Taylor Empire Airways

First published 03 October 2005.

But lazy, half-assed reporting was not.

Sunday’s Toronto Star had a nice three-page story about the September 2nd incident involving Air France 358 (web version here).  As citizens of this city are no doubt aware, the A340-313X arrived at Toronto Pearson’s Runway 24L in the midst of a severe thunderstorm.  The aircraft did not touch down until almost halfway down the 9,000ft runway, and it did not manage to stop, but ran off the end of the runway and went careening down into the Etobicoke Creek ravine.  The plane caught fire and the composite-based fuselage was almost completely destroyed, although no passengers were killed.

The Star leads off its coverage with a splashy front-page colour photo of the burning aircraft, with the bold headline "THIS COULD HAVE BEEN AVOIDED: An exclusive investigative report into this summer’s Air France crash at Pearson". CTV News also summarized the story in their web edition here.  Naturally, I was eager to see what Star reporters Scott Simmie, Kevin McGran and Catherine Porter had to say about it.  I thought I was going to hear interesting anecdotes from current A340 pilots, runway design enhancements from GTAA staff, a discussion of the typical decision tree when flying in severe weather, and speculation on what might have occurred in the cockpit during the event itself.

What I got instead was at least one full page devoted to the detailed story of two particular passengers, some idle speculation from lawyers handling the class action suit, quotes from exactly two pilots, four specific recommendations for field improvements to the airport, blasé rebuttals from a GTAA official that at least one of the runway improvements may not be effective in winter, and some very, very sloppy reporting.

I am disappointed in the Star’s coverage for a number of reasons:

1) Focusing exclusively on airport safety measures and not pilot decision support systems.  Here are the Star’s four recommendations:

A) Terminal Doppler radar.  Doppler radar allows meteorologists to see inside the storm structure and determine where wind shear is likely to occur.  This predictive function can be automated, with alerts passed to air traffic controllers when wind shear has been detected (or likelihood of it is very high).  ATC would then pass on the warning to inbound aircraft, who would then make decisions about the wisdom of an approach into such conditions.  Pearson International does not have its own Doppler radar but instead relies upon the Environment Canada doppler system based in King City, Ontario.  While this is a great suggestion and Pearson ought to have its own terminal Doppler radar in any event, it does not take rocket science to predict the occurrence of unsafe wind shear within a thunderstorm.  That assumption is routine for GA pilots whose smaller aircraft do not carry their own weather radar.  Although large commercial aircraft regularly fly into severe weather conditions, private pilots in smaller aircraft generally do not.  Why?  Because there are certain life-threatening risks associated with doing so, and you learn that in ground school before they ever let you near your very first prop-driven, piston-engined aircraft.  You don’t need Doppler radar to figure out that thunderstorms are inherently dangerous to flight.

B) Grooved runway surfaces.  Cutting small 6mm grooves, spaced 24-38cm apart, into a runway’s surface can help drain water and prevent hydroplaning.  The grooves are perpendicular to the runway’s direction of travel.  The FAA requires grooving of concrete and asphalt runway surfaces, but Transport Canada does not.  Pearson’s concrete runways were grooved at the ends in the 1970s, but its current asphalt runways are not.  Transport Canada believes the grooving technique may be effective in the summertime, but results in increased maintenance and marginal effectiveness during the winter.  Note also that TSB lead investigator Real Lavasseur has publicly cast doubt on speculation that hydroplaning was a decisive factor in the AF358 incident.  This did not stop the Star’s crack team of reporters from quoting a retired pilot and other experts who speculate that hydroplaning did occur and was in fact a decisive factor in the erosion of AF358’s braking performance.

C) Engineered Material Arresting System (EMAS).  ICAO recommends a 1000ft overrun area and the end of each runway, but not all airports have enough real estate to accomodate large overrun areas.  EMAS mitigates that limiting factor to a degree.  It is special, crushable concrete placed at the end of a runway in the overrun area, designed to crush under the weight of an aircraft, but support the weight of emergency vehicles.  EMAS slows down an overrunning aircraft rapidly by absorbing the aircraft’s inertia while denying its wheels traction, much like a car trying to drive on loose sand.  Airports whose runways end in a body of water have been most of the early adopters for this technology, and it is credited with some impressive saves.  Since the EMAS concrete is very porous, the GTAA has questions about its effectiveness in winter.  In wintertime, the EMAS concrete is likely to fill with water and freeze, increasing its structural strength slightly and raising its weight threshold.

D) Covering over the Etobicoke Creek.  Putting any kind of effective overrun area (EMAS or otherwise) on the 06-24 runways would involve covering over a portion of the Etobicoke Creek ravine.  What the Star doesn’t mention is that covering the ravine and installing a 1000ft overrun area without EMAS would be a big-ticket Bad Idea, as Highway 401 angles northward west of the runway end.  Any aircraft travelling fast enough to have a significant overrun on Rwy 24L may just end up on the 401 westbound collector lanes.  If we install any kind of major overrun area on the 06-24s, it has to have some kind of inherent decelerative properties, like EMAS.

In the Star’s world, the installation of these four technical solutions at Pearson International would have saved this flight.  And to its credit, the paper’s discussion of the possible field improvements is informative, largely factually accurate and relevant.  But these systems (with the exception of Doppler radar) are essentially a last-minute fail-safe.  Technical means provide some buffer from bad decisions made in the cockpit, but on their own they are utterly inadequate to prevent future tragedy.

