Stroking the Razor's Edge - A possible explanation for Air France Flt 447?

If crash investigators would like to see what sort of history AF (and other airlines') pilots have in respect of operating proximity to coffin corner, all they'd have to do is review historic FDM/FOQA data. That should make it clear whether or not there was any generic (or in the case of the AF447 captain) specific proclivity for "riding the tiger" in any misguided attempt to achieve greater economies.

French plane lost over Atlantic

Air France flight AF 447 has gone missing over the Atlantic. The flight left Rio at 2200 GMT on Sunday, and was due to land at 0910 GMT in Paris, but contact was lost at 0130 GMT, some 186 miles northeast of the Brazilian city of Natal. It had 216 passengers and 12 crew on board, including three pilots. The passengers included one infant, seven chi...

  During harsh economic times, when jobs are at risk, it's understandable that pilots are going to kick in and assist by making operations more economical. The best way for long-haul pilots to do that is to fly right up there in coffin corner despite their heavy weights, i.e. where the margin between the aerodynamic stall and mach buffet is minimal (or even negligible). That gives best engine and airframe fuel economy (air nautical miles per pound) however it's also an exercise in stroking the razor's edge. Do that above the 55,000 foot tops of the ITCZ (Inter-Tropic Convergence Zone) and your environment-related controllability hazards become very real. Even at night, you can fly around the intense cells visually (or by radar) dodging the CB tops; however it's also possible to strike Clear Air Turbulence associated with the ITCZ..... or to stumble into a CB thunderhead that you didn't see ahead (courtesy of its electrical activity and sheet lightning flashes).

If that happens, what's next? fact you don't even need to bumble into a thunderstorm's tops in order to come unstuck. All you need is some turbulence or a sudden unexpected air temperature change (affecting both airspeed and thrust output) - or an engine surge. Quite simply it becomes a sudden loss of control on a dark night..... and only a few thousand feet below you are the solid unfriendly cloud tops of the ITCZ's solid thunderstorm activity. Once inside that maelstrom of cloud below, with its severe turbulence, hail, lightning and electrical activity, a successful recovery from an unusual attitude would be unlikely. At this point you must recall that a FBW Airbus is limited in its manoeuvrability by its control laws. That feature is designed to stop pilots creating situations - however equally it can limit their ability to recover from out-of-control "unusual attitudes"  (e.g. inverted or spinning). Control Laws cannot stop cataclysmic overstress however.

A non-Airbus airliner may be able to recover from an upset that an Airbus cannot (due to the limits of its control laws).

What about the ACARS auto-transmitted message of a serious electrical failure? Well if the aircraft was being overstressed and breaking up, you'd only need a few seconds for the generated faults to self-precipitate a safety message to the Company OPS Centre via ACARS. Unfortunately, unless you have a data-dump capability (remember my Iridian/Roadshow proposal?) you don't know what led up to that message (and as for what came after? ...well, without Iridian/Roadshow, you have to recover the DFDR, QAR and CVR) for laboratory analysis.
So is Air France (and obviously, other airlines) allowing (or directing) their crews to get high early and start stroking that razor's edge whilst still at high All-Up-Weight and very vulnerable? Is the Airbus FBW design more vulnerable than others? Do crews practice tangles with coffin corner in the simulator?
Daresay all that will come out of the investigation. I somehow suspect that it will prove to be the case. i.e. Crews have been over-extending their necks, bt withy the vry best of intentions (protecting their jobs?).
Previous examples of Coffin Corner Encounters?
Roughly corresponds to the time interval that a loss of control scenario culminating in a structural failure, would take to impact (i.e. 4 minutes). The cascade of system failures is about what you'd expect from a structural failure following an an overstress stemming from a high-level loss of control (i.e. a coffin corner stall-spin entry).
"advisory regarding cabin vertical speed." tends to confirm a high rate of descent (and nil cabin pressurization because of hull rupture) towards the end of the episode.

Coffin corner (aviation) - Wikipedia, the free encyclopedia

The coffin corner or Q-Corner is the altitude at or near which an aircraft's stall speed is equal to the critical Mach number, at a given gross weight and G ... - 28k - Cached - Similar pages

