What happens when an air traffic controller is asleep?

Friends have been asking me what the practical implications are of air traffic controllers falling asleep during late-night shifts. News reports make it seem as though the controllers’ primary job is “guiding pilots” and therefore, without a Tower controller, the two airline pilots in their state-of-the-art plane wouldn’t be able to find their way to the airport/runway.

In fact, if properly programmed, a modern airliner can more or less fly itself from runway to runway. The vast majority of airports in the U.S. don’t have control towers. Of the towered airports, most have a part-time tower, e.g., from 7 am to 11 pm at our local (very busy) Hanscom Field. If the airport is non-towered to begin with or the tower is closed, the standard procedure is for an arriving airplane to broadcast its intentions on the common traffic advisory frequency (CTAF), which is typically the same as the Tower frequency when the controllers are off-duty. A broadcast might take the form “Hanscom Traffic, Cirrus 5 miles south, entering left downwind Runway 29. Hanscom.” If an airplane were taking off, the Cirrus pilot would hear “Hanscom Traffic, Baron departing Runway 5 straight out. Hanscom” and understand that a small twin-engine plane was about to take off and fly northeast towards Maine. http://www.airnav.com/airport/KDED is an example of an airport with 213 operations per day and no tower (that is probably one of the busiest non-towered airports in the U.S. and, in fact, a tower is being put in).

If pilots can find the airport, choose a runway depending on the wind, announce their own positions, and see and avoid each other, what purpose does a Tower controller serve? At a busy airport, the controller figures out the best way to sequence and separate airplanes so that they can make maximum use of the available runways while minimizing hazards such as wake turbulence. It can be a very demanding job but, aside from helping the occasional student pilot who gets lost, has little to do with “guiding” pilots.

What if it is 3 a.m. and traffic is so light that the controller has fallen asleep? Pilots in that situation are so few and far between that they should not have any difficulty in separating themselves. What if the weather is cloudy and the airplanes are flying on instruments? This is a very common situation and it is handled by the Approach controllers issuing only one instrument approach clearance at a time. Until they hear that the cleared airplane is on the ground or has “gone missed” (elected to climb out and try again at that airport or elsewhere; a standard procedure if the weather does not meet minimum requirements or if anything unusual occurs during the approach), any other airplane that wishes to use the airport will be forced to wait (vectors or holding up in the sky; parked short of the runway on the ground).

The main challenge in the reported situations is that the pilots expected the controllers be on duty and did not expect to have to follow the established procedures for non-towered airports. At the end of a four-day trip sleeping 5-6 hours/night in Hilton Garden Inns, the last thing that an airline crew needs is to be confronted with the unexpected (i.e., not everyone can be Captain Sully!). The government has decided to address the situation by adding more personnel (at an average cost to taxpayers of at least $250,000 per year per controller, including pension and health care benefits; not clear why the “keep the first person awake” person needs to be a fully trained controller, but apparently the plan is to use the highest cost and qualified people available) but it could also have been addressed by changing the overnight procedures so that pilots would more readily revert to the non-towered procedures if unable to contact the Tower. The only existing procedure that currently applies is “loss of radio communications” in which case a pilot in instrument conditions is expected to continue the flight and land at approximately the time expected.

How would I solve the problem? Start by purchasing a treadmill for the overnight controller so that he or she could be moving at a slow walking pace rather than sitting/slumping at a desk. Maybe add some productive work that the controller could do while waiting 15-30 minutes between radio calls. Finally hire an $8/hour intern who wanted to learn about air traffic control to assist with the overnight shift (an intern would ask a lot more questions than a fully trained controller and therefore be a greater aid to staying awake). Hiring a second controller to handle a situation that arises because there is only enough work to keep 1/10th of a person busy does not seem like a wise use of taxpayer funds.

[I have some personal experience with this. For example, I landed a regional jet at the Burlington, Vermont airport after the Tower was closed as scheduled. I don’t think that our flight was late; it was simply that our schedule and the Tower’s schedule did not overlap. It felt a little strange that we were clicking the microphone 5 times (to turn on the runway and taxiway lights for 10 minutes) and making calls to “Burlington Traffic” while traveling at 230 mph with 50 passengers in the back, but we managed to land and find our way to the gate nonetheless. A corporate jet was about 3 minutes ahead of us. I don’t think anyone landed between the time that we touched down and the time that we shut off the radios.]