For these fail-safe systems to come into play, your flight crew must already have shown questionable judgment and landed under marginal weather conditions… on a very wet runway… without enough room to stop.  With all due respect to the Star, air travellers generally want the avoidance of disaster to consist of more than last minute, on-the-ground airport and runway improvements.  There is no ground-based system which can definitively prevent errors in judgment.  The best flight management systems in the world are up there in the airplane with you, housed inside the skulls of the flight crew.  Getting the right information into that grey matter, at the right time, is critical to flight safety and the best way to avoid screwups.  If your flight crew makes a series of bad enough calls, even the world’s best weather detection systems and a three-mile runway with a thousand-foot EMAS overrun zone can’t help you.  You have to catch each little questionable decision long before they collectively add up to something approaching disaster.  This is generally well ahead of the point where the flight crew is staring at the rapidly approaching opposite end of the runway.

And what about other safety measures, like distance-to-go markers.  In addition to grooving the runway and installing this EMAS overrun zone stuff, wouldn’t it be a whole lot easier and cheaper to give pilots the tools to know what’s left in front of them?  Then they can make their own decisions about whether a full stop or touch-and-go is viable.  Why make them guess and wait for the concrete embrace of Nanny GTAA’s overrun zone?  They are pilots for heaven’s sakes, trained and accustomed to multi-tasking — talking to ATC, maneuvering the aircraft, fiddling with the FMS and avionics all at the same time.  They are not your typical poorly-trained civilian driver who can barely manage to safely maneuver his or her commuter vehicle down the highway while talking on a mobile phone.

2) Non-explicit accreditation of the two quoted pilots.  Neither were Air France pilots, and neither were credited as being currently or formerly rated to fly the A340.  I’m going to excuse the first pilot because he was quoted in context as the head of the Canadian safety branch of the Air Line Pilots Association, and he was speaking about weather detection capabilities at Pearson.  Although he only flies regional jets and turboprops, he has additional knowledge and a role dealing with air safety in a professional pilots’ organisation.  That’s a reasonable linkage.  The second pilot, however, is credited only as being retired from Air Canada, and he was speaking about this particular landing incident.  Look, guys, I know you’re just mere reporters but not all pilots are created equal.  Different airlines have slightly different operating procedures, even for the same type of aircraft.  Not all pilots are qualified to fly every aircraft in the airline’s inventory.  Some pilots are captains and some are FOs (first officers).  Some have additional training duties, like checkrides and pilot qualification.  All of these things have a bearing on the relevance of their commentary to a news story like this.

For instance, my own experience in single and twin piston-engine general aviation aircraft does not make me an expert witness in terms of multi-engine turbofan-powered commercial air transports.  I know what I know, but more importantly, I know what I don’t know.  If you want to know about a particular aircraft’s performance limitations, then talk to someone qualified in that aircraft type, plain and simple.  Don’t find the first off-duty guy with a uniform and ask him to comment about flight systems, avionics and procedures he’s not familiar with.

What made the retired Air Canada pilot worth talking to?  Special training?  Involvement in a similar accident?  Familiarity with Air France’s training doctrine and procedures?  Similar tastes in the choice of neighborhod pub?  His kids play with yours in day care?  Where does his expertise, aside from being a generic ATP-rated commercial pilot, bear relevance to this incident and your story?  Anyone?  Anyone?  Bueller?

3) Failing to follow up on obvious questions.  Leave the fluffy human interest crap at home.  I’m not interested in the devout Iraqi Christian woman who prayed through the whole flight and busted her femur getting out of the burning plane.  Thrilling to some, no doubt, but leave the Reader’s Digest "Drama in Real Life" tear-jerkers for Oprah.  The whole article is supposed to be about how THIS COULD HAVE BEEN AVOIDED, remember?  Where was your editor when the three of you decided to phone in this under-researched story?  Ask some real frickin’ questions for a change:

• The Star mentions the unsourced Figaro story I mention here, and then goes on to say that the twelve-second delay in reverse-thrust deployment may have been "software related", even though Airbus’ safety circular after the incident noted no software problems.  This is the simpleton’s method of saying that all of the thrust reverser interlock conditions (wheel rotation speed, main gear compression, throttle setting, radio altitude) may not have been met.  Airbus would not have indicated any software problems because the software is not at fault — even if hydroplaning occurred, no one is going to recommend disabling the interlock criteria for wheel rotation speed — it was operating properly as designed.  Disallowing reverse thrust because not all interlock criteria are present is not a fault with the software (which can not detect a hydroplaning condition) — but with the pilots, who are supposed to assess the risk of hydroplaning long before touchdown, and adjust their landing and rollout procedures accordingly.

•     * What’s really important is not whether or not the reversers engaged twelve, ten, or even two seconds after touchdown.  Reversers will slice off, at most, 11% of the landing distance on a very wet runway.  Other conditions, such as weather and runway contamination (water, oil, etc.) can totally nullify any landing distance reduction from operational thrust reversers.  Significant tailwinds, for instance, can increase the landing rollout by as much as 24%.  Didn’t any of your pilot sources tell you?  Perhaps none of them were rated to fly the A340, were they?  So I guess they wouldn’t be able to approximate the landing weight of an A340 flying a transoceanic LFPG-CYYZ route with 309 souls (and luggage) aboard, nor estimate precisely how runway and weather conditions affect the length of the A340’s landing rollout.  You know what remedies lack of personal knowledge? Research.  Where the hell was yours?  You could at least find the last reliable METAR, and figure out how much of a tailwind there was.  Would reversers (delayed or otherwise) have made any impact at all, given the late touchdown, tailwind and poor runway conditions?

