and shall remain so, just as the odds of your plane going down, no matter how bad a day a pilot or air traffic controller might be having, shall remain ridiculously in your favor.

In keeping with the laws of fallibility, airplanes do, on occasion, breach the confines of one another's protective space. Such incidents usually involve brief transgressions, a tangential grazing of restricted territory. Sometimes a crew has misread an instruction. A pilot may turn to the wrong heading or assume an incorrect altitude. Sometimes the controller issues an incorrect command. Almost always, however, the mistake is caught. Safeguards are in place for just these sorts of trespasses. Read-backs of headings and altitudes must be verified. Controllers work in pairs. Alarms are cocked to sound if flights venture too close to one another. And so on.

But then, every once in a while, something goes terribly wrong.

Almost every air crash seems to involves not one, but two or more unusual breakdowns of technology and/or procedure. Investigators' final reports are full of uncanny coincidences that, seemingly against all odds, came to be in the moments before an accident. To ask, "Well, what are the chances of that?" is to invite some depressing retrospective amazement.

In the case of the July 1 crash, the coincidences included an understaffed control sector, a disabled alarm system and even a malfunctioning phone line. But to understand the incident and all it will inevitably embroil, we should first take a look at the workings of the air traffic control system.

In the case of my near-miss in 1986, we were flying under visual flight rules (VFR), using nothing more elaborate than what a flight instructor might call "see and avoid" technique. Keeping clear of other airplanes, in the frequently down-home, low-tech world of general aviation, is not a lot different, conceptually, from what's exercised on the highway. But airline flying is a another story. In this realm, flights operate under "positive control." Simply put, this means a flight is not only in the careful care of the flight crew, but also under the competent watching eyes of air traffic control (ATC) as well.

ATC is hardly just the control tower -- that tall concrete structure rising above the terminals with its spinning radar and cluster of antennae, at the top of which you imagine a group of men standing in a windowed room saying things like, "cleared for takeoff." ATC is a collective term for a host of control facilities. Towers are responsible only for ground movements and planes within the immediate area and below certain altitudes. Tower staff clear flights to taxi, takeoff and land. But away from the airport, flights are watched over by other, often remotely located staff.

Let's follow a flight from start to finish. In the meantime, keep in mind that airplanes use something known as a transponder (two or more are usually on board) to detail whereabouts, speed and altitude back to ATC radar. At some airports, radar also can be used to follow planes along foggy taxiways. Though it happens infrequently, a flight can operate gate-to-gate without a single controller ever once having the plane in actual, physical sight.

Departing from New York, bound for San Francisco, a flight will obtain weather and flight-plan clearances either directly over the radio or, more frequently, via computer. (Any takeoff delays also can be relayed this way.) When ready to taxi, the crew will call for "pushback" clearance and will next receive taxi instructions. Getting to the runway can involve a half-dozen radio frequency changes and various conversations with the likes of "clearance delivery," "gate control," "ground control" and possibly others. These may or may not be physically located in the tower, but are always somewhere at the airport. Finally a flight will be cleared for takeoff by the control tower itself.

Only moments after liftoff we are handed over to "departure control," which follows the plane on a radar screen issuing various turns, altitudes and so forth as the aircraft is sequenced into the higher altitude route structure. During a single interaction with departure control, a flight can progress through several different sectors, each maintained by a separate controller.

Once at higher altitude we are guided by a series of air route traffic control centers (ARTCC), commonly called "Center" by crews and controllers. Boston Center, New York Center, etc. Centers maintain huge swaths of sky, rendering them somewhat irrelevant to their namesake identifiers, and all the airliners within. Again, these are broken down into various sectors, each in the hands of individual controllers. Often, ARTCC facilities are located in secure buildings far from an airport. The Boston Center, for instance, in charge of airspace extending all the way to the Canadian Maritimes, resides in a building in Nashua, New Hampshire.

Flying from Center to Center across the country, the crew might make upwards of 10 or more radio frequency changes, in touch with a different person each time. Pilots may be asked to change altitudes or take a slightly different route from what the flight plan originally called for. While controllers do not tell pilots how to get from city to city, amendments to altitudes and routings are very common.

Descending into San Francisco, the above routine happens more or less in reverse. The flight is given over to "approach control," then to the control tower for landing clearance, followed by another rigmarole with ground personnel before docking safely (and, of course, on time) at the gate.

