Runway Incursions

by Philip Greenspun, ATP, CFII; June 2008

Site Home : Flying : One Article


A potentially hazardous runway incursion occurs when an unauthorized aircraft taxis onto a runway that has already been allocated to another aircraft for takeoff or landing. Over the last few decades, the FAA has invested a lot of money and time in trying to eliminate runway incursions, but they still occur every day.

How are ground operations at a busy airport regulated? A voice communication link is established via radio between a controller and the pilots of the aircraft on the field. What could go wrong?

In short, all of the things that can go wrong when humans rely purely on memory and oral/aural communication do go wrong. Add to this the fact that taxiing a $50 million airplane around a busy airport tends to result in nervousness. Add to this the fact of distractions, e.g., running checklists, starting additional engines, responding to questions from flight attendants, trying to reach maintenance or dispatch, etc.

How complex is the task? Look at the Boston Logan Airport Diagram. Consider an airplane that has just landed on Runway 4R and comes to a stop just after the intersection with Runway 15-33, i.e., abeam the "satellite fire and crash station". This airplane now needs to get to the terminal. It might be instructed to "Turn left on Yankee [Y], left on Runway 33R [using another runway as a taxiway temporarily], hold short of Runway 4L." After the conflicting airplane has completed its landing on 4L the original airplane would be instructed to "Cross Runway 4L then turn left on November [N], cross Runway 33L, hold short of Bravo." Then "Airbus crossing left to right on Bravo. Behind him taxi via Alpha to the gate."

How often do these problems occur? You can monitor the ground control frequency at any big airport in the U.S. using liveatc.net. In the course of a few hours you will certainly hear confused pilots requesting clarifications. You will hear controllers reprimand pilots for failing to turn or hold short in the correct place. These are usually not serious problems, involving taxiways rather than runways, but they are indicative of a system that cannot be perfected.

The FAA Way

To reduce runway incursions, the FAA has done a lot of good work on airport signage, line painting, and lights. Turning onto a runway now looks a lot different than turning onto a taxiway and this has reduced the problem considerably. The other prong of the FAA effort has been training. This really means hectoring pilots and telling them to do a better job. There is no simulator training for taxiing around a busy airport. Airline pilots do this all day every day anyway. Airline procedures already require that pilots have runway diagrams out and available for reference and that pilots brief each other on likely taxi routes.

Pilots already receive extremely thorough training, tend to follow procedures to the maximum extent possible, and try their hardest to do what they thought they heard. The crew of a major U.S. airline will have at least 40 years of flying experience divided between the captain and first officer, yet these crews have made errors and some of those errors have been very dangerous. In fact, the world's deadliest aviation accident, the collision of two Boeing 747s in Tenerife, was caused by a runway incursion (more). The crews of those airplanes were among the best and best-trained pilots in the world. There is no reason to expect that today's pilots, regardless of any amount of FAA-mandated training, will do any better.

The Modern Technology Way

What do airliners have on board to provide situational awareness around an airport? A piece of paper, i.e., the airport diagram, the pilots' eyes to look at signs, and the pilots' ears to listen to the ground controller on the radio. In other words, exactly what an airliner would have had in the 1930s despite the fact that airports are vastly more complex and nearly 100 times busier.

What does a $20,000 four-seat airplane with a propeller in front have? A $1500 GPS stuck on top of the dashboard that shows (1) a backlit airport diagram, and (2) the position of the airplane on the airport diagram.

Would it be sufficient to put a $1500 handheld GPS in every airliner? No. The cheap GPS units do eliminate any potential confusion about "Where am I right now?" but don't do anything about the problems of miscommunications and misunderstandings inherent in using voice communication.

The basics of any successful system are the following:

How would you build it? You'd start with the $1500 handheld GPS. Now we have a computer that knows the structure of the airport and the position of the airplane at any time. You'd add a wireless computer network, perhaps WiMax, to cover the airport and an 802.11 modem inside the GPS. Now the controller can send text instructions to the GPS. Finally you'd add a bit of software to the GPS to cause it to sound an alert if the airplane deviated from a highlighted taxi route.

How difficult would it be to build this? Because airports are so structured it turns out that you could use off-the-shelf open-source instant messaging software. An oral instruction of "Delta 223 taxi via Echo cross 4R hold short 4L" could be typed as "DL223 E *C 4R *HS 4L" and easily parsed by the computer in the GPS. A ground controller could either type the text into an instant messaging client or use a mouse to pick up an airplane and drag its authorized path on the management screen.

Will it ever happen?