• Why did the crew elect to execute an approach when there was significant storm activity on the field?  Significant convection activity (like cumulonimbus clouds [CBs] and lightning) are reliable indicators of the presence of wind shear — i.e. rapid wind direction or velocity changes.  Wind shear is not exactly a new phenomenon to pilots, especially those with enough experience and training to hold an ATP rating and type-certification on multiple varieties of turbofan-powered airliners.  GA pilots are generally advised to stay away from an airport that is showing cumulonimbus (CB) activity within 15 nautical miles of the field.  What sort of restrictions or advisements exist for commercial aircraft?  Were any of these in effect at the time?  What are Air France’s guidelines for approaches into marginal weather.  What, if any, are the decision criteria for diversion to an alternate?

• Newer A340s, like the A340-600, carry enhanced weather radar with predictive wind shear detection capability.  These systems warn pilots as much as a minute in advance of expected wind shear activity.  Was the A340-313X fitted with these systems, or not?  What sort of weather detection capability (besides the aircrew’s Mk I Mod 0 eyeballs) does it have?  How would these systems have helped or hindered the aircrew in their decision to divert or execute a full-stop landing?

• The retired pilot indicates that landing with 40% of the runway behind you is a classic "Don’t do it" situation.  The preferred solution to a rejected landing is applying power (TOGA or take off/go around thrust), executing a touch-and-go followed by the missed approach procedure.  Since an aircrew cannot see how much runway lies behind them, how do they know how much runway remains in front?  At what point does the landing become rejected and TOGA thrust applied?  What decision factors lead an aircrew to make this call?  Does Air France include these scenarios in routine simulator drill?  Do any other airlines?  How frequently do they practice it?  How many flights, on average, ever execute a touch-and-go?

• One of the lawyers suing the airline speculates the plane was low on fuel and had to land or risk fuel exhaustion.  What is the basis for this claim?  Has the lawyer ever heard of a thing called "reserve fuel"?  Does the laywer even know the regulations for flight and fuel planning for commercial transoceanic flight?  Has Air France or any of its aircrews violated any of these fuel planning or must-divert regulations in the recent past?

• The Star mentions the loss of the anemometer, which gives wind speed and direction information to air traffic control (and hence also to the pilots).  Did air traffic control pass on the anemometer failure to Air France 358, and how would this have affected the aircrew’s land-or-divert decision?  What level of equipment failure is sufficient to jeopardise flight safety and/or shut down the runway?  What are Air France’s regulations for landing at fields in severe weather that have non-operational weather detection gear?

I am just a plebian, non-ATP-rated civilian but I have managed to answer a few of these questions on my own, here on this blog.  I could literally write volumes about this stuff.  If I somehow — horror of horrors — looked upon blogging as a journalistic effort I might even be motivated to write more, but I’m not paid to do this and frankly the idea of being a reporter does not and never will excite me.  I like to do things, not write stories about other guys doing things.  Most of my pilot pals fly GA (general aviation), and the commercial guys I know mostly fly Boeings.  And none of them work for Air France, so I will never be able to answer all the airline-specific questions.

But some folks do get paid to write stories, and among those people are the ones who forgot to ask all sorts of pertinent questions in the course of this so-called investigation.  In the Star’s case, it took not one but three J-school grads to forget or avoid asking all of these questions.  Yet they still managed to milk this weak article in a major Canadian daily for a full three pages, and many other media outlets picked it up and ran with it — without asking questions of their own.

I stopped reading the Star a while ago, because sometimes while scanning a half-baked, poorly-researched story I could actually feel a net loss of brain cells.  News coverage is supposed to inform the public, not make us dumber for having read it.

See also:

• Good Landing? (3)
• Toronto’s 2008 budget: balanced or bailout? (0)
• T&C Calendar modifications (0)
• Porter Airlines (9)
• Neighborhood of Make-Believe (0)

First published 04 August 2005.

A good landing is one that you can walk away from.
A great landing is one where you can use the airplane again.

– Anonymous

By now we’ve all seen and heard about Air France 358, the Paris-to-Toronto transoceanic flight that skidded off a runway at Toronto’s Pearson International Airport, bursting into flames a hundred feet or so away from a busy section of Highway 401.  Miraculously, out of 309 passengers and crew, there were only 43 casualties (minor injuries only) — and no fatalities.  We may even applaud the quick reaction time of the aircrew and cabin attendants who managed to get everyone off the plane, safely.  (Click here and scroll down a bit to see some truly harrowing amateur photos of people evacuating the burning aircraft.)

Sadly, the media did not perform nearly so well.  The aviation “experts” called in to comment for Global News and Pulse24 were uniformly dismal, and the questions lobbed at them by the anchors were positively asinine.  At first they could not even get the aircraft type and airline right.  Then everyone was focusing on mechanical factors, like problems with the aircraft, problems due to lightning strikes, type of aircraft and its safety record, et cetera.  Broadly speaking these are the sorts of problems envisioned by the general public and the news media.  Pilots and real aviation experts think about much, much more.  Like weather conditions on the field, possible alternates, crew resource management, and the aircrew’s decision tree that led up to such an incident.

Worst of all though, the “experts” did not add any value to the commentary provided by the anchors.  Here’s a clue, guys.  You are asked to be on these programs because you supposedly add value and insight to the discussion.  If you are unwilling to stick your neck out there and make informed suppositions, on the assumption that the following TSB official investigation will leave you looking like a fool, then perhaps you ought to decline the invitation to appear on TV.  If you are unable to make informed suppositions, because you did not prepare adequately, or your reservoir of knowledge is laughably shallow, then again, perhaps you ought to decline these sorts of offers.  You should be able to provide significantly better insight into a situation than the all-purpose evening news anchor — if you can’t, find another line of work.

So let’s take a moment and dispel some of the casually stupid questions that crop up during these sorts of incidents.

