Monday, September 18, 2006
A Tale of Two T-Tails
Inescapably Floored by Design Flaws
When Russia's Pravda disclosed that the Pulkovo TU154 (reg: RA-85185) that crashed on Aug. 22, forty minutes after takeoff, had been trying to outclimb a thunderstorm, the air safety cognoscenti were immediately aware of a likely scenario for what had transpired. The upshot: the aircraft was doomed by the kinds of design flaws that call for eternal vigilance from air safety regulators.
The latest data from the MAK (Interstate Aviation Committee) reports that Flight 612 had step-climbed to 12,400 m (41,000ft). Maximum allowed weight at 12,100 m is 85t for the TU154. Take-off weight must have been around 93.5t, given the distance from Anapa to St. Petersburg and with 160 passengers. This would mean a weight of about 88t at the time of the crash.
The thunderstorm was reported as having been a heavy one, reaching up to 12-15 km. It looked like the aircraft had stalled and entered a flat spin when it encountered turbulence at a low indicated airspeed (IAS), because it was way too high up for its weight. This was confirmed two weeks later in a statement by the Ukrainian Attorney General.
There have been three previous fatal flat spin accidents in large jet airliners with passengers. All of them were TU154s. There's a transcript of the last one, and it's chilling. After the stall warning, and even though the crew reacted immediately, it was already too late. Apparently the Soviet authorities didn't believe in stickpushers, or even the earlier lesser measure of alerting stickshakers (apparently the TU134 lacks a stickshaker as well).
Any T-tail aircraft has a critical wing Angle of Attack (AoA) where a horizontal stabilizer mounted atop the vertical fin, and the all-important rudder also, will be blanked by the turbulent airflow off the wings. Experience shows that the resulting embedded stall outcome will produce an attitude that you won't recover from -- not without a tail- chute deployment that could change the balance of forces. The critical wing AoA for deep stall on the Trident, 727 and DC9 was quite high, around 30 to 35 degrees. The major player in determining this is the wing-to-T-tail relationship.
However, this crisis AoA can also be affected by icing buildups, engine thrust, IAS, leading and trailing edge flap, speedbrake, stabilizer position, Centre of Gravity placement, as well as any dynamics in the pitching plane -- e.g., back-stick input at entry to the condition or the rapid onset of induced drag as an autopilot's auto-trim seeks to maintain a selected altitude). As a clincher it's known that, courtesy of its tail fuel tank, the TU154 is usually flown fairly tail-heavy in the cruise. The MD-11 operates similarly as a way of minimizing aerodynamic drag. Entering a stall/spin in that aft weight-trim would be like a tsunami washing over an earthquake zone.
Deep Stalls and Bogus Stick-Pushes
A Deep Stall occurs when the turbulent wake of a stalled mainplane "blanks" the horizontal stabilizer, rendering the elevators ineffective and preventing the aircraft from recovering. For T-tails, the deep stall is the equivalent of the helicopter's vortex ring condition, as they can both generate very high rates of descent in a near normal wings- level flight attitude. It's only the relative airflow that counts. For a T-tail aircraft, the precursor to a flat spin is usually a deep stall. The British BAC One-Eleven prototype was lost on an early test flight in 1963 due to a deep stall, well before the two later Trident deep stalls.
The first crash of a 727 (United, near Chicago, August 1965) has never been properly explained and may have been a deep stall on leveling off as well. However, without any thrust, drag or lift asymmetry, there need not be any follow-on autorotation (i.e., a spin entry). The BEA Flt548 Trident deep stall at Staines UK (G-ARPI on June 18, 1972) occurred two minutes after takeoff when the leading-edge lift augmentation droop (aka slat) was prematurely retracted by the 22-year-old second officer.
Despite the stick-shaker operating its warning, this led to a series of stalls and eventually a deep stall from which recovery was impossible. The Trident stick-pusher was prone to false alarms so its actuation together with the stick-shaker did not necessarily alert pilots to a bona fide cause, such as early droop retraction. On G-ARPI, the pusher's intervention was instinctively cancelled as a nuisance warning well before it could be effective. Cardiologists found that the captain had probably had a simultaneous heart attack. Subtle pilot incapacitation wasn't in the 1960s first officer syllabus.
Stalling for (All) Time
In an earlier 1966 Trident crash, the aircraft was carrying out the first of a series of production test flights to qualify for a Series Certificate of Airworthiness. After completing a large part of the required tests, the stall tests were begun. Three approaches to stall were made to check the stall warning and stall recovery systems. The fourth stall test was made at an altitude of 11,600 ft in the landing configuration and with the stall warning and recovery systems inoperative. The Trident entered a deep stall with the nose going up to a 30� to 40� attitude. The aircraft yawed to the left, the right wing dropped, and the plane went into a flat spin to the right.
The investigation board concluded: "During a stalling test, decisive recovery action was delayed too long to prevent the aircraft from entering a superstall (deep stall) from which recovery was not possible." After that disaster the U.K. CAA identified this T-tail flight condition as a terminal one requiring incipient stage stepped alerting and avoidance measures.
