The Air France crash in Toronto: Let’s all practice go-arounds

Friends have been asking me for an explanation of the Air France Airbus crash in Toronto yesterday.  It is tough to say without a full accident report.  Whatever the report concludes the accident will underscore the important point that a pilot should never commit to a landing.  There is a temptation in aviation, especially after a long flight, to conduct the approach and landing as though it is inevitable that the plane must continue to descend and stop on the runway.  Unless you’ve run out of gas or suffered some sort of catastrophic engine failure in a single-engine plane, however, there is nothing inevitable about continuing the approach and landing.  A pilot can at any time make the decision to add power and go around.

When does it make sense to go around?  The airlines have rules about stabilized approaches.  The plane should be at the right airspeed, at the right position laterally and vertically, at the right descent rate, and configured for landing when it is still about 1000′ above the runway.  If not within tolerances, the pilots are directed to go around and, at least in training, they always adhere to these rules.  A light airplane is more subject to turbulence and has less inertia and therefore is easier to adjust.  The tolerances can probably be relaxed a bit and the deck lowered to about 500′ for a single-engine four- or six-seater.

Suppose that the winds are super gusty over an entire region, as is common here in the Northeast.  Or suppose that you’ve just flown many hours and are tired.  Or suppose that there are thunderstorms and it isn’t obvious where you could go that would be totally safe to land.  The rules get bent.  In general not nearly enough go-arounds are made by either small or big airplane pilots.  Usually the decision to fix a bad approach doesn’t result in an accident.  And indeed in Toronto these guys might have been okay if only they’d been on the 11,000′ runway 23 instead of the 9000′ 24L (the runways are oriented facing magnetic direction 230 and 240, i.e., almost the same, so it is unclear why the Airbus wasn’t assigned the longer runway).  They might also have been okay if they hadn’t been in the famously underpowered A340-300 (the newer versions of this plane have almost twice as much power) and therefore requiring more runway remaining before going around.

The cause of the crash might yet prove to have nothing to do with the pilots’ decisions.  Nonetheless it is probably a good idea for all of us who fly to remember to keep a hand on the throttle and tell ourselves that this is only an approach and that we won’t commit to the landing until we’re very nearly stopped on the runway.

12 thoughts on “The Air France crash in Toronto: Let’s all practice go-arounds

  1. We obviously don’t know all the facts, but they seemed to be on a stabilized approach, at least before they were handed off to the tower (from ATC feed). Seems it could have been wind shear or maybe they kept too much speed because of the nearby thunderstorms? But major airports have windshear detectors surrounding them so could probably rule that out. In a jet, where power changes take a lot longer to take effect, there must be a small window of opportunity to decide on the go around before its too late to do. Imagine you’ve touched down (assuming first 1/3 of the runway), I’d bet that once you’re on the last 40%-50% of the remaining runway, a go around wouldn’t be possible. But you would think their checklists would list target airspeeds to be for a given amount of remaining runway. Plus one would think that there would be a requirement to touchdown within so many feet of +/- the touchdown marker or go around. 9000′ is pretty long. Glad everyone made it out ok.

  2. I’ve heard news mention something about a failure of the electrical system just before the touchdown. Would it make the plane unable to perform a missed approach?

  3. “the electrical system” doesn’t really make sense as a concept for an airplane. Even my simple little four-seat plane has two alternators, two batteries, two buses. An airliner has a lot more redundancy and a fly-by-wire airliner like the A340 yet more. It would take a tremendous number of failures before the pilots of an A340 would be unable to add power and operate the control surfaces. The fact that the cabin lighting might have failed doesn’t say much. We’ll know more when they retrieve the black boxes and interview the pilots.

    Most of these accidents are blamed on pilot error of some sort. And in a sense the blame is fair because if the pilot had made different decisions the plane wouldn’t have crashed. On the other hand the standard the pilots are held to is often unreasonable: a perfect pilot, perfectly rested, with perfect knowledge of the future, would have done something different and therefore not crashed. In this case the obvious things that they could have done were divert to Montreal, or, assuming that there was not some catastrophic failure of the Airbus, execute a go-around and try again at Toronto or go somewhere else. But if after 100 years of aviation we know that not enough go-arounds are happening then maybe we need to accept that this is the best humans can do and it is time to change the procedures and training.

    What could one change? You could have a totally fresh pilot and copilot on board, sleeping in the back. They wake up 1 hour before landing and have a cup of coffee. Then take their seats in the front of the plane 30 minutes before landing. Now you have guys and gals who are 100% fresh doing the landing. They would presumably be pretty happy about going around because they might enjoy flying a bit longer. Given that copilots can be hired for $18,000 per year and captains for less than $40,000 per year this wouldn’t be out of the question for airlines (though they would miss having those two extra seats to sell).

    Hardware junkies could install some more stuff in the cockpit watching the plane’s position on the runway, the ground speed, the runway remaining, the deceleration rate, and shouting “go around” into the pilots’ headsets if it seemed like a good idea to the computer.