29 thoughts on “What happens when an air traffic controller is asleep?

  1. Have a webcam watching each controller. For 200k/year I will personally watch multiple feeds (thus saving much money) and press a button activating a klaxon to be installed with the webcam. For further savings, a klaxon from decommissioned Navy subs could be recycled thus incurring further glee for paid overseer/sleep police who could also shout “Dive, Dive, Dive”…

  2. Another solution is to recognize the human element. If a controller really has no work to do apart from the very occasional aircraft, let them nap. Just make sure there is an enormous alarm guaranteed to wake them up.

    Sometimes a 10-20 minute nap results in far better performance than sleep-deprived for the entire shift worker.

  3. You said “In fact, if properly programmed, a modern airliner can more or less fly itself from runway to runway.”

    Is the GPS resolution, altitude and elevation of runway data really accurate (or even known) enough for this to happen with no human intervention? Or did you mean by properly programmed with the necessary algorithms and supplied with good enough terrain data this is possible?

    I think of something like a Predator drones and it has a built in camera for the “pilots” to take-off/fly/land it properly. (Although you wonder, if for any reason the camera goes out, is that $4.5M of (future) taxpayer’s money gone?)

  4. Jitesh: How accurate does the GPS have to be for avionics to get an airplane where it needs to go? You don’t need a GPS at all! For the en-route portion of the trip, the ground-based VORs are sufficiently accurate. For landing, the ground-based localizer and glide slope navigation aids will get an airplane down to 200′ above the runway and all the way onto the runway if augmented by radar altimeter (on the airplane). No terrain data is required whether using GPS or ground-based navaids, since the airplane is flying published instrument procedures that keep it well clear of the ground at all times (except when at the bottom of the glide path toward the runway).

  5. In Australia the first page of the ATC manual states that ATC is responsible for providing the “safe, orderly and expeditious” operation of air traffic in controlled airspace. Now the orderly and expeditious parts pretty much fall to the wayside in low traffic situation – I personally have directed an aircraft to track direct to Brisbane whilst still taxi-ing at Adelaide due to a lack of crossing traffic. But ATC have safety concerns that pilots may not be aware if separating themselves e.g. ground staff working on runway lights, wake-turbulence separation requirements, maintenance crews taxi-ing aircraft to a hanger on the other side of the runway, planned outages to nav-aids, airspace closures for military or civilian tests etc.
    Having two qualified staff on latenight dog-shifts follows the same pattern as two crew on the flight deck – they check and cross-check each other as the traffic patterns emerge around them, and just like 99% of all flights dont push the skills and abilities of the two pilots, two controllers aren’t going to have to make life-saving decisions in 99% of their shifts. But come the night a tired controller clears a plane to land just as a catering truck enters the runway, we’ll all regret not spending the small amount of the money required to adequately staff ATC.

  6. Why was the controller tired in the first place?
    Could it be that he works a 40 hour week when the rest of the world’s ATCS works only 32? Studies have shown the stress of the job, and that MORE time away is needed to maintain alertness.
    Or could it have been that he works a rotating shift schedule? 2nd shift, 2nd shift, 2nd or 1st shift, 1st shift, 1st shift or mid shift (just 8 hrs off between the last two shifts). The FAA’s own study stated that it was a stressful schedule, their suggestion, give up the family, give up school, give up sports, give everything up and just work!
    Or maybe it is due to staffing, after all Regan fired the ATCSs in 1982, 80% of controllers where hired in the following three years. After 25 years you could retire, so starting in 2010 every controller hired to replace the controllers fired, can retire. Most have been retiring over the last few years, and have not been replaced fast enough. Of course it is worse than that, if you where hired in 1982 at the age of 30, you would retire in 2002.
    Staffing at my old facility has fallen below 50% since I retired. Every controller that retires takes 20+ years of experience with them, to be replaced by someone with less than 1 year.
    Just a interesting note for me, the first reports stated that a FAA SUPERVISOR was working the midshift. This would be part of management, not the controller workforce, and management is always over staffed.
    As for one controller or two. Two. In case of sickness or family emergency you would be left with one controller instead of none. A accident on the airport is dangerous even after it happens with only one controller, who would be very busy controlling the field, doing the paperwork, and making the required calls. Then there is the physical needs, more than once as the waste paper basket served a dual purpose, as the biological waste receptical.

