February 22, 2019

Commercial Aviation Part 8

Neal Schier, a pilot for a major US airline, has graciously agreed to continue my series on commercial aviation.

Airline delays and the idea of scalability

Woah! What kind of boring title is that? Ok, I admit it is rather boring, but delays and scalability frequently play a role when you travel—and an important role at that, for who likes to be late? With that old maxim in mind of “with every factor added, complexities grow exponentially”, I will look at why delays are an anathema to the air transportation system. In part 2 on this topic, I will look at how the authorities get the system back on track—an effort that to me as an airline pilot is still nothing short of miraculous.

Imagine for a moment that it is a Wednesday evening and you are sitting in a full airliner at New York’s La Guardia airport. It has been a sweltering August day and you have had three days packed with meetings that did not go as well as you had hoped. Add in the annoyances of a chatty cab driver and delays going through security, and you are entitled to want nothing more than to get back home to Chicago. You want to get into the office early tomorrow morning for an “after action” report of all that went wrong in New York. It has not been a good week…

Except that you are not going anywhere—at least for the time being. One of the pilots has just come over the speakers with the dreaded “Well folks” announcement. As you fidget with the gasper above your head in an attempt to get some fresh air, you hear the bad news. Air Traffic Control (ATC) has advised the cockpit that the Airbus A320 you have just boarded will not be going anywhere for at least the next hour. The pilot, chatting away in terms that sound a bit too technical to fit your mood, describes thunderstorms that are requiring the westbound flights be re-routed further to the north. This is causing a bottleneck in the sky and to relieve this congestion, similar to merging three lanes into one on the highway, the controllers are metering the aircraft over this new route with more spacing. Unfortunately, everyone will have to wait their turn and thus the holdup.

You peek out the window and see the taxiways packed with airplanes that are going nowhere fast. You reach for your phone to look at the weather map. What kind of nonsense is this? Chicago looks to be as clear as a bell! Well, that’s the question until you slide the map over to the area between New York and Chicago and you see a line of severe storms extending from Toronto all the way down to Virginia. Maybe the pilot was right after all…best to be here on the ground instead of doing battle with Mother Nature.

Apart from the food vendors in the terminal, no one likes delays. Not you, not the airport, not ATC, and believe it or not, certainly not the airlines. Every factor, such as a delay, introduces complexities into their passenger, aircraft, and crew routing systems that can, and do, grow exponentially.

Here is where the term scalability comes into the discussion. Scalability brings both benefits and challenges. Think of Henry Ford finding that he could just as easily build 1000 cars a day on a production line as opposed to 5 by hand. In the airline industry the idea is the same. If you are only operating one aircraft between Chicago’s O’Hare airport and La Guardia then life is simple but perhaps not optimal for profits. You will need X number of pilots, Y number of flight attendants, Z number of mechanics and so on down the line. But interestingly, as we know scaling up can help with costs—if you lease five aircraft instead of just one you might get a better payment and interest rate. Plus, your passengers might want to go to other places than just between these city pairs.

However, as you add crew, aircraft, and new city pairs (“stations” in the language of airline operations) you have to manage the complexities that are inherent in growth. If you are not good at this, then your revenue opportunities will not keep pace with the costs and that is not good for the bottom line. Airline managers have to constantly work toward finding this balance and it is not an easy task to be sure.

The route network of United Air Lines, 19311

Although I do not have experience in the field myself, I know that mathematicians and operations management researchers have crafted computer models that track the growth of complexity within systems as factors are added. Not surprisingly, airlines are very susceptible to these complications when one of those factors is a delay—there is a cascading effect that grows with breathtaking speed.

Yet it is not only weather that adds troubles into the mix. An aircraft might have a mechanical issue that requires the mechanics to fix it before takeoff. This fix might require just a few minutes or an hour—it all depends on what went wrong. The worst case, of course, is that the aircraft will have to be taken out of service completely—and if that doesn’t just spoil your day!

Flow control into and out of the big airports is another cause of delays. While US airports are normally not formally slot controlled as they are at places like London’s Heathrow, there are a lot of planes competing for the runway and they cannot all arrive at the same time. This means that ATC needs to feed them in at what is known as the “acceptance rate.” Most often this rate depends on things like how many runways the airport has in use at any one given time. This means, sadly, that no matter how good the controllers are, there is not a lot of flexibility in simply taking on more arrivals or departures.