Q.  Was there a mechanical problem with the aircraft?

A.   No one can say for certain right now until the CVR (cockpit voice recorder) and FDR (flight data recorder) are examined, but I highly doubt it. The pilot didn’t report any mechanical failures to air traffic control, nor did his approach indicate that the aircraft was suffering a serious malfunction.  Keep in mind that runway overruns happen all the time, and usually they result in minor damage to the landing gear.  The only reason this became a dramatic fire and total hull loss is because the Etobicoke Creek runs right through the west side of the airport, a few hundred metres off the end of Runways 24L and 24R.  The plane went barrelling into the ravine and suffered fuselage fractures and probably severed fuel lines.  If there had been straight, level ground for 300-400m off the runway end, then the aircraft would likely be intact instead of a total writeoff.  Of course creating straight, level ground means filling in (or building over) the environmentally sensitive Etobicoke Creek, and you can imagine how popular that move would be with our local tree-huggers and politicians.

Q.  Could lightning be a factor?

A.  Not likely.  In fact, the odds are that most airliners in North America will be hit by lightning about once a year. Given how frequently civil airliners operate near or around severe weather, lightning strikes are not rare; but serious incidents as a result of lightning strikes are rare.  Back in the olden days of jet travel — December 8, 1962 — lightning struck Pan American 214, a Boeing 707-121 in a holding pattern over Elkton, Maryland.   The lightning strike caused a spark which ignited fuel vapour in the fuel tanks, and the explosion brought the aircraft down, killing all 81 people aboard.  This accident led to all sorts of regulations requiring that aircraft manufacturers ensure sparks could not ignite fuel vapour in tanks or fuel lines.  It also led to the installation of lightning diverters, designed to deflect the conducted electricity away from the people and flight-critical systems in the fuselage, out toward the wingtips where it can be safely dissipated.

Q.  What about passenger reports of a lightning strike and the lights going off just before touchdown?

A.  It could certainly indicate a problem, or it could indicate routine actions taken belatedly. Cabin lights are routinely dimmed prior to takeoff and landing, most especially in the early morning or late evening.  Takeoff and landing are the most dangerous phases of flight (not counting the part where the flight crew breaks out the laptops and starts playing air combat games).  If an accident is going to occur it is most likely to occur during those phases of flight, and so the cabin crew dims the lights so that everyone’s eyes become used to reduced light levels.  Why?  Well, if an accident happens it’s safe to say that the aircraft will probably not be producing power, and lights (aside from those dim aisle emergency lights) are out of the question.  The crew want your eyes to adjust in case you have to evacuate the aircraft in conditions of low light or darkness.

This aircraft was certainly not experiencing total power loss.  An aircraft without power would not be able to deploy the thrust reversers on its engines, and if you look closely at this photo from Airliners.net, you can see that the thrust reverser on the portside engine is fully deployed. Look for the rust-coloured section on the outboard engine nacelle.  And this, by the way, is what deployed thrust reversers look like on a fully functional A340.

Q.  There are reports of the aircraft weaving after touchdown.  What might cause this?

A.  Hydroplaning, excessive steering correction by the pilot, loss of several tires on one side of the main landing gear trucks. If this plane weaved at all while on the runway, my gut says “hydroplaning”.  But first, take a look at this photo and tell me if it seems like the plane was weaving.  The nose wheel tire tracks exit the runway just a little bit off the runway centerline.  That’s a pretty good steering job for a plane that was supposedly without power, weaving and bumping and going all over the place.  It looks like she didn’t start weaving until after she’d shot off the end of the runway… And that’s perfectly natural, because there is no centerline painted down into the ravine, and it’s hard to maintain a straight line over uneven terrain.

Q.  Was the pilot “rushing” due to fuel concerns?

A.  No.  Fuel planning is the most important component of flight planning. It is a complex balancing act of ensuring there is enough fuel, while at the same time trying to minimise fuel use for the sake of economy.  Fuel is the second most costly expense for an airline (number one being headcount — people), and the heavier an airplane is (whether with people, cargo, or fuel itself), the more gas it burns getting off the ground and to its destination.

Flight dispatchers and pilots use computers to calculate every variable — weather, altitude, speed, and route — and decide upon the least costly solution.  This is of course subject to certain limitations such as the desired flight schedule, passenger connections and convenience.

Then there are other considerations mandated by law.  Each flight, domestic or international, must have sufficient fuel to fly to the destination, then to an alternate, and have sufficient reserves to accomodate delays at the alternate.  This is the absolute minimum fuel planning requirement, and realistically no commercial flight leaves with only that basic amount of fuel.  Extra fuel is always added for anticipated delays or holding at the destination, taxiing, takeoff delays, delays enroute, and so on.  Additional fuel reserves are also required when undertaking transoceanic flights or crossing any body of water for longer than an hour.

Q.  Air France 358 was in a holding delay for 20 minutes waiting for the weather to clear.  How would this affect their fuel situation and decision to go to an alternate?

A.  A pilot must never fly himself into a corner, where he has only one choice of airport.  This maxim is law for the airlines. If fuel reserves decrease close to the point where an alternate airport cannot be reached (including holding delays and several landing attempts) the pilot must divert to the alternate.  He can not continue holding.