  7. Leif: Is it standard practice in Australia to have two controllers in the tower at all times? Even 11 pm to 7 am or a similar late night shift?

    [As far as catering trucks entering the runway at midnight, the layout of U.S. airports pretty much precludes that. Since we’ve had so much funding for airport construction, there are almost always taxiways and service roads that can get vehicles where they need to go. At our airport, vehicles enter the runway only to inspect lights and check braking action. Any vehicle that has access to the controlled surfaces (i.e., taxiways and runways) must have an aviation radio installed. Although as a pilot I prefer the additional set of eyes of a tower or ground controller, I’m not aware of any studies that have found a safety improvement at lightly used airports due to a control tower being established.]

  8. I’ve already enacted a plan to solve this problem. More than 2 dozen controllers have registered for a free bag of Flying High Coffee: http://bit.ly/atc-java through my “free coffee for ATC initiative” (I’m looking for sponsors for this, BTW)

    These guys and gals are hardworking professionals who work odd schedules and are often left isolated, late at night in a darkened room with very little to keep them alert. Coffee is not going to replace adequate rest by any stretch but it wouldn’t hurt. Probably what should be done is to allow for somewhat less demanding schedules. It doesn’t seem right to duplicate the resources…

  9. Amen to that. I’ve been waiting for someone to provide some reason to the media hysteria. Why are we even paying controllers to sit at an empty airport in the first place?

  10. @Joel Many times this is mandated by the airlines. Here in Burlington, VT (KBTV) we have one flight that comes in around midnight (might be different now, I’m thinking some years back). The tower had to stay open until that plane was on the gate. Then they went home and left the field non-towered. Practically nothing happened between 9PM and that last airline arrival.

  11. Rob: I’m not sure how the airlines and the FAA agree on tower staffing. I know that that the airlines prefer to have a control tower in operation, but it is not a requirement and the FAA has the final word on schedules. The airlines also like it when the FAA builds an ILS, but again they can nag, not insist. Looking at http://www.airnav.com/airport/kbtv I see that the tower is scheduled to be open from 5:30 am to midnight. So probably we were running late (no surprise given that we were based at JFK and the economy had not yet collapsed).

    Anonymous: I wouldn’t draw the conclusion from the European midair that sleep deprivation was a factor. The accident occurred at 11:35 pm, presumably right near the beginning of the shift. Nor is the obvious conclusion that more humans should be added to the system. The accident occurred partly because of the mixture of TCAS (telling the Tupolev to climb) and the human controller (telling him to descend; we wouldn’t be reading about this if the pilots had listened to the TCAS computer) and also because of the combination of our archaic airway system with modern avionics that enable planes to fly right in the middle of airways. There was also some radio miscommunication due to our archaic broadcast system that enables transmitters to step on each other.

    A systems engineer might look at the accident and say “We need to get rid of all of the humans above 18,000′ [where planes fly at 400 mph and faster, leaving little chance for human vision and reaction time to avoid collisions] and substitute a fully automated system.”

    Anyway, I’m not sure that a midair collision in some of the world’s busiest airspace is relevant to the question of separating 3 airplanes per hour that have previously been sequenced onto the same approach (i.e., they’re all going in the same direction) by the Approach controllers.

  12. Wait….you have an ATP cert!? I’m so jealous! Not of the hours or pay I suppose, but I love Bombardier CRJs. I don’t know if it’s my imagination, but they seem to push you back in your seat harder than an airbus or Boeing.

  13. Thanks, Jeffrey. $19,000/year for working 22 days/month, 16 hours/day is not the greatest way to support a family, that’s for sure. Coming from the world of four-seat piston airplanes, I loved the quiet of the CRJ cockpit, especially with a Lightspeed Zulu headset and, most of all, the massive air conditioning ducts (about one third of all of the A/C capacity of the airplane comes through the front ducts and then flows back through the airplane). It is ironic that the public is so worried about these government workers falling asleep in their comfortable chairs when the flight crew up front might be inexperienced and therefore exhausted (see http://philip.greenspun.com/blog/2009/05/21/to-improve-airline-safety-give-all-pilots-the-same-schedule/ for why these tend to go hand in hand; by union agreement, the airlines give the toughest schedules to the pilots who are still learning how to do the job).