Cruising in the coffin corner the red stripe is the overspeed and the yellow stripe is the low speed resulting from a high wing loading.  (to see original image with "mouse-over" capabilities)  definitions and graphics
same text in English
also see attached coffin corner imagery.
Iridian/Roadshow (1999 original and subsequent links):
Aerodynamic "Coffin Corner" Tech Ops Forum |
12 posts - 9 authors - Last post: 21 Mar 2006
Can someone explain the "coffin corner" effect and give examples of the width of the safe operating speed between Vc, Vmo and Vs ? ... - 109k - Cached - Similar pages
  This Mach/ASI shows a margin of 35 knots; what margin would one normally expect to have for a safe cruise in a high and heavy aircraft?
"what margin?" A minimum of at least twice any observed (or anticipated) indicated airspeed fluctuations. Marrying the needles is never a good idea because all it takes to "hit the wall" and strike instant controllability issues is a sudden along-track temperature rise, convective or clear air turbulence, a TCAS RA event or an engine hiccup. Engines operating at coffin corner type heights and speeds are running quite close to the edge of their performance envelopes also (their design surge line) - because of the reduced air density and mass through-flow. Turbulence can momentarily disrupt intake flows and of course any sudden yaw induced by one engine momentarily surging can also blank the other engine's intake, precipitating a dual engine stall. That's what happened in the Oct 04 Jefferson City crash of the Pinnacle Airlines CRJ200 (link).
Pilots should (and generally do) observe the weight versus height and ambient temperature guide-lines available to them as both tables and graphs in their aircraft operating manuals. However those numbers don't factor in such wild-card variables as crossing the ITCZ. The vertical instability in the atmosphere in the vicinity of that Zone is well known to long-haul pilots. But, over the space of a year, the ITCZ moves longitudinally North and south of the Equator. It also varies in intensity with the seasons. In some months and particularly on some days, like a volcano, it's much more active..... and dangerous (see attached treatment of that subject).
Some bravado may be evident where the pilots vow to "keep a close eye" on their coffin corner exposure, but then fail to do so adequately. It's just another form of conscious risk-taking and risk buffer erosion. It's very similar to (say) conducting a short-field STOL approach to (for instance) London City Airport. That's a very short duration risk exposure and so both pilots are very much on the qui vive and ready to react if things start looking border-line on finals. Coffin corner exposures (by contrast) tend to be a very long-term risk-taking venture and it would be very easy for fatigued crews to let their guard down (or relief crews taking over to just not realise how close to "the edge" they were operating).
I've no doubt that many "insiders" are already aware of this likely scenario but quite unwilling to even bring it up - beyond referring vaguely to turbulence and lightning etc. I'd expect that the well-known French reluctance to associate any incident or accident to the Airbus control laws will continue to mask this probable scenario. Because it is just a tangible threat it was always going to claim someone/sometime. It's as likely a scenario as any other that's out there at the moment. Recovery of the recorders should easily prove (or disprove) this loss of control scenario. Prior to that happening, I'd expect other AF and long-haul pilots enerally to simply "keep their own counsel" on the subject. But they'd also have quietly learnt from it.
Another thought.
If crash investigators would like to see what sort of history AF (and other airline's) pilots have in respect of operating proximity to coffin corner, all they'd have to do is review historic FDM/FOQA data. That should make it clear whether or not there was any generic (or in the case of the AF447 captain) specific proclivity for "riding the tiger" in any misguided attempt to achieve greater economies.
This A330 was 3 hrs into a flight fuelled for an 11hr trip, and level at FL350, with 228 POB. What was the likely weight and what would be the low-end buffet boundaries for that weight?  What max turbulent air penetration speed should they've been observing? What was the ISA deviation at FL350 in that area that night? Was it ISA+10? ISA+15? (i.e. warm air?). Lots of questions that should eventually be answered.
Hitting Mach buffet alone (in an intentional "Mach Run" for instance) can create "tuck" problems. In the old Vampire you could roll inverted at
Met synopsis