For example, Toronto Pearson (CYYZ) has zero-zero visbility so your aircraft is in a holding pattern over Simcoe VOR (YSO) waiting for approach clearance.  Meterologists are sure that the weather will lift within 10 minutes, but in 5 minutes you will no longer have sufficient fuel reserves to fly to your alternate destination, Ottawa (CYOW).  Remember, your reserve calculation includes the capability to endure the flying time, holding delays, and more than one landing attempt at Ottawa.  So what must you do?  Well, in five minutes time, you are required to leave for Ottawa.  It doesn’t matter that the passengers didn’t pay to go to Ottawa.  It doesn’t matter that in ten minutes you’ll be able to land at Pearson.  You are obligated to take the safest route, which is to divert to your alternate.  Airline pilots are not permitted to hold or delay to the point where diversion becomes impossible.

Why?  Well, let’s say you don’t go to Ottawa and remain in the hold over Simcoe.  What happens if the weather doesn’t clear over Pearson in ten minutes, or even an hour.  You’ve just gambled your passengers’ lives on a weather report, and you lost.  You don’t have the fuel to make it to your alternate, which is the only field with good weather and long enough runways to accomodate you.  You will have to try and find a clear field somewhere between Toronto and Ottawa to put yourself down, and hope that the EMS people can reach you quickly.

So we can deduce that since Air France 385 did not hightail it to an alternate, she was in no urgent fuel situation, and could have continued holding for some time longer.  Whatever reason the aircrew decided to try the approach, fuel shortage was not one of them.

Q. The Transportation Safety Board is calling this a “unique situation” because they will be able to interview all of the passengers and crew.  How unique is it?

A.  It’s not that unique at all. Remember, this is a runway overrun, and in the vast majority of overrun cases, the aircraft is in its takeoff or landing configuration and travelling slowly — 130 to 140 knots, or 241-259km/h.  It might seem fast compared to a car’s top speed, but compared to the A340’s economical cruising speed of mach 0.83 (535kts, or 991km/h), even 140kts is not exactly ripping up the pavement.  It’s entirely typical and even expected for the vast majority of passengers and crew to survive a runway overrun incident.  If the A340 had actually crashed — that is, come screaming in at VNE (velocity never exceed, or the point beyond which the aircraft may sustain structural damage), engines growling, bouncing along the runway then cartwheeling in, spraying bits of itself all over the Etobicoke Creek, then we should be surprised at the presence of survivors.  As it is, I am thankful that the Air France crew was devoted and well-trained, and made every effort to evacuate the entire passenger complement in a timely manner.  This incident is unusual in the sense that a mere runway overrun ended up in a total hull loss and massive insurance writeoff for Air France.  And that only happened because we have a ravine off the end of the runway.

Q. How safe is the Airbus A340-300?

A.  Full disclosure: I am a Boeing fan through-and-through, but there is no disputing that the A340’s record is exemplary. Airbuses, of course, use plastic and composites heavily throughout the airframe, much like the popular Canadian-built Diamond DA20 Katana .  Being a light plane, the Katana is a little less sturdy than an A340, and actually has a little indicator light that illuminates on really hot days when the airframe is too warm (squishy?) to fly!  My pilot buddies joke that Airbuses have such an indicator too, and often refer to them in a derogatory sense as “plastic jets”.  I’ve also heard air traffic controllers say in jest to A340 pilots “How about turning on those other two engines and giving me a decent rate of climb?”  My prejudice aside though, the A340 record speaks for itself.  Until August 2nd there had been no crashes involving the type, and even with this latest incident there are only three A340 hull losses:

• 20 January 1994, Air France F-GNIA, lost after maintenance-related cabin fire at Paris-Charles de Gaulle Airport (CDG).
• 24 July 2001, Sri Lankan Airlines 4R-ADD, lost due to rebel suicide attacks on Colombo-Bandaranayake International Airport (CMB).
• 03 August 2005, Air France F-GLZQ, lost after runway overrun at Toronto-Pearson International Airport (YYZ).

Out of nine air safety incidents involving A340 airframes (including those three hull losses above), there have been no fatalities.  And that ain’t bad.

Q.  So what might be a likely cause of this runway overrun, then?

A.  A combination of highly variable weather conditions on the field, and CRM (crew resource management) factors. Weather is outdoubtedly a contributor.  General aviation (GA) pilots are taught to stay well away from thunderstorms and any kind of severe weather activity.  Our planes are smaller, lighter, and less robust; we can’t climb over the cloud layer because our planes are not pressurised, and we can’t outrun the clouds because piston engines are not that fast.  Small GA planes can get tossed around the sky by the convection updrafts and downdrafts in a cumulonimbus cloud, and it is considered the height of folly to depart or arrive at an airfield that has CB activity within a few miles.  Wind direction can change at a moment’s notice (known as windshear), and microbursts can slam a plane into the ground during the critical low-speed moments of takeoff or landing.  Commercial airliners are a little more robust than your average Cessna, have better autopilot, better weather radar and windshear detection capability (a Cessna has none), and as a result are more willing to brave these dangers.

MICROBURSTS

When a microburst’s fast-moving column of cool air strikes the ground it fans out, and an aircraft encountering the strong winds of the leading edge will suffer a sudden reduction in airspeed.  Normally, this will necessitate an increase in airspeed so that you do not stall and fall out of the sky.  As the aircraft passes through the microburst and encounters the opposite side, its airspeed suddenly increases as it is hit from the rear by strong tailwinds.  At this point the smart and safe thing to do is abort the approach, firewall the throttles and climb like mad.  Oh, and get thee to an alternate, ASAP.  If a microburst happens after the approach and while in the landing phase, then you will find your indicated airspeed either drastically reduced or drastically increased, and disaster is moments away.  A suddenly decreased landing speed can mean a stall and pancaking hard into the runway.  An increased landing speed could mean landing long (floating airborne down the runway a little longer than you intended to), broken tires, frantic braking, and possibly overrunning the runway (hey, does that sound familiar?).  Neither result is a happy one.