    The CRJ is a challenging plane to fly compared to the smaller Airbuses or Boeings because of the lack of leading edge devices. Our final approach (Vref) speed was 145 knots and a large attitude change is required during the last 25′ of descent. The Boeing 737 will approach slower and at an attitude that is closer to the attitude just prior to touching the wheels.

  14. “Is it standard practice in Australia to have two controllers in the tower at all times?”

    Towers in Australia, what towers? Except for the major international airports, the norm is uncontrolled airfields all day, every day. The hand full of busy GA airfields with towers shut late in the evening, having no night shift to fall asleep on.

    Some of those uncontrolled airfields do see a reasonable amount of airline traffic too, using CTAF. Anything from Dash-8s to 737s mixing it with light GA day and night. All perfectly safe.

    Two controllers seems excessive. Two people even. I reckon an electric collar around the neck: if it doesn’t detect motion for a few minutes, it zaps the controller to wake them up.

  15. I’ve been an ATC for almost 30 years and I never failed to wake up instantly from a nap during the night to answer any call, simply because I kept all 3 VHF transceivers on and checking for the volume to be loud enough. Some pilots were nice enough to key the mic quickly before calling, while others would literally scream their first call as some kind of joke. In the first place, taking a nap helped me a lot on maintaining my awareness and the capacity to focus on my job.
    On the other hand, there was another controller who possibly suffered from narcolepsy, because once he got asleep I could not get him awake. Maybe the US controller had to go to the restroom, in that case I would take a portable transceiver along.
    I agree with the ideas mentioned above, 2 ATCers for a number of reasons.

    Nowadays I use the night shifts to read stuff when there aren’t any ACFT near.
    And I keep the VHF volume loud for the next call.

    The need for sleeping is insidious, it just happens and at times we can’t help it; otherwise nobody would ever crash behind the wheel.

    Sleeping on the job: it’s not about being guilty, it’s more about human biology.

  16. Freight locomotives have a vigilance control device. Basically a randomized alarm prompts you to push a button to prove that you are awake. I realize that this would be a real pain in the controller’s aft, but it’s cheaper than a second controller.

  17. Here in the Midwest we have several RJ flights into and out of uncontrolled airports each day. Other than an irritating habit of not reporting their landings as promptly as the GA crowd (many but not all RJ flights), operations in IMC are uneventful. My opinion is that this issue of sleeping controllers has been capitalized on by the ATC union for members’ gain and over-reported by media lacking even a simplistic understanding of air traffic operations.

  18. Jitesh’s Predator question combined with Salvador’s observation about human nature made me think about this a bit more. If a pilot at a desk in Nevada can fly a plane in Afghanistan, why can’t a controller in Australia, fully rested and awake during what would be daytime there, help control an airport in the U.S.? So you’d probably still have someone on-site overnight in the U.S., his or her co-worker would be connected to the airport via Internet (cameras to view the runways and taxiways, audio link for the radio, video chat link for communication with co-worker).

  19. Jose: http://www.wsws.org/articles/2011/feb2011/spai-f10.shtml is an article about the sufferings of the controllers in your country and the lack of support from other labor unions. It seems that they have suffered a 40 percent pay cut because they are no longer able to work overtime. The article does not mention that it is a 40 percent cut from a maximum of about $1.3M and an average of about $500,000 per year (see http://www.nytimes.com/2010/12/06/world/europe/06spain.html ). I wrote about this earlier. Most of the airports in Spain that have controllers have less traffic than non-towered Deland, Florida (cited above).

  20. A combination of Philg’s post and the first comment:

    Why not have the “clicking the microphone five times” procedure that turns on the runway lights also activate a loud buzzer in the tower? Clearly the equipment is already in place to [do something] based on incoming radio communications when no one is around. It would cost, what, $100 to attach a buzzer into the same circuit as the lights?

    If no one is truly around – the tower is closed as it is supposed to be – no harm no foul, the buzzer sounds in an empty tower. If some dude is sleeping on the job, well, wakey wakey!