Sadly I doubt that this scheme, or anything like it, will ever be implemented. Due to regulatory hoops and fear of litigation, something that would cost $200 to do for a consumer costs $2000 to do for a four-seat Cessna and $200,000 to do for an airliner. The current method of voice communication and paper diagrams has proven to be a failure. Nonetheless, it is an FAA-approved failure. In the event of an incident or accident, the FAA and airlines can demonstrate that everyone was complying with the current best practice.

More


Text and photos (if any) Copyright 2008 Philip Greenspun.
philg@mit.edu

Reader's Comments

What about a sort of reverse 'bread-crumb' solution? A way for the pilot(s), once on the ground to be actively shown the way to go. It seems that when an airplane is on the ground, you need three pieces of information: who they are, where they are at a given point in time, and where they need to end up. From that you can calculate everything else and make dynamic adjustments along the way.

Let's assume who they are can be identified through their transponder code. Where they are can be established by placing them at a known spot, say a given taxiway with a bright green 'COME HERE' light. Since runways are essentially serialized pipes and one airplane at a time can be on them, you only need to time it so as soon as airplane A lands, the proper taxiway is lit for that plane and that one alone and as soon as they cross the threshold, the light turns off, ready for the next arrival.

Where they want to go depends on whether they are a big plane with a fixed pre-determined destination or a smaller plane where a dialogue with ground control is required to establish intent. So you get the big planes off the runway quickly which is how it should be. If it's a known route, just light up the rest of the bread-crumb for them to follow. Turn the breadcrumbs red to indicate they need to stop and let someone else go by.

The breadcrumbs can be in the form of bright lights (or indicators) to the side of the runway/taxiways, or embedded lights right down the centerline (that a lot of airports already have installed). The only thing is it needs to be visible in hard sunlight. Either way, some thing that can be cheaply installed and wired for active control.

If it's not a known route, then a system needs to calculate a best-fit path, given where they are and where they need to be. Since the number of options are fairly finite in a given airport, it's not a hard technical problem.

To fine-tune the system, a series of location spot-checks need to be performed, meaning if they tell Airplane A to go from point B to point Z, then there should be a way to make sure they crossed over points C, D, and E in a timely manner so as not to hold the other people up. The solution can be as simple as your lowly FastPass-like traffic toll transponder/RFID chip (attachable with velcro to the windshield, natch :-) Pass the sensor at spot C, check. Taking too long to get to D. Your breadcrumb light just turned red so you stop to let an airplane cross the taxiway. If you have one of these, now you don't even have to rely on the transponder code to identify the plane.

This way, the only thing the pilots have to pay for is a FastPass-like transponder. Not too onerous a burden especially if it help speed up their coming and going. If someone doesn't carry one of these, they get a lot of slow red lights while ground control routes them in short-burst segments to their destination.

This way, the pilots only have to pay attention to the breadcrumb lights on the roadway, instead of dividing attention between the outside and the screen inside the cabin -- this after a possibly long, tiring flight.

The whole cost of development and installation would be much less than making every pilot/airline buy an expensive computer/GPS device for each aircraft and it all can be done using existing time-tested technologies.

My $.02.

-- Ramin Firoozye, July 2, 2008

Take ASDE-X or ASDE-3X. These are supposed FAA blessed improvements over AMASS. some people don't think 3X is worth it (see http://www.natca.org/legislationcenter/testimony-detail.aspx?id=226).

AMASS has formal rules about aircraft-aircraft encounters. it has formal rules about aircraft-ground vehicle encounters. true, it is limited by its sensors. but even when it is not, several of the safety rules are disabled in the operational deployment because Controllers and the owners-operators of airports WANT them to be. in short, a busy, operating airport, to maintain its throughput, needs aircraft juggled and stacked in a pattern that violates AMASS' safety rules. the resulting alarms and alerts are annoying, so the rules are disabled.

the same kinds of approaches are taken to the software in ASDE-X and ASDE-3X. there are a lot of ambiguities, judging by sensor-only data. and configurations of sensors need to be tuned to particular airports in order to remove effects of multipath and reflections at that airport. no doubt these could be configured properly if the original software were more transparent.



-- Jan Theodore Galkowski, July 2, 2008

I think the fundamental problem is that we're using the same people and the same communications techniques on the ground as in the air. That's convenient, but perhaps not as efficient as it could be, taking into modern technology and energy costs. Airplanes are designed to fly, and pilots trained to fly them.

What if, once clear of the active, airport-provided tugs were waiting to hookup to the just-landed aircraft? Operators of these would be extremely familiar with that single airport and they would have radio and other technologies specifically designed for moving airplanes around the ground efficiently. The tugs would be the type which completely lift the front wheel off the ground, not the push-bar type you usually see at the gate. The lifter type allows for much faster taxiing and is more like highway-capable truck than a tug.