HYDROPLANING

Another big issue is hydroplaning.  Hydroplaning occurs when your tires move too fast across a wet surface; fast enough that they do not have sufficient time to channel away the water and moisture from the center of the tire.  The tires then ride on the water, not the road (or runway), and traction is lost.  Every driver knows the dangers of it, and believe it or not, even something as big and heavy as an aircraft can hydroplane very easily on a wet runway.  Those big white runway stripes and markings, for instance, usually become as slick as greased glass when they are wet.  Pilots use more caution on a wet runway, even when performing low-speed maneuvers like taxiing.  Many pilots are taught to try and plunk the plane down a little harder on a wet runway.  Not so hard that you bounce back into the air, mind you.  Some instructors will tell you not to “grease it in” as smoothly for a nice bump-less landing like you would on a dry runway.  The objective is to break through the thin layer of water and grease and have the wheels make definitive contact with the runway.  This method has its advantages, but can be tough on the tires.  You don’t want to land so hard that you suffer a blowout, either.

WET RUNWAY PERFORMANCE

Everybody knows that your car’s braking performance is substantially reduced in wet-weather conditions, and this is especially true of aircraft.  Storms impose a double hazard on aircrews; very often you will have to increase airspeed to deal with buffets and gales on the approach, and at the same time your braking efficiency is lessened by the wet runway conditions.  If you land long (that is, make contact with the runway well past the touchdown zone) under wet conditions then your increased airspeed and reduced braking power may leave you sailing off the end of the runway.  You should always have a touchdown area mentally selected, based upon the wet runway performance of your aircraft type.  If you do not touch down early enough, then you have no choice but to apply power and execute a “touch-and-go”, climbing out on the missed approach procedure.  It’s a lot safer than plunking it down with half or two-thirds of the runway behind you.  It may also be time to think about going to an alternate with better weather conditions, although it would be wiser if you made this decision before beginning the actual approach.

CREW RESOURCE MANAGEMENT (CRM)

CRM issues could include the cockpit environment, how well the flight crew functioned as a team, and how well they dealt with differences of opinion or methodology.  For instance, on January 13, 1982, Air Florida 190 was improperly de-iced at Washington National (now Reagan National) Airport.  Ice built up on the wing surfaces and blocked some of the critical pitot-static instruments (which measure things like airspeed and altitude).  As a result, the 737’s airspeed indications were not correct, and it took a longer than average time to accelerate down the runway.  The first officer expressed concern to the captain that something was “not right” no less than four times, and yet the captain dismissed these concerns and continued with the takeoff.  The aircraft stalled on takeoff 0.75nm from the end of the runway, struck the 14th Street bridge over the Potomac, killing 75 of the 79 people on board and 4 people on the ground.

It is critical that a flight crew not only be technically competent, but that they have good people skills.  They need to be able to communicate well with each other, and reason well, even when they disagree on the proper course of action in a given situation.  The captain, although he has overall responsibility for the passengers and crew, is not always correct in matters of flying.  Sometimes he (or she) will miss something and another member of the aircrew or cabin crew will detect a problem.  The crew’s ability to assess the information, discuss solutions, and act, is critical.  It can literally make the difference between losing an aircraft with all souls aboard, and saving them in the nick of time.

A textbook example of good aircrew communication is United Airlines 232 on July 19, 1989.  A little over an hour after departure from Denver, Colorado, the center engine’s fan blade disintegrated, severing hydraulics lines to the rudder and ailerons needed to maneuver the plane.  In just 34 minutes the crew recruited a fourth pilot flying as passenger, devised a strategy for bringing the plane under control, assessed the damage, chose a landing site, and prepared the crew and passengers for the crash.  If you look at a transcript of the CVR (cockpit voice recorder), this crew communicated a lot.  As much as once a second, at points.  They prioritized their work well, kept each other aware of unfolding events and decisions, and most importantly accepted and acted on input from junior crew members.   They even joked around a fair bit, despite the stress.  Although they started to lose control of the aircraft 100 feet above the ground and just shy of the runway, their fiery crash-landing at Sioux City ultimately saved 185 of the 296 people on board.

We know from Air France’s statement that the 57-year-old captain logged 15,000 flight hours, 1,800 of them on the Airbus A340.  The 43-year-old first officer had 10,700 flight hours, 2,500 of which were on Airbus A340.  They were certainly experienced and technically proficient.  So how does a technically proficient crew end up coming in at a high landing speed and overrunning the runway?  The big question is, how well did they work together?  Did one of them notice the high speed, and mention it?  Air traffic controllers certainly did.

ARMCHAIR AVIATOR

So let’s take a look at what would normally happen during an approach into Pearson’s Runway 24L, and throw in the information that we do know, or can glean from sites like the Aviation Safety Network.

AF358 would have been in contact with Toronto Centre (CZYZ), the Flight Information Region (or FIR, equivalent to a U.S. ARTCC) responsible for monitoring and advising enroute air traffic flowing through its territory — which is basically the province of Ontario.  AF358’s flight plan would normally include a pre-defined jet route (above 18,000ft) and Victor airway (below 18,000ft) into the Simcoe (YSO) VOR.  From Simcoe she would have used a STAR (standard terminal arrival) procedure which is basically a pre-defined route and altitude steps to fly in order to arrive at the desired runway.  STARs are published for most major airports and simplify the jobs of both pilots and controllers by making the arrival routes into an airport predictable and easy to understand.