  21. While we’re at it, let’s get rid of the co-pilots in the cockpits. We could replace them with interns that want to learn about flying and train them on how to have the plane’s avionics fly it from runway to runway.

    Would you really propose this treadmill/intern exercise program to save a paltry $7 million dollars from the Federal budget? I have a counter proposal. How about we buy 1 less fighter jet and *double* the proposed number of Air Traffic Controllers to be added? I’m sure we could find $15 million dollars /year in our Defense Budget within 30 seconds with one eye closed.

  22. Christian: The first officer (what you’ve called “co-pilot”) in an airliner is not there simply to keep the captain awake. He or she will do all of the flying on every other leg. When serving as the pilot monitoring, the first officer will do at least the following: calculate weight and balance, performance, landing speeds; handle all radio communications and frequency changes; run roughly 10 checklists; manipulate switches such as the landing gear and flaps handle; check to make sure that the pilot flying is not in the wrong place at the wrong time at the wrong speed. It is true that there is really no need for two pilots to be on the plane during cruise flight, at least if nothing goes wrong mechanically, but a crew of two pilots is required by certificate for all airliners and is needed during takeoff, climb out, approach, and landing.

    In the particular case of these sleeping government workers, the second person’s job is not to handle half of the work (handling 1-4 flights per hour). The second person’s job is simply to keep the first person awake.

    If you want to compare government spending to military spending, you can justify almost any expense. Why not spend $1 trillion on a new elementary school? We spent that much on Iraq and received nothing of value. How about $1 trillion on a new agricultural subsidy? That’s cheaper than the war in Afghanistan’s ultimate cost and the money will be going to Americans.

  23. Thanks Phil for your well written piece and support for our profession. We hope for the best and are preparing for the worst. I thought you should be aware that as an ATCS at a very busy little VFR tower there is not much that can be done to change the work ethics or character of the folks who arrive at the facility to perform ATC duties. We have to accept the people HR sends us for whatever the reason, and once they are here it is very difficult process to train someone perform correctly as a specialist, whom may not have been the best person in the first place for the position. Many have a arrived with a sense of entitlement and little else to offer the rest of the team except to take what they can get for themselves. Having come out of the military myself, I feel that it is a privilege to have received this position, I continue to honor my commitment to the taxpayer to provide, to the best of my ability, the best possible service to the public I can. In return, I don’t expect to be paid a fortune or have some crazy working conditions including naps or excessive breaks. I don’t make the pay that you suggest we all make far from it. 80k is a grand sum for most people I get that. Around the field it is nothing, so I drive an hour each way. I watch people who make 10 times that much all day everyday, including yourself and say nothing. I work a second job to meet ends, to pay for a car that can keep up with the demands of the commute, and get me to work on time, every time. I stress every time something like this happens, and is exploited in the press, that I wont be able to complete my federal service ofwhich I am heavily invested in

  24. KBEDATCT: I did not mean to suggest that U.S. ATC are overpaid in general. http://www.bls.gov/oco/ocos108.htm says that in 2009 the average wage, before overtime, was about $110,000 per year (that’s more than 1/10th of my income, I can assure you, and so is $80,000 per year!). With a little overtime, plus health care and pension benefits, however, the cost to the taxpayer is going to be close to $250,000 per additional controller. Is that too much to pay? Obviously at a busy airport such as Hanscom, no. And I am almost always impressed by the capabilities of the controllers at Approach and Tower facilities up and down the east coast. But I think there has got to be a more creative way to ensure overnight coverage than “Pay two highly skilled people to handle 2-3 airplanes per hour.”

  25. Then there is FAR 91.185:
    “Sec. 91.185 — IFR operations: Two-way radio communications failure.
    (a) General. Unless otherwise authorized by ATC, each pilot who has two-way radio communications failure when operating under IFR shall comply with the rules of this section.

    (b) VFR conditions. If the failure occurs in VFR conditions, or if VFR conditions are encountered after the failure, each pilot shall continue the flight under VFR and land as soon as practicable. …”

    I realize that we are so accustomed to communicating with ATC, that we are somewhat taken aback by getting no response from the tower. Guess we should be prepared, on a regular basis, to go ahead and “land as soon as practicable.”

Comments are closed.