More Chilling Details: The Close Call At SFO Keeps Getting Closer

Filed Under: Air Canada

As I first wrote about a week ago, on Friday, July 7, 2017, an Air Canada flight had an incident at SFO. The A320 was flying from Toronto to San Francisco, and accidentally lined up with the taxiway instead of the runway. To make matters worse, there were four planes on the taxiway that were waiting to take off (a United 787 headed to Singapore, a Philippine Airlines A340 headed to Manila, a United 787 headed to Sydney, and a United 737 headed to Orlando), so you can imagine how much fuel they had.

The Air Canada pilots were clearly confused, because on final approach they asked air traffic control to confirm that the runway was clear, because they saw lights on it. Air traffic control confirmed the runway was clear. The Air Canada plane only realized it was about to land on the taxiway when the pilots of one of the planes waiting for takeoff told ATC what was going on.

One thing we didn’t exactly know shortly after the incident is how close the planes actually got to one another. We had heard that the plane initiated its go around at roughly 400 feet, so it was my assumption that this was as close as the planes got to one another. That’s close, but as it turns out, this was a much closer call than that.

A couple of days ago I wrote about how the NTSB estimated that the plane overflew the first two planes on the taxiway by just 100 feet. Well, there are even more details now, just when I thought this couldn’t have been any closer of a call. The East Bay Times has some even more alarming data on the incident:

New data obtained exclusively by this news organization add to the picture, showing that the Air Canada plane was just flying over a second fully loaded Philippine Airlines jet at 106 feet in the air — still continuing its descent — when an SFO air traffic controller finally warned him to abort his landing. The Transportation Safety Board of Canada said in its initial report that the Air Canada pilot did not begin his “go-around” until the air traffic controller told the pilot to pull up. It took a flight crew member from a jet on the taxiway to alert both the pilot and air traffic controller over the radio of the wayward Airbus 320.

Once past the second plane, the Air Canada jet continued to drop to as low as 81 feet before it began to climb, as aviation experts say such a late aborted landing takes a moment to stop the jet’s inertia and begin to ascend.

At the Air Canada flight’s lowest point of 81 feet — and headed straight for the third plane on the ground, United Airlines flight 863 — it was only 26 feet above the top of that airplane’s tail. A Boeing 787 is 55 feet tall.

The tail height of an A340 (which is what the Philippine Airlines plane was) is 57 feet, meaning that if this data is true, AC759 was less than 50 feet above the A340. From there it descended even further, to the point that it was 81 feet off the ground as it was headed straight for a fully loaded 787.

Here’s an excellent recreation of the situation, which frankly gives me chills (my apologies that this seems to be a video player with annoying ads, but it’s the best recreation I’ve seen):

So, what’s the likely cause here? Obviously the NTSB is doing a full investigation, and I’m sure we’ll have a lot more information soon. One potential factor is that there were two parallel runways, 28L and 28R, and AC759 were supposed to land on runway 28R. However, runway 28L was closed for the night, with the lights off.

So it’s possible that they mistook runway 28R for runway 28L, and mistook the taxiway to the right of the runway for runway 28R. In other words, they saw two parallel sets of lights, and (incorrectly) assumed that the set of lights on the right was runway 28R.

The issue is that runway lights and taxiway lights are completely different colors, so how could two experienced pilots make that mistake? I’m sure we’ll find out soon, but keep in mind that these pilots were coming from Toronto, and the plane was landing minutes before midnight Pacific Time, so it was 3AM for them. I imagine they were exhausted. We don’t know whether or not that’s directly related, but it certainly could be.

What a situation. Every time I think that this couldn’t possibly get closer, more details emerge.

The pilot who spoke up and asked what the pilots of AC759 were doing deserves some sort of a medal. One could only imagine could have happened if he hadn’t spoken up.

(Tip of the hat to Kay K)

  1. “but keep in mind that these pilots were coming from Toronto, and the plane was landing minutes before midnight Pacific Time, so it was 3AM for them. I imagine they were exhausted. We don’t know whether or not that’s directly related, but it certainly could be.”

    This is a thoroughly irresponsible statement. I’ve dated a 747 pilot previously, and what you’ve described is easily a fire-able offense.

  2. @ henry LAX — Sorry, what’s irresponsible? I’m in no way making excuses for what happened, but rather saying that fatigue could have been a factor. Some of the worst air disasters in history have been attributed to pilot error, including fatigue.

  3. Look like a UNITED pilot saved a lot of lives.

    Make that clear next time there’s whining about 1K status.

  4. My father is a retired air traffic controller and he says that complacency is usually to blame in these close calls. Even so, it’s hard to imagine how this got by both pilots in the cockpit.

  5. Chills down my spine watching and listening to that. Further reinforces “if you see something, say something”.

  6. Pilots around the world fly at all hours of the day or night on their body clocks. It’s more a matter of how long they have been working at a time rather than it being 3am where they flew from. If fatigue is the issue, it’s down to airline management not time of day. It also sounds like airport layout is another possible issue here. Commercial pilots flying into major airports should know the conditions of the situation they are coming in to – which runway they are using and what the lights look like.

