NextBurgh, PA          @VannevarB      About      Pittsburgh Murals
July 07, 2009

Solving the ATC Delay Problem

I am not a good writer, and I am certainly not a brief writer, but I believe that accuracy in naming and describing things is to be desired, and I am enough of a fan of George Orwell to believe that correcting a misleading phrase is a public service.

Recalibrating the Frame

So I must ask that we reconsider the frame suggested by the often used phrase, ATC Delays. In general, ATC does not delay aircraft; ATC just tries to keep them from hitting each other.

Weather is a primary cause of airplane delays. When the weather is bad and unsafe, planes wait on the ground. Don't you think this is wise? Not too many passengers are willing to say, Damn the weather! Tell that pilot to takeoff anyway!

Airline Schedules delay airplanes, which is a paradox of sorts. Usually, schedules keep things on time. (They may complain publicly, but watch what they do, not what they say.) When airlines schedule more departures in a given time than the airport can provide for, then the airport is overloaded and the schedule has caused a delay. The crew may tell the people in the back that "we've got an ATC delay for Boston", but it's probably a scheduled delay that somebody in the airline intentionally designed.

Delays are often an act of commission; the airline industry has clearly found delays to be cost effective over time. This is an external diseconomy: the cost of delays are borne by the public, and the profits of overscheduling accrue to the airlines.

I would like to change the title of this piece (and the public discourse) to,

Solving the ATC Delay Airline Scheduling Problem

Imagine if the railroads scheduled more trains than they had tracks for. Imagine if restaurants scheduled more reservations than they have tables for. And yet, we consider it acceptable for an airline to promise people that 23 jets will arrive in 15 minutes at an airport that, in perfect conditions, can only get 15 down.

Perhaps it shouldn't surprise anybody that the same industry sells 110 seats on a 105-seat aircraft, because they choose to "overbook" as a matter of policy.

Tragedy of the Commons

From economics we have the Tragedy of the Commons (by Garrett Hardin) which is what happens when a common, unregulated resource is overused by people acting in their own best interest. In its most simple expression, the Commons is a large field in the center of town. People let their cows graze on it.

It is in each herder's interest to put as many cows as possible onto the land, even if the commons are damaged as a result. The herder receives all of the benefits from the additional cows, while the damage to the commons is shared by the entire group. If all herders make this individually rational decision, however, the commons are destroyed and all herders suffer. This was literally the experience at Boston Common, but the phrase itself is extended to larger concepts.

I'm sorry to repeat myself, but if those herders had adhered to Kant's Categorical Imperative (only do something if it's okay for everybody to do it) there'd be no problem. If airline schedulers listened to Kant, then airports would be all right, too.

Clarifying the Commons

Let's clarify a few relationships, in the hope of avoiding confusion later.
Who builds airports? Local government, usually a County or Authority.
Who owns airports? Local government.
Who manages airports? Local government.
Can an airline fly into an airport without gate/ramp space? No.
Who sells the airlines gate/ramp space? Local government.
Who profits from the airport? Local government.
Who pays for 90% of the airport? Federal government.
Who works the airline's planes through the local gov't's airport? Federal gov't.
Does the federal government have influence over airline schedules? No.
Who gets blamed for airport congestion? Federal government.

You couldn't make this up, nobody would believe it. I'm open to correction, but it seems like the Feds get the blame for a situation they didn't create and don't have any authority over.

Protecting the Commons

In general, the response to abuse of the Commons is to regulate the commons. In fact, some economists believe that the phrase should be changed to, Tragedy of the Unregulated Commons. Hardin himself concurs with the revision.

So, to follow my points (if I have made any),
  • Runways are constraints
  • Constraints must be addressed
  • An unregulated commons extends benefits to the abusers
    and extends costs to the community

Let us consider what Regulation might bring, and see if it is more or less onerous than current conditions.

Airport Capacity

We have talked of runways and runway capacity, but we should really talk about Airports because that's where people go to fly. An airport is generally a collection of several runways. The configuration of those runways will lead to an "Airport Capacity", and variations in configuration lead to variations in airport capacity. For instance, an airport with widely spaced parallel runways can conduct independent operations even in low weather, but an airport with intersecting runways must run interdependent operations in nice weather, and may have conflicts between the runways in bad weather.