Following the midair crash over Europe, one Associated Press report stated, "There was only one controller in the Zurich tower at the time," setting off all sorts of anachronistic imagery. Yet, as just illustrated, the control tower at Zurich's international airport had nothing to do with the accident. The flights were in the upper, en-route portions of airspace. (Terminology changes somewhat when we leave the U.S., but the basic ATC concepts remain the same.) In truth, only one controller was monitoring that specific sector of sky.

That the jets were left on a collision course is unsettling, as is the disabling of the Swiss controllers' crash warning system and the missing second controller. But myths and hysteria over what some perceive to be an old-fashioned and dangerously overburdened system need first be reckoned with. The image of a lone man sitting in a tower with a microphone is foolishly misleading.

Independent from the air-to-ground link with ATC, airliners today also carry onboard technology to protect them from collisions. Linked into the transponder systems, TCAS (traffic collision avoidance system, pronounced "tea-cass" in an almost folksy slang) gives pilots a graphic representation of nearby aircraft. TCAS will issue progressively ominous oral and visual commands (to climb or descend) once certain thresholds of distance and altitudes are crossed. If two aircraft are erroneously flying toward each other, their TCAS units work together, issuing "CLIMB!" instructions to one, and "DESCEND!" to the other.

On July 1, for undetermined reasons, the Bashkirian Airlines plane, a Tupolev TU-154 (something of a 727 lookalike), was set on a collision course with the DHL 757. Not an imminently calamitous situation just yet. The Swiss controller below eventually realized this and issued a command, twice, for the Bashkirian crew to descend.

Meanwhile, both airliners' TCAS systems had correctly interpreted the impending hazard, and issued their own instructions. It told DHL to descend, and Bashkirian to climb. However, the Bashkirian crew disregarded the TCAS alarm, and chose instead to descend, in compliance with the controllers' request. Suddenly, both planes were descending and were still on a collision course.

Standard procedure is that a TCAS command override any from ATC. This protocol seems not to have been obeyed, and the two jets struck each other killing 71 people, a large group of children among them.

"A TCAS system makes no sense if it's overruled," said Herbert Schmell, a spokesman for the Swiss national airline (now called, well, Swiss, after the bankruptcy and dissolution of Swissair.) "We train for that," added Georg Fongern of the German pilots union, arguing that TCAS always be given top priority. True. Others have chimed in to remind us the planes should never have been allowed close enough to activate the warnings in the first place. Also true.

No fewer than five breakdowns came into play. 1) the Swiss controller not realizing the collision hazard in time. 2) the absence of the backup controller. 3) a disabled collision warning system at the facility. 4) malfunctioning phone lines that kept a German controller from notifying his Swiss counterpart of the hazard 5) the Bashkirian crew ignoring its TCAS command.

To wit: What are the chances of that?

For passengers, any night flight or bad-weather approach can be disconcerting. From the cabin, a plane may seem to be groping helplessly through darkness or drizzle. How do pilots know where the other planes are? Aren't the skies terribly overcrowded? A Salon reader recently voiced concern over landing in the fog at Heathrow: "Hundreds of planes following one another through the mist in a great spiral? Isn't it dangerous?"

In a word, no. While the spiral analogy isn't entirely far-fetched, it lends itself to a vision of closely-spaced airliners chasing each other blindly through the mist. In reality, the queuing of airliners for a landing through opaque skies is quite routine, and with as much as five miles separating them, "crowded" is something other than the wingtip-to-wingtip swarm you might be envisioning. Planes are monitored the entire way, choreographed along courses and led, ultimately, to an "instrument approach," whereby a plane is led by onboard instrumentation, both laterally and vertically, to the threshold of the runway with virtually unfailing accuracy.

But is it just a matter of time before a collision similar to the one in early July occurs on American soil? Maybe, but "matters of time" aren't always a fair way of evaluating such potentials. Airplanes occasionally crash and will continue to do so, for any number of causes. Can we improve our chances? Sure, and we must do so, but to suggest needed improvements is not to imply a situation rife with impending doom.

In 1978, a Pacific Southwest Airlines (PSA) 727 crashed while preparing to land at San Diego after colliding with another aircraft, killing all 135 people aboard the jet and seven more on the ground. The weather was excellent -- a cloudless morning with more than 10 miles visibility. The second aircraft was not another airliner, but a four-seat Cessna flown by a private pilot. He and his flight instructor also died.