    But instead of proposing changes like this most accident reports boil down to “it is a shame that these guys weren’t superhuman and made a big mistake after flying for 15,000 hours in their life and 10 hours on the day of the accident.”

  4. I agree in principle with keeping all your options open and not unnecessarily ruling any options out due to being stubborn or tired. However, I believe you’ve grossly over-generalized the options of pilots. For example, if you encounter any problems such as extreme wind sheer during the landing rollout, as is speculated in the Air France case yesterday, and you don’t have enough runway to takeoff, your options seem to be to either overrun at low speed or overrun at a high speed; I’d opt to continue the landing.

    As far as your assertion that except for running out of fuel or losing your only engine, “there is nothing inevitable about continuing the approach and landing,” how do you initiate a go-around on one engine in a twin-engine aircraft below Vmc? As soon as you reduce power in your operable engine to lose airspeed to land and the airspeed decreases below Vmc; you don’t have the option of a go-around/missed approach — the aircraft is going to go down either in the existing configuration or wherever you roll and crash it into if you add power and try to go-around.

    If you want to consider more scenarios where it’s not quite cut and dry if a go-around is the “always available” best option, what if you’re flying in IMC into an airport (elev. 8000) where the MAP is a mile or two from the runway threshold, the runway is 1.5 miles long, and three quarters of the way down the runway, you’ve just touched down and begin to think that that you might overrun. You’re now 2-3 miles past the MAP and you look down and see the missed approach procedure is a climbing 140-degree right turn to 14,000 to avoid obscured terrain that’s directly ahead.

    It’s easy to play the “he should’ve gone missed” card, but if flying was so easy, there probably wouldn’t be so many crashes.

  5. I have the vast experience of sitting one time in the jumpseat of a KC-135R, an airplane with four wing mounted engines, like an A340 but smaller, while it landed in a strong gusty crosswind. I recall the pilots huge concern with dragging an engine pod. I have to wonder if a gust caught them and caused an engine pod to drag, which would have the effect of pointing them at the weeds.

    I heard today that the pilots made sure they were the last persons on the plane before they left. Bravo.

  6. Jim Howard,
    When I was working in the windtunnel business more than 5 years ago we had an aircraft incident page up to chase sleep on long runs. What I remember most vividly was a Jal(?) cargo 745 landing at Hongkong in a typhoon crosswind. The series of photos show him crabbing at about 45 degrees to centerline just before touchdown. After touchdown he drags a pod quite damatically. He made it to full stop without further damage. I was amazed that a heavy could do that. It would have made an interesting landing for passengers, if any had been aboard.
    Other pictures of 747s cranking hard over the checkerboard were interesting too.
    Lots of other photos where guys made the wrong decision that bent equipment. The site did not show photos of fatals.

  7. Bud: Captains for major airlines usually make pretty big bucks. But that is because they have been good at negotiating union contracts not because it costs more than $18,000 to hire a competent jet pilot or $40,000 for an experienced and competent jet pilot. The captains at American Eagle earn as little as $40,000 per year yet they do a great job and fly an airplane (Embraer Regional Jet) that requires substantially the same skills as flying the Boeings and Airbuses. If not for all these international restraints on the airline trade they would probably all be like the cruise ships. You’d have Filipinos working the cabin for $2 per hour and South Africans, Australians, and New Zealanders flying the planes for $30,000 per year and everyone would retreat to Mexico at night to sleep so that they didn’t have to have green cards.

    Daniel: I’m not sure what you mean by “wind shear during landing rollout”. They had a ground speed just before touching down and, if the approach was safe to continue, the ground speed would have been low enough that they could have braked to a stop without reverse thrust (always a bonus in airline operations; never taken for granted that you have it). If the wind shifted after touchdown it shouldn’t have affected their ground speed much because they were rolling down the runway. Wind shear has a big effect on airspeed but ground speed is pretty stable due to inertia. And yes you’re right that there are some situations in low-powered piston twins where you don’t have many options. But the world of airliners is the world of power. American Airlines demonstrated that the Boeing 757 could do single-engine climb outs from Cusco, Peru at 11,000′ pressure altitude and maybe 13,000′ density altitude.

  8. Small Thought,
    After seeing this crash scene live on TV and an fire tender spraying foam on relatively small flames, why was the fire crash teams not able to extinguish the fire.
    The fire seemed to grow and engulf the whole plane even though foam was being poured on.

  9. What if I was to tell you I did business with a specialized contractor who worked on the runway after the accident to repair a crushed storm drain that was known about before the accident. Due to the high volume of rain, the runway ‘flooded’ causing the airplane to hydroplane. At this point, Im not even sure if this is widely known. I’m in the process of researching for myself right now.

  10. A suit has been filed filed on behalf of all passengers on board by representative plaintiff Suzanne Deak to the Ontario Superior Court of Justice. The attorneys representing Deak and the passengers are Gary R. Will and Paul Miller from Will Barristers in Toronto. The accusers are seeking payments for general and aggravated damages of about $75 million, and payments for special damages ($250 million).

Comments are closed.