Here is a chart of airport capacity at PHL, EWR, LGA, and JFK.
Airport Capacity in (Arrivals+Departures)/Hour, by Type and (Freq) of Weather
Optimal Weather110, (86%)81, (86%)88, (82%)82, (81%)
Marginal Weather99, (6%)81, (5%)80, (9%)79, (10%)
IFR Weather96, (8%)65, (9%)64, (9%)72, (9%)

As you can see, the airports have their highest capacity in optimal weather conditions, a reduced capacity in marginal conditions, and their lowest capacity in instrument conditions.

Which Capacity Should We Use To Regulate?

So we have three numbers for airport capacity/hour at PHL - 110 in optimal weather, 99 in marginal weather, 96 in instrument weather. Which value should be used to regulate the volume of airplanes?

WWED : What would an Engineer Do?

An engineer would probably pick the instrument capability as the constraining value, which would ensure a flow of traffic that was always within the capability of the airport. In response to questions about unused capacity during nice weather, the engineer would say (1) we don't design systems for optimal conditions, and (2) the unused capacity on nice days leaves room for unscheduled general aviation and charter flights.

WWGD : What would Goldilocks Do?

Goldilocks would avoid the extremes and take the middle choice, echoing Aristotle's Golden Mean. Goldilocks would pick the Marginal Weather Capability value.

WWVD : What would Vannevar Do?

I'd pick an airport capacity that would accomodate operations 95% of the time, generating delays only 5% of the time. This 95th percentile capacity value would result in airport capacity values of PHL: 98/hour, JFK: 73/hour, EWR: 72/hour, LGA: 75/hour.

WWSD: What would Solomon Do?

Solomon, known for his wisdom and pragmatism, would probably say: Adopting any of these values as the hourly capacity for the airport, and enforcing that limit, is so much improved over the chaotic status quo that any of them is acceptable.

How would you administer the limit?

There are multiple options available for administering the limit. Options include:

The Lottery

Usually, lotteries are appropriate to randomize the delivery of windfalls and tragedies. Who wins a million? Who gets drafted? A lottery would have a randomly distributed result, but probably not an effective result. If airlines were permitted to buy and swap slots, this would truly be gambling over public assets.

Federally Reregulated Airlines

Airline deregulation in 1978 caused a fair amount of this problem, although congested airports existed during deregulation. The federal government could regulate airlines on a national level, distributing slots per hour at the 35 major airports (the OEP airports).

Locally Regulated Airports

Federal government sets the hourly capability for the airport based on technical factors, and then the Feds are out of it. Airports will be responsible for assigning their available slots as they see fit. They may choose between arrivals and departures; they may choose among airlines. Local airport boards, who have control over airport construction and management, and who enter into airline leases for gate and terminal space, will be able to coordinate their airline activity as they see fit within federal capacity standards. Airports will not be allowed to tolerate airline schedules over and above their capability, on pain of losing their AIP (Airport Improvement Program) largess.

How do the Airlines Respond?

So now the airlines are regulated and restricted to a certain number of takeoffs and landings per hour. Those operations can be assigned by (1) lottery, (2) federal planning, (3) local airport preference/auction.

Last year, the airlines routinely scheduled 80 operations per hour.
Next year, the airlines are only authorized 60 operations per hour.
How do the airlines respond?

Airlines will respond rationally

If you give an airline 6 operations/hour rather than 8, they'll optimize their business to make profits within the new structure. Instead of twelve departures/day from LGA to Richmond, VA, maybe they'll have five and fill every seat. They may start flying 128-to-159-seat Boeing Classic 737s instead of 50-to-100-seat Regional Jets. (They started flying the RJ's for union busting anyway, not their operating costs.)

The focus will go from airplanes to people

Really, nobody should care about airplanes; we should care about people. When we reduce Newark to 72 operations/hour, and Continental thinks it can sell the same number of tickets, they'll schedule bigger airplanes. The result? Same number of passengers moving, fewer aircraft, fewer people delayed.