Toronto Centre would either confirm this routing or assign a new one, and give AF358 instructions on when to commence its descent in preparation for the arrival procedure.  As a general rule, you can guess at when to begin the descent by multiplying your cruising altitude (in tens of thousands of feet) by three, and that result is the distance (in nautical miles) from the airport you should commence the descent.  Actual descent profiles for airliners vary greatly depending on the specific aerodynamic performance of the aircraft, but the “times three” guesstimate is a good rule of thumb.  We’ll assume that since she was flying a long overseas flight, her westbound cruising altitude was either flight level 350 (FL350, or 35,000ft) or flight level 390 (FL390, or 39,000ft).  This would mean that she would have to begin her descent somewhere between 105 and 117 nautical miles from Pearson (39 x 3 = 117) — well in advance of the Simcoe VOR.

The chatter would go something like this:

Toronto Centre (CZYZ): Air France 3-5-8 Heavy, direct Simcoe VOR, descend and maintain flight level 2-0-0, expect Simcoe Two arrival into Pearson.

Air France 358 (AF358): Out of flight level 3-9-0 for flight level 2-0-0, direct Simcoe on the Simcoe Two arrival, 3-5-8 Heavy.

Figure 1. SIMCOE TWO STAR (Standard Terminal Arrival)

(click to view larger image)

Once below FL230, AF358 would have been handed off to Toronto Arrivals, and she would have received a short weather briefing from ATC along with additional instructions on the STAR and her arrival runway.

The Arrivals controller would receive a METAR weather report looking like this:

CYYZ 022000Z 29011KT 4SM +TSRA BKN051TCU BKN140 23/22 A3002 RMK TCU6AC1 CB ASOCTD LTGCC VIS LWR SW-NW 2 SLP164

And he would have read it to Air France 358 something like this:

“Toronto weather, winds from 290 degrees at 11 kts, visibility 4 statute miles, heavy thunderstorms containing rain, cloud base at 5100ft with towering cumulus, cloud ceiling 14000ft, temperature 23C, dewpoint 22C, altimeter setting 30.02.  Cumulonimbus clouds with cloud to cloud lightning in vicinity of the airport, sea level air pressure is 1016.4 millibars.”

The pilot would also have pulled up the same weather report from the airport’s automated ATIS service and his own airline via the onboard computer’s ACARS system.

Toronto Arrivals would also have cleared them down to a lower altitude, like 10,000 feet, in order to cross the Simcoe VOR and begin the STAR.  Once AF358 dipped below 18,000 feet, her crew would set their altimeter to Toronto’s local setting, 30.02 inches.  At 10,000 feet they would decelerate to 250kts (about 463 km/h).  This is a sort of North-America-wide speed limit; the FAA requires it, and Nav Canada adopted it too.  There are a number of reasons for it.  The airspace below 10,000ft is where low and slow general aviation planes live; slower airplanes means less engine noise; and slower speed means a windshield bird strike is less likely to punch through the cockpit windows and end up embedded in (or at least all over) you.  And yes, that last one has happened to military aircraft, who routinely fly a lot faster than 250kts at altitudes a lot lower than 10,000ft.

Arrivals would also have briefed AF358 on her expected runway (24L), the waypoints to cross (YSO, WASIE, TULOT and LISDU), and so on.  Like the STAR indicates she would have been prepared to cross the WASIE intersection at 210kts and 7000ft, in preparation for a Rwy 24L arrival.

Runways, incidentally, are numbered according to their compass direction.  Add a zero to the end of the two runway numbers and you get a rough idea of its compass heading.  Runways 24L and 24R, for instance, have a true heading of 237 degrees, so they are rounded up to 240 and assigned the runway number 24.  Landing the opposite way, on Runway 06L or 06R, means following a rough heading of 060 (true heading 057).  The actual runway headings are, as you can see, indicated on the approach charts and the airport diagram, so there is no confusion for an aircrew arriving at an airport for the very first time.

The Aviation Safety Network report includes some details on the radio traffic following her last waypoint:

Toronto Arrivals: Air France 3-5-8 Heavy roger, 2-4 Left is your runway, the altimeter 3-0-0-0 and when you are able fly heading 2-10 and intercept the localizer.

Air France 358: When able within… five nautical miles we can intercept the localizer, Air France 3-5-8.

Figure 2. ILS Rwy 24L Approach

(click to view larger image)

The localizer is one of two parts of the ILS (instrument landing system) installed on most runways at major airports. The localizer (using VHF) provides lateral guidance, helping the pilot align the aircraft properly with the runway centerline.  The second part, the glideslope (using UHF) helps the pilot find the proper descent angle.

In the cockpit, the pilot tunes one of his NAV receivers to the localizer frequency.  A second receiver tunes the proper glideslope frequency; the pairing is automatic.  On the instrument panel, the pilot sees a little cross on one of his MFDs (multi-function displays).  A vertical bar indicates the localizer and a horizontal bar indicates the glideslope.  When both of these bars are centered in the MFD, making a plus sign or cross, the pilot knows he is tracking the ILS properly.  The autopilot can also be slaved to the ILS, allowing the aircraft to intercept and follow the ILS to landing all by itself.

In fact, the aircraft’s onboard systems are so good and so responsive to the slightest deviation that they can usually fly an ILS approach better than a human pilot.  In severe weather or low visibility, airlines usually require the aircrew to engage the autopilot and allow the aircraft to execute the ILS approach itself.  This “autoland” feature often includes post-touchdown activities like deploying spoilers and brakes to slow the aircraft down.  The auto-brake features even include settings for wet runways.