While under control of the tug, the airliner could have both engines off, saving quite a bit of fuel. You'd need a lot of these tugs ("Roger, number 16 in sequence for takeoff"), but between reduced fuel and engine wear, perhaps it would at least be a wash?

-- Trenton Lipscomb, July 2, 2008

So if a major part of the problem is that the per-plane expense gets to high why not deploy some technology at the airport itself?

Would some system of text boards next to the various junctions directing planes by their tail-number (or whatever the common id they use is)? I'm thinking boards that reinforce the existing system, saying things like "XXXXX: Hold here" and "XXXXX: Left on November" with a zonking great arrow pointing down November.

Of course when you're done you have only done that airport, but at least you don't let the old skills rust. You would reinforce the existing comms, and if the system was out, or a pilot arrived an airport that didn't do it they ought to still know what to do.

-- Ashok Argent-Katwala, July 2, 2008

Phil,

There must be some more immediate ways to reach many of the same objectives without the burden of certified equipment. Why not change FAR 91.129 so that controllers are required to give explicit clearance to taxi on the runway? Also, airports can (easily?) deploy dynamic signs (e.g. traffic lights) which might be set automatically when a runway is active or manually to control arbitrary surface movements. Finally, in addition to the current system of named taxi routes, you might have color coding ("continue on blue, turn left on green") to make some information easier to convey.

Aside from Tenerife, is there a correlation between RVR and the incidence of incursions? How many airports actually have ground radar? I understand that it, too, has problems in the rain.

-- Vik Bajaj, July 2, 2008

Very clever idea, and I bet we'll see more GA innovations driving airline operational concepts in the future. Some feedback for what it's worth:

-- Are you proposing a short-term fix before NextGen becomes reality, or a long-term solution that's part of (or in place of) NextGen? See section 2.6 of the FAA's conops (PDF file)

-- In your "what could go wrong" list, I suggest adding:
---- "controller misinterprets pilot's replies" (contributed to Tenerife)
---- "controller forgets about an instruction" (happened at LAX, USAir 1493)
---- "controller gets confused about where the aircraft is" (contributed to incursions at Detroit, NW1482 and Lexington, Delta 5191)

-- Regarding Trenton's comment, after landing I think most jet engines need some cooldown time before shutting down, so it might be more economical to self-taxi to the gate. Plus, powered nosewheels might soon replace tugs.

-- Andrew Schrauben, July 2, 2008
Many of these problems also relate to other problems being worked out in traffic control.

A simple first step would be a constant digital stream indicating what runways and taxiways are clear or in use, and who they are in use by. If the gps/computer doesn't get a signal that a runway is cleared for taxi, it will flash occupied, and sound a warning if the aircraft taxis towards it.

If the ATC forgets to clear the runway, the pilot will call and ask "when can I use that runway, I'm waiting..." If the ATC marks a runway as clear when traffic is landing, you could get an incursion, though ideally a system should know what runways have been assigned to traffic, and will create a warning if the ATC tries to assign them to another aircraft.

Fancy traffic control systems involve a "reservation system" where each vehicle gets a reservation to be in a particular place in space at a particular time. Vehicles ask for reservations, and get them, and then warn if for some reason they fall behind and won't be able to use the reservation. You don't go where you are not the reserved owner, nor head on a vector that will put you somewhere you aren't the reserved owner.

For ground vehicles, systems have been designed that even let you run a high-speed road intersection at grade (no overpass) where vehicles zoom through at the same time, each in their reserved spot. You slow down or speed up to assure you go through the intersection only in your window. That's for computer controlled vehicles, though.

You make most of it automatic. A plane is cleared to land, it owns the runway around its expected landing time. Ditto cleared for takeoff. For taxi, it owns a smaller moving rectangle. Any attempt to have two vehicles in the same spot is a red flag.

-- Brad Templeton, July 3, 2008

I suggest the use of optical encoding on the surface of the runways and taxi areas itself. A simple low-resolution camera on the underside of the plane (or even beside the pilot's window in a smaller plane) would suffice.

Maintenance would be minimal (just paint) and there is so much ground that high tolerances (read: dpi of resolution) would not be required.

That way, the pilot could know where he was to be at all times, and an alarm could sound if the optical encoding indicated that he was on the wrong section of road.

As well, any other equipment traveling over the surface could also read the same information.

-- Patrick Giagnocavo, July 6, 2008

Add a comment | Add a link