Nor would it help if one or more airlines voluntarily decided to be a nice guy and just draw down the number of flights to a destination so as to “help out.” If American, Delta, or United were to withdraw a couple of flights from La Guardia for example, another airline would be quick to try to pick up the open space. Other airports around the nation, just like those in the New York area, frequently operate for a good chunk of the day at very close to capacity.

Understandably, there is always economic pressure for flights into and out of these “high demand” airports—it is where people want to go. The airlines and their passengers certainly desire a robust frequency of service and, when the weather is good, this more or less works. Throw in a complication though, and it gets ugly. Where is the right balance? The fact is that no one quite knows and, to use the cliché, it depends on whom you ask. A hot button topic to be sure!

Now we toss in yet another factor. Pilots are normally assigned only one type of aircraft to fly at a time. This means that when a pilot is delayed from another flight it can disturb the schedule. Contrary to what some believe, you cannot just yank a pilot from the terminal and have him fly any aircraft in the company’s inventory. Not only that, but pilots have their own duty time clock running.2 The pilot you see grabbing a coffee at the snack stand across from your gate might not have enough duty period to fly you to Los Angeles—and that assumes it would even be on an aircraft on which he was qualified.

Ah, but there is even more! If one aircraft needs to be substituted for another (an E-sub in the airline lingo), yet more work needs to be done—work such as getting passengers in the correct seat, pilots to fly the plane, the mechanics to ensure it is ready to go, new flight plans to be printed, etc. You are now working an algebraic equation with a lot more than just solve for x.

I realize that airlines are not always faultless. I dislike making the hated-by-all announcement that “We are going to have to slip our departure time” as much as you dislike hearing it. Passengers hear this all too often and it is a major, and understandable, cause of frustration. Yes, we all wish that airlines would build bigger cushions into the schedule to allow for the unforeseen, but just how big should they be? Ah, but again we meet another of those eternal airline questions. As an aside, one exception to this is on late night international flights where we often wait for passengers if we can make up that lost time enroute. These are critical connections and, when it is the last departure over the ocean on that day, the airline will try to get any delayed passengers on board.

As you can see, there are many causes for delays when flying. Good and honest information is a must when the airline, and we in the cockpit, communicate with you. While it might be small comfort, I want to stress that everyone in the air transportation system dislikes delays just as much as you do. Domestically alone, we are moving the population of Philadelphia to Minneapolis (that many air miles) every 24 hours and so even a slight disruption to that flow can mean a lot of work for everyone—those in the ATC system, the pilots and cabin staff, and particularly the aircraft routers who have to roll up their sleeves and get it back to normal.

Next time I will take a look at how they get things back to normal—how the disruptions are unraveled and how the system is “recovered” or “reconstituted” as I like to say.

In the meantime, clear skies!

Neal Schier/February 2019

1 From the astonishing collection of old airline timetables at timetableimages.com.

2 Bean: There are strict rules about how long pilots are allowed to work. They're rather complicated, but the short version is that a pilot has to have parked the airplane after their last flight within 9 to 14 hours of reporting for duty, depending on a bunch of factors, and can't be in the air more than 8 or 9 of those hours. He then needs 10 hours of rest (including the opportunity for 8 hours of uninterrupted sleep) before going back on duty. There are also weekly, monthly and yearly limits.


  1. February 22, 2019Chuck said...

    Is there any kind of scheduling priority that gives larger planes more priority over tiny planes when it comes to bottlenecks like the one described? Do landing fees keep smaller planes from sucking up all the capacity at major airports, or is there something else at work?

  2. February 22, 2019bean said...

    AIUI (and I'm sure Neal can say more on this), the bigger airplanes are given priority by the airlines themselves. If you can only get one plane in right now, it's obviously better to pick a 777 with 300 people onboard than a 737 with half that number. Fewer annoyed customers, fewer hotel vouchers, and less lost revenue from having to cancel the next flight. I once heard of someone who was flying from LA to New York during a major storm rerouting through Europe because international flights typically get priority and it got them out of the mess that was the US air travel system.

