Comair Crash: When Systems Fail

Aug. 28, 2006 — -- How on earth could two qualified airline pilots take off from the wrong runway?

The question, of course, is the one being asked nationwide as the search for a cause in the tragic crash of Comair 5191 ramps up in Lexington, Ky. The answer, well, that will surprise you: There is no single cause in this or any airline accident.

Air accidents (like most accidents in complex human systems) are caused by a long series of problems, mistakes, missed opportunities, and other glitches that linked together like a chain, set up the potential for a tragedy long before someone came along and added the final link.

The Comair crash is no exception.

There are, in fact, many things that can contribute to a professional airline pilot starting down the wrong ribbon of concrete.

Whether confused by a forest of sometimes conflicting runway signs, hurried along by pressures of schedule and other aircraft, or victimized by your own assumptions of where you are in a complex airport, most experienced pilots have done something similar at least a few times in their careers, me included.

In fact, I can recall three instances in 13,000 hours of professional flying when my airplane ended up on the wrong runway. Fortunately, in each instance, we (or the tower) caught it in time.

When it's as dark at an airport as it was on Sunday morning when Comair 5191 left its gate, the visual clues and markings on taxiways become significantly more difficult to read.

Runways and landmarks that are so easily seen during the day may literally disappear at night, leaving the pilots with a combination of their own internal compass supplemented by taxiway lights and whatever lighted signs loom at them carrying the numbers of runways and taxiways alike.

And Sunday, there may have been an added problem created by the airport work crews changing the markings leading toward the main runway, though it's much too early to know if that was a factor.

But it gets more complex.

While pilots taxi their departing airplanes toward the runways they expect to use, they're also working their way through a lengthy series of preflight steps and checklists to make certain everything is properly set for takeoff.

In addition, the pilot may be giving the co-pilot a briefing on the departure plan (which way to turn, altitude to fly, etc), while fielding radio calls and listening to other aircraft on the field. And, as is often the case in an early morning departure at a smaller airport, a clearance to take off may be issued early by the control tower before the aircraft has reached the runway area.

If anything close to that sequence happened Sunday morning, one of the many questions for National Transportation Safety Board investigators will be whether providing the takeoff clearance early (which is perfectly legal) could have created a bit of additional pressure on the pilots to get off the ground.

The question has already been asked why the air traffic controller wasn't monitoring the flight after issuing the takeoff clearance, but the system has never required that, and controllers have many heads-down duties in the tower as well.

True, every year many sharp controllers catch human errors in progress in time to radio a quick warning (and there is a prestigious recognition called the Archie League Award for such heroics), but baby-sitting every pilot's level of compliance with every clearance just isn't possible.

And then there's the actual layout of the Lexington airport. When there's just one runway, or two runways crossing in the middle with the ends far apart, the opportunity to get mixed up is minimal. But when two different runways begin from nearly the same spot, the opportunity for confusion becomes possible.

That does not mean Lexington is somehow designed wrong or is unsafe, but it means there's an added potential for trouble.

At Lexington, the Comair crew was cleared to use the 7,000-foot runway that was long enough for a safe ground roll and liftoff. That's Runway 22 (add a zero and you have the magnetic compass heading, 220).

But the starting point for Runway 22 -- the one the plane was supposed to be on -- lies approximately 900 feet beyond the start point for Runway 26, which is half the length of 22.

In addition, there is a small jog to the left before reaching the end of Runway 26, just like there is before Runway 22, and even someone familiar with the airfield might be fooled.

Now, put these factors together and you begin to get a rough idea of why a simple mistake turns out to be so complex.

Add in a possible half-remembered reference to runway lights being out, and the opportunity to fixate on the first runway you come to as the correct one (even though unlighted) and it all becomes more explicable.

Then, too, there's the reality that you can't necessarily see the runway lights of the intersecting runway (the one they really wanted, Runway 22) from the end of 26, and thus the only thing that could have absolutely prevented a disastrous plunge down a too-short runway is a procedure the FAA has never mandated: A final crosscheck by the pilot and co-pilot before pushing up the power.

Pilots are human, and humans -- once we decide something is true -- tend to discount subtle clues to the contrary.

There is little doubt that both pilots Sunday morning were dumbstruck as they accelerated down what they thought was a 7,000-foot runway and suddenly found themselves facing the end of the concrete while still far below flying speed.

There would have been only two options: Try to stop and guarantee a major crash, or try to fly, and maybe, just maybe, be able to lift away from the nightmare.

Most of us would have chosen to try to fly, because that's where pilots live.

In this case, the Comair jet may have been pulled up at too slow a speed, in which case it would have been able to lift off the ground, riding a cushion of compressed air referred to as "ground effect."

But it would not have been able to climb above that cushion -- which is, at best, half the wingspan -- and skimming above the ground no more than 20 feet high at more than 100 miles per hour with no ability to get over the trees and farm buildings ahead, the physics of the situation would have been impossible to overcome.

In the aftermath of the deadly postcrash fire, one of the NTSB's most important duties will be to answer the questions of why, after 30 years of the FAA's mandating safer, slow-to-burn materials in airliners, the jet cabin burst into flames so intense that only one person, co-pilot James M. Polehinke, survived.

The accident at Lexington ended almost five years without a major airline accident in the United States, but the rest of the story is that the five-year safety streak was an unprecedented result of finally learning to anticipate human mistakes and insulate against them before they metastasize into an accident.

In other words, it's never enough to just order people not to fail. We have to build the system to safely absorb human screwups, and for the most part our success rate in doing that has been spectacular.

In the 15 to 25 major contributing factors that the NTSB will eventually discuss in issuing a final report on Comair 5191 (in perhaps a year or more), every one will need to be addressed and solved.

But in this case, we're very lucky, because a single, universal procedure can be added that will prevent this type of accident from ever occurring again -- a procedure prohibiting takeoff unless the nonflying pilot repeats the runway the flight is cleared to use, and the flying pilot reads in full the compass heading of the aircraft as it sits aligned with the runway.