    Fatigue or not, middle of the night or not, this is not acceptable.

  7. Air Canada tried to pull a Harrison Ford. It’s lack of communication in the cockpit. First Officers never have the courage to speak up and correct their Captain.

  8. This would have been the second air disaster to involve a flight 759 – that was the number of a Pan Am 727 that crashed in New Orleans in 1982.

  9. Pilots aside, I’m curious to know if this will also have fallout for SFO. It was previously mentioned there’s a FlyerTalk thread wherein pilots are blasting SFO as being a pretty terrible place to land insomuch as the runway situation and the frequent shutting-off of ILS and other seemingly important systems that would make the whole environment safer.

    I hate SFO with a burning passion because I think it’s just an awful airport in addition to the horrid runway situation, sonid love to see it receive some sort of admonishment as part of the investigation.

  10. “First Officers never have the courage to speak up and correct their Captain.”
    – that’s a broad statement that was more true 30 years ago, especially when pilots mostly came from the military ( in the US) and from more hierarchical cultures.

    A series of crashes in the 70s and 80s which were the result of bad cockpit communication resulted in the implementation of FAA mandated CRM (crew resource management) emphasizing teamwork in the cockpit. Though veteran captains resisted initially, Captain Al Hayes has said it was the reason he and his crew were able to successfully land United 232 in Iowa after all flight control systems were destroyed midair. Moreover, after a string of KAL crashes, major changes and implementation of CRM turned around its record.

  11. Mavis – circadian rhythms do have an effect. We don’t disagree on it being irresponsible if that is what it was, but do know time of day plays a role, just as number of hours do. Plus, CRM as Seattle Eric mentioned above oddly has more error-prone possibility at early parts of a multi-flight team.

    There are many factors…

  12. I imagine the aircraft holding on the taxiway could have concealed the taxiway lights. Perhaps the beacon and position lights were misunderstood for runway flight somehow?

  13. Where is Air Canada in all of this? I’m frankly outraged by this and they’ve been silent. I’d like to know, among other things:

    – Is the pilot still flying right now? Clearly, he’s posing a great danger to Air Canada’s pax as well as everyone else in his vicinity.

    – If the NTSB rules this was pilot error, what actions will be taken to ensure this lazy pilot is never again allowed in a cockpit?

    – How will Air Canada make sure to not endanger thousands of innocent non-customers going forward?

    WHY are we not hearing anything from them? Why is there no pressure on them? I guess what you’re doing here is pressure, Lucky, and I sincerely appreciate that.

  14. “First Officers never have the courage to speak up and correct their Captain.”

    Thanks for the laugh, I needed that!

    @Seattle Eric – I was going to mention KAL, but you beat me to it. Great minds and all that…:)

  15. Whatever the cause, the idea that the flight crew “must have been exhausted” seems off the wall. The flight from Toronto is maybe 5 hours and started from a 3 hour time difference.

    What about international flights that are 10 or 15 hours long, starting 8 to 10 time zones away? Wouldn’t they be three times as exhausted?

    Now, maybe the judgement and acuity of this crew may have been compromised by something we don’t know. But it sure as heck isn’t a 5 hour flight. It wasn’t even delayed.

  16. @William Y

    Those (and many others) are questions that the FAA and NTSB ask in moments like this. But all of them would be very irresponsable if they give any statement before the investigation is finish, and that kind of investigations could take far more time than a few days.

  17. Back when Lucky posted this story, I contacted United to tell them to find the pilot of UAL #1 who was the guy who alerted everyone to what was happening. I have not heard back if they found him or if he’s gotten some type of commendation. He should.

  18. maybe they thought it was a runway because of the combination of taxiway lights and lights of planes waiting in a row for departure, the biggest overlook and mistake by me would be not anticipating that something like that could happen when you leave one of two parallel runways completely in the dark, airport is to blame minimum at same level as Canada’s pilots.

  19. As a 777 pilot based in SFO, a lot of these comments show total ignorance of aviation. Just because you dated a pilot or fly to Disneyland once a year doesn’t mean you have the slightest clue

  20. A small correction:

    “A couple of days ago I wrote about how the NTSB estimated”

    That report was from the TSB of Canada, not the US NTSB.

  21. As a former Air Traffic Controller who worked at SFO I have a couple of observations. First the center line distance from 28L to 28R is 750′, not a lot of separation. Looking at a map of the runways and taxiways at SFO it appears that the distance from the centerline of 28R to the centerline of taxiway C is about half that distance. The controllers do have a radar that, in my day, went out to about 15 – 20 miles on final. However, It was sometimes not easy to confirm that an airplane was lined up on the correct runway, either 28L or 28R. It would be even harder for the controller to determine if an airplane was lined up for 28R or taxiway C.

    As a side note, I would love to be a fly on the wall in the Air Canada Flight Management office when this gets discussed.

  22. The “he’s on the taxiway” was said so casually, for a comment that probably saved hundreds of lives. Maybe it’s just the movies, but you’d think you’d have a bit more urgency in your voice! Still, who cares how he did it, he did it.