  3. February 22, 2019Neal said...


    That is a really good question and, as weak an explanation as it might seem, the answer is it depends on a lot of factors.

    I can say yes in that when a delay strikes the airline wants to move as many people as it can and thus they will want to launch the bigger metal if it is available to clear out any backlog. Often however, it is not at hand at the airline has to cobble together a plan B. Plus, as every traveler knows to his/her frustration a lot of the smaller aircraft are outsourced to a separate operator. This is in no way a dig at the regional carriers, but merely that it sometimes adds a layer of complexity into the decision making process and what can be used to move passengers.

    With Bean's good allowance perhaps I can address this at length in the future but I will bring up two of those "it depends" factors.

    The first is that air traffic control system is, more or less, a first come first served arrangement. Say that Airline A has two 737s out on the taxiway at Dulles getting ready to go to Chicago. There are two because the delay has been weather related and it just backed up over the next departure. Airline B also has a couple narrow-bodies or regional jets destined for Chicago. Airline C cannot step ahead of the queue just because it has a 777 that holds more people. Airborne it works the same way in that a faster 777 cannot just "pass" a slower aircraft into Chicago. The controllers can, and do, let this happen at times, but they are trying to get it all sorted out so that the planes are spaced roughly a couple of minutes from each other on final approach.

    I know this is not exactly the question you asked, but I use it to stress that the bigger plane does not automatically get priority to the same destination. For landing it also works the same way unless a flight is getting low on fuel and announces that to the controllers.

    That said, the airline dispatch section has a desk that works directly with ATC (more on this in part 2) and they can often work to get a flight in--I have seen this going into airports with curfews--Orange County/John Wayne on the West Coast for example.

    The second big factor is how many "connectors" the airline might have on any one flight. Going into La Guardia for example there will not be many connecting flights. But a flight into one of the airline's major hubs might have a good chunk of passengers trying to reach the flight to Tokyo or Shanghai or Frankfurt. Here is where the airline really has to do some thinking and there is now software to help. In other words, that 777 that is destined for Atlanta might be full of connectors that are no problem to accommodate. That 737 however, might be full of passengers that only have one option that day to get toward where they are going...and yes, revenue yield does come into play from what I hear but the airlines keep that very close to the chest apparently as I have only heard it described in the terms of "we have X number of Asia or Europe or South America connectors on board."

    To show you how forward thinking this has to be, think of a passenger going from Albuquerque to some place like Tashkent. They have to go through Denver or Dallas or Atlanta and then probably to Europe (maybe Asia but I have seen Lufthansa handle a lot of these less-traveled city pairs. From there an alliance partner airline will pick them up--but there might be just one flight a day. Sometimes it just doesn't work, but the airline is expected to make the best efforts to make it happen.

    Bean brings out a good example of how creative the agents can get to route a passenger onward--sometimes it is just jaw-dropping how they piece it together.

  4. February 22, 2019bean said...

    Bean brings out a good example of how creative the agents can get to route a passenger onward--sometimes it is just jaw-dropping how they piece it together.

    That wasn't the airline agents. It was the work of someone who runs a points booking service for a living, so they were at the very least semi-pro at doing this sort of stuff themselves.

  5. February 22, 2019LordNelson said...

    @bean Although I enjoyed reading the re-routing via international flights story, I feel there is a limit to the amount of time any sane person should want to spend on a plane.

  6. February 23, 2019bean said...

    There are some people who are clearly not sane about such things. People who want to do things like break the record for speed around the world using scheduled airline flights.

  7. February 26, 2019Neal said...

    This is a reply that retired airline Captain R. Chaber offered as a round-out of my views on delays. He makes a great point as to what the customer is looking for in the commercial bargain with the air carrier for transport.

    His reply:

    I’m very much involved in sales---wine sales---and that’s the way I think.

    Sales are a negotiation. A year ago I read a book and now I get weekly emails from a man who was formerly an FBI hostage negotiator.

    Most people speak from the point of view of the vendor. The negotiator says you have to speak from the point of view of the consumer.

    Example: this wine is from grapes from one of the finest vineyards in Sonoma. It tastes of blackberries, black cherries, and has an element of baking spice.