According to the Aviation Safety Network report on AF358:

The crew then received further instructions to descend to 5000 and to reduce their speed to 190 knots. About 15:55 they were cleared down to 4000 feet and one minute later the controller cleared the flight for the ILS approach to runway 24 Left. Within less than a minute the controller asked “…Air France 3-5-8 reduce speed now to 1-60 knots”, which was correctly read back. Thirty seconds later the controller radioed: “Air France 3-5-8 slow to your final approach speed”. Then, about, 15:58 they were instructed to contact the Toronto Tower: “Air France 3-5-8 contact Toronto tower at the KIREX fix on frequency 1-18 point 3-5″.

This is not exactly damning stuff — it could be significant, but it’s not out of the ordinary for ATC to issue speed instructions.  It is, of course, routine for pilots to well, know their own aircraft’s speed requirements in each phase of flight, but they will not always slow to approach speed immediately upon intercepting the ILS.  It really depends on the flight conditions, ATC’s spacing requirements, and so on.  ATC may ask them to maintain a higher speed for a few moments, or to reduce speed to allow departing or arriving aircraft ahead to clear the runway in time.

At the KIREX intersection, AF358 would have contacted Pearson Tower and confirmed they were locked on the Rwy 24L ILS.  The tower would have sent them a brief weather update and cleared them to land on Rwy 24L.

Now CAT I ILS approaches like Rwy 24L do have limits, though.  For safety reasons the crew have to be able to see the runway before they are permitted to land on it.  ILS approaches have a certain decision height (DH), and it is called that because at that point the pilot must make a go/no-go decision about whether to land.  The elevation at the touchdown zone of Rwy 24L is 547ft MSL (mean sea level), and the altimeter on board the aircraft is also set to mean sea level, so that a plane sitting on the ground at Pearson will actually indicate 569 feet (which is the actual elevation above sea level).

So, according to the Rwy 24L ILS chart above, if the crew can not see the runway by the time they have descended to 797ft MSL (mean sea level), or in other words 250 ft AGL (above ground level), they must abort the approach and execute a missed approach procedure.  The weather report said visibility was 4 statute miles, and Air France did not abort, so they must have seen the runway at the decision height, and gone ahead with the landing.

Figure 3. ILS CAT II or CAT IIIa Rwy 06L Approach

(click to view larger image)

Normal ILS approaches are rated Category I (CAT I).  In extremely low visibility, you have to use a special kind of ILS approach, Category II or Category III (CAT II or CAT III).  These types of approaches include extra equipment on the field that allows the aircraft to descend safely to a lower decision height, or in some cases (CAT IIIc), to execute a landing under zero-zero visbility conditions.  The aircraft type, the airline, and the pilot must all be certified to use these types of precision approaches.

Toronto Pearson has only one runway suitable for CAT II and CAT IIIa operations, Runway 06L.  CAT II decision height is 100ft AGL.  CAT IIIa requires a runway visibility of 700 feet at a decision height as low as zero feet, CAT IIIb requires a runway visbility range of 150 feet at a decision height as low as zero feet, and CAT IIIc allows a “blind” landing with no runway visbility and decision height minima at all.  I personally find it a little astounding that Pearson has no CAT IIIc capability, considering how bad the weather gets here, and how many planes cycle through all the time.

The selection of 06L for CAT IIIa equipment is also a little ass-backwards to me, because the prevailing winds across the north shore of Lake Ontario blow primarily west-to-east — in other words they favour the 24s and not the 06s.  This wind direction reverses in the winter, but aircraft do have to land facing into the wind, and the 06s are frequently impractical, like they were for Air France 358 on Tuesday.

I would really love to know why we don’t have CAT II or III equipment in each direction on a couple of the east-west runways, to allow for better poor-vis capability.

The north-south runways are kind of a write-off.  They are used, and they are useful, but using them generates a lot of noise complaints from the surrounding suburbs.  Don’t get me started on the futility of even listening to noise complaints from people stupid enough to buy expensive urban homes directly under the approach and departure paths of the nation’s busiest airport.  It’s not as if the airport (and noisy jets!) didn’t exist 30 or 40 years ago.

Pearson International, like all urban airports, has instituted noise abatement procedures.  This means that shortly after takeoff, at the moment they most need additional acceleration and stability, heavily-loaded domestic or international flights have to dramatically reduce throttle and engage in some maneuvering to avoid flying over the noise-sensitive real estate.  From an air safety perspective, it’s not the wisest thing to do, but we have been living with it for the past 30 years and no one has fallen out of the sky because of it yet.  Or at least not in a directly attributable way.  Who knows how many stalls and maneuvering accidents may have been prevented.  Fortunately, new turbine engines are getting quieter all the time, and real estate west of Toronto is getting more expensive all the time.

Figure 4. CYYZ Pearson Airport Diagram

(click to view larger image)

Rwy 24L indicated on the airport diagram in red.
CAT III-capable Rwy 06L/24R indicated in blue.

As history records, AF358 did touch down on 24L, albeit a little faster and a little later than expected, sliding 200-300m off the end of Rwy 24L and plunging into the Etobicoke Creek ravine. Whether a touch-and-go and missed approach should have been executed, I can’t say.  Investigators will make that known very soon, I’m sure.  For now it appears that events simply unfolded too quickly for the flight crew to really comprehend what was happening, and what their escape route should have been.

It would also be interesting to know why they were slotted in to Rwy 24L and not the longer, higher-visbility 24R next door.  Probably just a matter of timing in the cycle of landing operations.  Runway 24R is 9,697ft long — 212m longer than 24L.  If AF358 had been assigned to land on 24R instead, there’s a chance she might still be intact, with only minor damage to her landing gear.

See also:

• Air France 358 crash was avoidable (0)
• Toronto’s 2008 budget: balanced or bailout? (0)
• T&C Calendar modifications (0)
• Porter Airlines (9)
• Neighborhood of Make-Believe (0)