This shift to people-focus is complex. For instance, if Continental decided to use B767s instead of B737s, the B767 will need additional wake turbulence spacing, and it becomes a multivariable problem. They have lots of computers, they'll figure it out.

The airlines and the airports have been overbooking the runways just like the airlines overbook seats on their planes. They do it because it's cost-effective; the profits go to the airlines and airports, and the costs go to the travelling public and the economy. When too many show up, they bump some back to the next hour, and so on, and so on.

What Won't Solve the Airline Scheduling Problem

Flow Control

For at least twenty years, Flow Control has attempted to stand between the unrestricted, uncoordinated airline schedule and the known capacity of airports. Perversely, federal Flow Control imposes delays on a situation they did not design and have no authority over, to keep the operational situation from snowballing into ramp gridlock and unsafe situations. We throw more money and attention into Flow Control each year, and it has not improved the situation since at least 1985. We should abandon this flawed concept and deal with the root issue.


The current aviation system can deliver any number of aircraft you'd like to the airport. What would you like - 40/hour, 80/hour, 120/hour, 180/hour? The "outdated WW2 system" can do that now. The constraint is the runways, and NextGen will not do anything about runways.

Look at the 9/11 grounding of all airborne aircraft, an unprecedented feat of choreography. The legacy system and the people at the scopes made it look easy. NextGen would not have passed the 9/11 test.

Vannevar's Recommendation

  • Calculate hourly airport capacity based on 95th-percentile weather conditions.
  • Manage the constraint with Locally Regulated Airports. Let the people who own the airport, who profit from the airport, and who rent space to the airlines manage their own capacity as they see fit. Take AIP funds away from those that don't.
  • Incentivize desired behavior. Any person delayed more than 20 minutes should get a $100 cash payment from the airline. Any person delayed more than 40 minutes should get a $200 cash payment from the airline. When delay costs become internal rather than external, those smart airline guys will become concerned about delays.
Until we identify, quantify, and manage constraints, the Airports and Airlines deserve the delays that they're designing. The costs are born by the public, and the benefits go to the Airports and Airlines.

Don Brown, a highly respected, thoughtful guy with an operational perspective, presented his recommendations in Sept. 2007.

Overton Window Alert: On the web today, 7/7/09: The United States... ...has adopted a pragmatic approach to the future of its national airspace system, based on a desired mid-term outcome that makes the best of technologies already available and gives industry a leading role in showing the way forward....


Marc Cote said...

My compliments on a well written, balanced, plain english article that does a great job explaining how the airport system currently works versus how it could/should work. I follow your logic very nicely about Nextgen and how it will not solve the issue.

Anonymous said...

just a thought
large airlines use schedules to block out competitors from regional markets, by over booking and airport and causing a delay a guy with a lot of jets can hurt a guy with fewer jets. in other words the delays kill the little guy, while i can us my larger network and fleet to soak up cost and delays.
airports make money when people are delayed, its simple the longer you are stuck at an airport the more money you will spend

Caribou Barbie said...

Golly!! I am sick and tired of this kinda "gotcha" media. I quit reading your blog.

Daniel said...

What does that mean - "gotcha" media?

Anonymous said...

Hi Daniel,
I might be wrong. I think Caribou Barbie is a joke about ex-Gov Palin, who's been quitting things. "kinda gotcha media" is not too different from her syntax. You could google "gotcha media". Just a guess.

Krak@United said...

Look Vannevar, it's working. We get it, the focus is going to the people. We just needed a shove.


Vannevar said...

Hello Krak - that was my post for Friday! I've just moved it online now. Thanks for your comment. Vannevar.

Daniel said...

Anonymous -

Thanks, that makes sense. I'm not sure what the Palin jab has to do with anything in this context, though.


Orville Wright said...

Runway 31L at JFK will be closing for 4 months.,0,4645096.story

Funny how all of Vannevar's points are validated and not one sign of Robert Poole saying how NextGen could solve the problem.


Post a Comment

Comments and Feedback? Love that stuff. Please leave your thoughts in the box below--