  23. As an aerospace medicine physician with an interest in fatigue, this is a concern that is often on our minds. I agree with the assessments that fatigue likely played a role in this near-miss (along with other factors such as 28L being closed and unlit). As noted, coming from Toronto, the pilots were landing at ~3am their time. They were close to their circadian nadir in terms of alertness (generally, alertness is lowest in the middle of the night, at approximately 4-6am). I don’t aim to make excuses for the pilots, but rather to point out that there are extenuating circumstances to take into account when considering the whole picture.

  24. Warren,

    You may be right but you need to explain why. And it’s not unusual for pilots to exonerate each other.


    No disrespect or nothing, but let the inspectors do their job.

  25. I recall reading in the local newspaper that there was only one air traffic controller on duty at the time and that he was busy taking care of some other matters. Can there be only one controller in the tower? Are there still staffing issues post strike? could this have contributed to the slow realization of what was happening?

  26. I am a SFO based 787 pilot for UAL and would like to clear up a few things.
    1) Yes there was 1 ATC guy running ground and tower and possibly dep or arrivals. Not that uncommon for that time frame. I think you will find this at many ATC locations. However, right after this happened a new policy was implemented that does not allow a single controller after 1AM.
    2) Due to noise restrictions (and recently construction on 28L) ILS 28R is the preferred runway at night due to it being further away from San Carlos, Burlingame, and Pablo Alto area. Usually at night you are taken out into the bay between the San Mateo and Dumbarton bridges before your base vector to final. There Is a specific approach chart called the Quiet Bridge Visual (it is chart SFO 19-2 In Jeppesen if anyone cares). It is a visual approach meaning that the weather has to be 2100 ft ceiling and 5 mile visibility for this to be handed out. It doesn’t mean the ILS has be turned off though. Also there are GPS or RNAV overlay approaches that could even be plugged in to give even more situational awareness.
    If you are coming from the East you normally are given an approach called FMS Bridge Visual 28R (Chart 19-3-1). It is a GPS taylored approach that will link you up from the arrival all the way to the runway. This approach gives you both lateral and vertical guidance. Again just because it says visual does not mean components are turned off.
    28L probably was closed and the ILS was switched off which is not that uncommon.
    Sometimes you do run into an ILS being out of service for repair or maintanence but again even if it is there are usually other GPS or RNAV approaches that can be utilized. This is in US only. Other countries may not have this but seeing how US has been in the lengthy process of switching to a more satellite based approach system vs a land based like ILS the capabilities are there.
    3) I think SFO gets a bad rap. It was built on the old military style of airport configuration. There was a time that the airport toyed with building an island similar to Osaka’s Kansai airport to utilized a runway that would be further in the bay. Thus giving additional spacing, and dealing with noise restrictions but that was shot down rather quickly. It can get crowded on the ramp and you may be waiting a bit to takeoff but everyone who flies out of EWR, JFK, LGA knows all about that as well. The controllers do an excellent job of pushing tin around giving the close proximity of runways , weather, and nearby airports.

    To Zero-G we are active in a current FRMS study dealing with long haul flying. Initially the SFO-SIN volunteers were given the monitors to wear and capture the data but I believe they are branching it out. I do know because of the data it has changed how we allocate crew rest breaks on the long haul segments.

  27. Lucky – any chance the report is failing to take into account the fact that field elevationat SFO is 13.1 feet? If so, that could halve the distance / near miss

  28. I’ll guess this could have been a billion dollar accident. Planes, lives, lawsuits, loss of airport use, etc.
    Possible fix to be considered: How about rows of red and green flashing lights at the beginning of taxiways and runways that shine directly at incoming flights when they are on final approach for only the last couple of minutes. Not so bright they blind a pilot but another caution in addition to white and blue runway and taxiway lights. It would probably cost a couple of million but let’s not let this happen again.

    Discussions about how and why it happened are interesting but prevention is most important at this time.

    I would love to fly with that same AC pilot at the controls, he will surely be the most aware pilot in the sky from now on. That is, unless he was drunk, color blind, or otherwise possessed.

  29. Fred there are already several lighting indications in use to properly taxi, hold short, cross and align yourself with a runway.
    Basic human error will always be there.

  30. James, Yes, I know of the various lighting indications at airports. They obviously were not enough to prevent the close call at SFO. We all know red means stop and green means go. Flashing taxiway red lights shinning towards a pilot would tell him not to go there. While flashing green lights at the beginning of a runway would indicate it’s ok to land. The control tower should have a method to turn off the green lights and turn on red lights when there suddenly becomes any reason not to land. Such as for the parallel runways when the pilot is directed to one of the runways and not the other one.

    Redundant safety systems have proven to prevent accidents. Spend a million, save a billion, and all those lives, maybe even mine or yours.

  31. @James I’m glad to hear that crew fatigue is being taken seriously and that changes have been made based on the observations!

Leave a Reply

Your email address will not be published. Required fields are marked *