    This wine would go perfectly with your Thanksgiving dinner. You’ll see it complements the turkey wonderfully; it goes with the sweet potatoes and even blends with the cranberries. This is a wine you’ll enjoy and be pleased to serve to your friends and relatives!

    Your essay is mostly from the point of view of the vendor.

    What the customer is interested in is: I’m booked on a 6 P.M. flight out of La Guardia. I should arrive in Chicago at 7:45 PM. That’s what he was sold and that’s what he expects. Anything less than that and he’s not getting what he was sold. He may be polite about it, but in the end, he’s not getting what he was sold. That’s the customer’s point of view.

    The airline has a different point of view:

    • Weather-it happens every day and delays, cancels or diverts flights • Mechanical-things break, and delay, cancel or divert flights • Crew legalities-they happen, and this includes misconnects and running out of crews • Air Traffic Control delays-may be due to weather, but also radar and communications failures and lack of staff • Airports-curfews, runway construction, mechanical failures, security interruptions or closures • Passengers-illness or misbehavior • Environmental-volcanoes, earthquakes, tsunamis • Other-everything else

    From the point of view of the airline it’s a wonder any flights get completed.

    The customer doesn’t really care. He was promised to arrive in Chicago at 7:45 PM.

    The airline can take steps to minimize some of the above problems, but they certainly cannot address all of them.

    One of the things the airline can address has to do with building the schedule. As it is now, they put the whole problem in a large computer and it builds a schedule which has a crew for every flight and does it in the least expensive manner.

    I’ve always questioned this and I use a San Diego bank I flew as an example. Three B-737 flights are due in at the same time. They come from SFO (me), SMF and DEN. The outbound flights go to the same three airports. Each flight crew changes airplanes and two of the three flight attendant crews move to a different airplane. What this means is that none of the airplanes can depart until all three flights arrive. One day I arrived on time and the other two flights were delayed more than an hour. So there we sat at SAN with a good airplane, a pilot crew, and, a flight attendant crew, and we couldn’t go anywhere. I called the company and suggested we do whatever reassigning necessary and get at least one flight out on time. I pushed this enough that they wouldn’t talk to me anymore and over 300 people sat in the terminal not going anywhere.

    Normally we don’t get to see things like this, but this was a far corner of the country; there were only three flights inbound, and the outbounds all returned to the airport they came from. In this case, and probably many other cases, if airplane changes and flight attendant changes were minimized, although the planned schedule would cost more, in actuality the operational costs would be much less. But, the people who construct the schedule work from a different budget than the operational people. You don’t get points for constructing anything other than the least expensive planned schedule.

    So, can you work out anything from the point of view of the customer?

    All I can think of is giving the customer odds of getting there on time. If you’re booked from LGA on the 6PM flight what are your odds of getting to Chicago at 7:45PM? They’re certainly not 100%, but any odds are based on past performance. The odds in April are certainly better than January or July. How far back do you go in constructing odds? Maybe a flight that was always late was the last flight of a pilot’s long day. Any delay they incurred all day would affect their last flight. But then, last month, the airline reordered the schedules and now the flight in question was the pilot’s first flight of the day. The old odds don’t apply any more. I don’t know how any odds-makers would calculate this. Maybe you need a matrix of odds: last seven days, month to date, last 30 days, year to date, one year.

    You know, if people can make stock market predictions and other people bet on horse-races, maybe someone can publish odds of any flight being on time, and, if late, how late might it be?

  8. February 26, 2019bean said...

    Interesting. That San Diego bank sounds weird. Normally, I'd suggest there was substantial connecting traffic, which is why you wouldn't let someone go early if the others were delayed, but you're not going to see that through SAN from SMF, SFO and DEN.

    On-time arrivals are tracked both by airlines and the feds, and at least some data is available to anyone who wants to look for it. In fact, at least some travel search engines show the on-time percentage of flights they offer, if you really care about it. Or you could just fly Delta, who has the best operation, despite their rampant hypocrisy and high prices.

    Or the average passenger could continue buying whatever's cheapest on Expedia, then getting mad when things go wrong.

Comments from SlateStarCodex:

Leave a comment

All comments are reviewed before being displayed.

Name (required):

E-mail (required, will not be published):


You can use Markdown in comments!

Enter value: Captcha