Then, in August of 1986 an Aeromexico DC-9 crashed on approach to Los Angeles on a similarly clear morning. Sixty-four passengers and crew, as well as 15 residents below, perished when the jet plunged into the L.A. suburb of Cerritos. The other plane? A four-seat Piper flown by a private pilot. The Piper had strayed, without permission, into LAX's restricted airspace.

Granted, an off-course private plane should not cause a crash. Indeed, both small aircraft above, whether or not they belonged where they were, were under ATC radar surveillance up to their points of impact. And in both cases ATC procedures were deemed inadequate by the National Transportation Safety Board following post-crash investigations. But these high-profile crashes helped usher in the use of TCAS and various ATC system enhancements. Thus, rather than a harbinger of even more deadly incidents to come, these events can be interpreted as the price already paid for upgrades now in place.

With all the fuss over congestion around major terminals, it's interesting that the most infamous collisions over the U.S. involved private planes venturing into the realm of airliners. And as you can see, bad weather does not necessarily increase the likelihood of a disaster, and in fact may improve things by limiting the number of private planes operating under visual flight rules.

In addition to what happened on July 1, the three most notorious collisions involving passenger aircraft all took place overseas. In September of 1976 two airliners, one of them in the colors of British Airways, hit above Zagreb, Yugoslavia. Improper ATC procedures were to blame. Had TCAS been available at the time, it could have prevented the crash. In 1996 a Russian-built cargo jet struck a Saudia 747 over Delhi, killing 349 people. Another TCAS-preventable catastrophe.

And while the Delhi disaster stands as the third-worst crash ever in terms of fatalities, No. 1 on that list belongs to the collision of two 747s at Tenerife, in the Canary Islands, in 1977. Ironically, this accident was a runway collision in which neither plane had yet left the ground, and many experts will warn that the greater danger is not overhead but on the runways and taxiways of our busiest airports. At Chicago's O'Hare airport, a world-class calamity was narrowly avoided after an Air China 747 misunderstood instructions and ambled onto an active runway and directly into the path of a Korean Air 747, which leapfrogged over the top of it.

Close calls like the one at O'Hare notwithstanding, our record since 1986 is a good one. However much of our fortune is a debt to fate, it nonetheless stands as a testament to the reliability and safety of the ATC system, maligned as it is.

Ground radar, discussed earlier, is one way to avoid a second Tenerife. TCAS, meanwhile, can be extremely helpful to crews. And while the purpose of TCAS is a defensive one -- a backup in the event of things gone wrong -- its value cannot be overstated. (And while more systemic changes are its priority, the FAA deserves a black star for dragging its feet on the mandating of TCAS for cargo planes. In Europe TCAS was a requirement well before the FAA began any serious talk of our own freighters following suit, and most are not yet equipped.)

With respect to what happened over Europe on July 1, to indict the Bashkirian crew seems an easy call, though its predicament was one most pilots hope to avoid. The controller, suddenly noticing the danger, had barked an urgent command to descend. TCAS, meanwhile, demanded a climb. He shouldn't have been at this crossroads had ATC done its job correctly, but his duty was to follow the blares of TCAS.

The captain's response was incorrect, but not entirely surprising. "He knows the air traffic controller is telling him to descend, and he knows the TCAS system is telling him to climb, and that is a difficult spot to be in," said Peter Quaint-Mere, director of a consulting firm that specializes in airport safety.

Thus we see that the air traffic control is not, and never will be, reliant fully on either technology or human factors, but a well-managed combination.

Are the airways and airports congested? Yes. Are enhancements to an outdated and understaffed system needed, both here and abroad? Yes. In the aftermath of July 1, Swiss authorities acknowledged understaffing of its facilities (a privately operated ATC system, it should be noted) and announced a 20 percent cutback in the number of overflights.

In the meantime, though, do we put ourselves in grave danger each time we fly? No, and you may dispense with images of chain-smoking men bathed in the green glow of a radar screen, and captains lurching through the overcast with crossed fingers. It's a bit more advanced than that, thank goodness.

By Patrick Smith

Patrick Smith is an airline pilot.

MORE FROM Patrick Smith

Related Topics ------------------------------------------

Air Travel Ask The Pilot Business