Tim Vasquez's report, whilst being fairly technical, concludes well. We don't at the moment know the ADD status of AF447 and what if any MEL's the a/c dispatched under. So called ETOP's dispatch with single radar, IRS fault or a prim/sec problem is feasible. Comment was made that many crews frequently attempt to fly over weather; but in an almost max weighted 330 or even 340, this would be folly. Pilots who have little experience of other types may attempt this manoeuvre but for those with L1011/747 classic/DC10 time, that would be a rare judgement call. The digi radar on the A330-200 requires more analysis and very careful gain and tilt management than earlier analogue types. The buffet thresholds at close to or even at, max recommended (or above optimum) altitude levels are also narrower than non FBW types. For these reasons, lateral wx deviation, even by track displacements of over 100nm are not uncommon within the ITCZ that I operate through. It appears through Tim's track analysis of Flt447, that no significant wx avoidance was applied. As an afterthought, wrt CRM and crew rest issues, were the senior crew resting at the time of the 'event'? What was the crew pilot complement/experience level? Was this a training sector for either the Capt or F/O?
As a senior check and trainer on the 330 and 340, I always, without exception employ two cardinal rules.
. Rough air speed protocols versus altitude limitations must be applied early on in the piece. If the air is rough, (med/light, med turb and above), select M.78 without delay. "Chunking down" of mach to finally achieve target .78 is not recommended.
. Early wx deviation and crew lockdown is key to successful wx damage avoidance. Never attempt to out-climb weather. Cherry picking the weaker cells and risking penetration of invisible gust regions, again is total folly. Keep the big picture in mind, take the additional fuel at flight planning and large track mile deviations.
The above is intended to complement Tim's fine piece and in no way wishes to undermine the severity of the event and its tragic outcome.
40,000 feet so that the M0.84 pitch-up was actually a pitch-down (towards terra firma). That ensured a continued rapid (violently pitching) height loss but eventually you'd lose Mach No in the thicker air lower down and the compressibility effects would just evaporate - after quite a wild ride, but with no hazardous portents.
Hitting both Mach buffet and the aerodynamic stall at the same time in the cruise (by marrying the needles) can give you the worst of both worlds. Autopilot will disconnect (if not already done so by the pilot) and the aircraft is likely to roll and pitch violently. This could be complicated by an already stalled engine or the yaw/high AoA is likely to induce an engine stall/loss of thrust - introducing asymmetry and compounding the loss of control challenge. As I've said before, the Airbus control laws are designed to stop the pilot from putting himself in harm's way, however nothing in those arcane control laws can stop an aircraft adopting weird and unusual attitudes during a coffin corner encounter. And there's the rub. Will the Airbus control laws help or hinder in such a scenario? You can guess my suspicions on that score.
You have to use some imagination when trying to envisage the combinational mach buffet and stall effects on the aircraft during a cruise encounter. When a wing stalls there's an airflow breakdown over the wing's upper surface - and that alone will cause the compressibility pressure wave to move rapidly and violently fore/aft. Both phenomena will change the fore-aft position of the centre of pressure, initially causing wild pitch gyrations. Inevitably any yaw component or mismatch between the wings will generate a rapidly divergent (dutch?) yaw/roll/pitch cycle. If the nose drops below the horizon then mach tuck (pitch "up" into the turn = re-stall) is a likely outcome.
The ultimate conundrum lies in the 50% chance that a pilot caught by surprise (or disorientated or suffering 
sudden vertigo) will attempt a jet upset recovery by rolling the wrong way - i.e. long way round.... (or perhaps an Airbus control law will limit his choice to one way - possibly the wrong way). He may even be opposed by his copilot attempting his own version of "recovery". What will rolling the wrong way do? Modern jets are so clean that even if the pilot got the power off, an aircraft entering an extreme nose-low attitude will quickly accelerate to very high speeds (>Vne) and the ensuing pull-out is likely to overstress (component shedding, hull rupture, loss of pressurization, zoom back up into hypoxia country? = pilot incapacitation) . Who knows. Always better to not go there. Exceed the limits and you're in test-pilot country - in fact, invading areas that you couldn't pay an experimental test pilot enough to go. Coffin Corner is well named and best avoided by sensible margins. I'm guessing that those margins will soon be re-defined. The accident A330, if it reverted to direct law, would lose most (all?) of its control law protections and thus be liable to be easily "broken" by a pilot pulling hard - but with nil experience in flying a BUS in Direct Law.
It's not definite that this was the AF447 scenario - but it's a top contender. I'd suggest to the investigators that they closely review the crash captain's (and other captains') FDM/FOQA records for their style of operations on that route (or any other equatorial transits across the ITCZ). That might be revelational. If long-haul pilots are getting high early at higher weights, then they'd be doing it for a reason. Flight plans don't count. It's what you decide to do as you burn off your gas that decides how close you're operating to your airplane's limits. Riding those limits at a low level of personal awareness is to whimsically stroke the razor's edge.

John S
An EXEMPLAR Coffin Corner Accident
Date: 22 AUG 2006
Time: 15:39
Type: Tupolev 154M
Operator: Pulkovo Aviation Enterprise
Registration: RA-85185
C/n / msn: 91A-894
First flight: 1991
Total airframe hrs: 24215
Engines: 3 Soloviev D-30KU-154-II
Crew: Fatalities: 10 / Occupants: 10
Passengers: Fatalities: 160 / Occupants: 160
Total: Fatalities: 170 / Occupants: 170
Airplane damage: Destroyed
Airplane fate: Written off (damaged beyond repair)
Location: 45 km (28.1 mls) NW of Donetsk (Ukraine) show on map
Phase: En route (ENR)
Nature: Domestic Scheduled Passenger
Departure airport: Anapa Airport (AAQ/URKA), Russia
Destination airport: Saint Petersburg-Pulkovo Airport (LED/ULLI), Russia
Flightnumber: 612
Pulkovo flight 612 departed Anapa (AAQ) for St. Petersburg (LED) at 15:05. The Tupolev climbed to the cruise altitude of 35,100 feet (10.700 m). Because of storm cells ahead, the pilot decided to change course laterally by 20 km and attempted to climb over the storm cells. However, the thunderstorm front was unusually high, extending up to 15 km (49,000 feet). The Tu-154 entered an area of severe turbulence, pushing up the airplane from 11.961 m to 12.794 m within just 10 seconds. The angle of attack increased to 46 degrees and the airspeed dropped to zero. It entered a deep stall from which the crew could not recover. The plane crashed and burned in a field.


'Climb over storm' doomed Tu-154 (Flight 05/09/06)
RIA Novosti