From the National Air Traffic Controllers Association FAQ we can see some interesting numbers.
On any given day, more than 87,000 flights are in the skies in the United States. Only one-third are commercial carriers, like American, United or Southwest. On an average day, air traffic controllers handle 28,537 commercial flights (major and regional airlines), 27,178 general aviation flights (private planes), 24,548 air taxi flights (planes for hire), 5,260 military flights and 2,148 air cargo flights (Federal Express, UPS, etc.). At any given moment, roughly 5,000 planes are in the skies above the United States. In one year, controllers handle an average of 64 million takeoffs and landings. ...There are 14,305 air traffic controllers that work for the Federal Aviation Administration, according to FAA data (dated Aug. 2006).
Assuming the air traffic controllers don't work much overtime (unlikely) we can guess that 20% of the ATCs may be on-duty at any particular moment -- they work 24 hours a day, seven days a week -- so about 3,000. Does it really take 3,000 air traffic controllers to coordinate 5,000 flights?
I'm not a pilot, so am I missing something? Is our air traffic management system so inefficient that we need a controller for almost every flight?
Something is fishy with those numbers. How can there be 87,000 flights per day, but only 5,000 planes in the air at any given time? One of those numbers is seriously misleading.
On air controller workload, consider: 1) they must have capacity to handle peak load times, and 2) they must manage the planes on the ground as well as the planes in the air. And if 87,000 flights per day is right, each with a takeoff and landing that must be managed, then 14,000 controllers would make sense.
It's an average flight-time of about 1.5 hours, isn't it? Doesn't seem so crazy. Also, a large fraction of flights leave or enter the US air-space, thus being either a take-off or a landing but not both, and only the time spent in US air-space counting towards the 5000 flights at any given time. eg. a flight to Europe from New York leaves US air pretty much immediately, but would count as a flight I guess. If 1.5 hours still seems low, perhaps that can explain it.
It seems reasonable to expect that the marginal cost in air traffic workload of adding a plane to an airfield space grows with the density of planes already near that airfield. If so given that density is unevenly distributed among airfields, just a few of the most dense airfield would account for a disproportionately large portion of the controller staff.
J.
mauyr: I think the work of preventing collisions would only increase linearly with the number of planes, since they mostly fly straight-line routes and are spaced out pretty far. I'd have to think about the problem more though.
BB: Peak times don't just sneak up on them though, the flights are scheduled in advance. And yes, one plane will use multiple controllers over the course of a flight, but at any given time it only needs one. Controllers must have a lot of down time.
mauyr: An average of 1.5 hours per flight doesn't sound too outrageous, but you can only get that number if you assume that they're doing it 24 hours a day, and that they don't spend any time on the ground refueling and taking on passengers and cargo.
On takeoffs and landings, the math indicates that there aren't many international flights fouling up the ratio of takeoffs and landings to the number of flights. 87,000 flights per day times 365 days per year, times two (one takeoff and one landing) yields the 64 million figure from the FAQ.
I'm not saying that there's any sinister conspiracy behind the 5,000 plane figure. I just can't make sense of it. Maybe it means that 5,000 planes are in the takeoff and landing zones and being actively managed by the controllers. I dunno.