A precursor/associate of the deep stall is the high sink-rate that caused many early 727 and DC9 accidents, (two in one week in Nov. 1965). A NorthWest Orient 727-251 deep- stalled fatally at Stony Point NY on Dec. 1, 1974, losing 24,000ft in 83 seconds. A Canadair Challenger prototype (C-FCRJ) was lost July 26, 1993 and its pilots killed when the aircraft entered a deep stall from a high sink-rate and the misconfigured anti-spin tail-chute simultaneously deployed and jettisoned.
In an earlier April 3, 1980 tail-chute deployment, the Challenger C-GCGR was lost when the tail-chute wouldn't jettison (and a pilot then died when his personal chute failed to deploy). Tail-chutes are now more reliable and can be drogue, spring or rocket deployed. One did save an MD-80 during that model's flight-test program.
However, these chutes aren't necessarily a panacea for larger T-tailed aircraft. Pilots being flung forward against their yokes by heavy centripetal forces in a spin might not manage to deploy a chute, and reliable auto-deployment systems would likely have an "unintended consequences" downside in the event of an accidental triggering or a failure to later jettison. Deep stall prevention is better and much easier to accomplish at the design phase.
An additional factor in a deep stall with aircraft such as the BAC1-11 and DC-9 is that the turbulent airflow from the stalled wing also blanks the engine intakes and precipitates engine stalling (i.e., an inability to develop sufficient thrust to accelerate the aircraft out of its stall). Consequently, you end up with a "mushing" high angle of attack, no pitch authority to correct that condition -- and no thrust to accelerate out of it.
The prototype BAC1-11 had tab-actuated ailerons and elevators, so it not only had no elevator authority in deep stall, but no roll capacity either. The solution for the production aircraft included not only powered rudder and elevator controls, but modified wingtips and wing roots to alter airflow, much more efficient stall warning vanes, a stick pusher and fuel dip to eliminate compressor stalling (i.e., overfuelling).
In theory, in the model 217 anyway, one could power out of a deep stall if all other measures failed to stop you reaching that condition. The BAC1-11 would show auto-ignition first, then the shake, followed by the push. Nitrogen pushed the controls via a ram piston. By the early 1970s, the U.K. CAA was acutely aware of the T-tail threat and made 727 operator DanAir fit both shakers and pushers. It was an expensive exercise, even though the 727 was FAA certified for only the shakers.
The question is now really whether, with its "three strikes and you're out" record, the Tu154 can continue to be considered airworthy in this day and age without either a stick-shaker (optional on the TU154M) or more importantly, a stick-pusher. There are still over 400 TU154 in service, with more still being produced.
The Relevance to WCA Flt 708
One further notable aspect of the recent TU154 crash at Donetsk is that the Ukraine Attorney General has released a factual report indicating that two of the three engines, probably the outboards, lost thrust as the aircraft failed in its struggle to top the thunderstorm. This highlights the other peril awaiting those climbing into "coffin corner". At those "corner of the operating envelope" heights, not only is the aircraft aerodynamically challenged by a very limited operating speed band between stall and limiting Mach number, but in the thin air the engines are running very close to their surge line (the well-known N/vT stall).
A Pinnacle airlines CRJ200 crew found that to be true when they lost both engines at a low IAS at FL410 on Oct. 14, 2004, about 22 minutes before impacting a few miles short of the runway at Jefferson City. The Pulkovo Airlines TU154 climbed into an outside air temperature of around minus 36� C rather than the standard minus 55� C for that height. At the warmer temps, the air is less dense and an engine N/vT stall is more likely.
Yet even without the unfriendly air temps, just draw an imaginary line between the wing-root leading edge and the horizontal stabilizer atop the DC-9/MD80 T-tail and you will see that the engine inlets are effectively blanked by the wing-roots whenever the T-tail is. The same turbulence that can induce a stall/spin will also readily stall an engine near to its surge line (and it will then drop to negligible thrust and
In fact, some types of "locked" compressor stalls will require engines to be shut down and relit in order to clear the stall. A final report on the Aug. 16, 2005 West Caribbean Airways Flight 708 crash of an MD-82 in Venezuela is still awaited. However, its common denominators and the familiar scenario are sufficient evidence to prima facie identify it as a deep stall event.
A Reluctance to Surrender Thrust
After taking their >6000lb overloaded jet high to avoid the weather and neglecting to turn on de-ice/anti-ice, their two MD82 tail-mount engines, one after the other, lost thrust but were still running. Both rotors showed evidence of turning at high speed upon impact. Unfortunately, once locked in a deep-stall's strangulated and stifling airflows, the engines wouldn't have recovered until gulping much denser air at the lower altitudes, shortly before impact at their descent rate of over 7000ft/min.
According to the Flight Recorder, WCA708 descended at a high angle-of-attack and this was due to two main factors. The crew foolishly trimmed the stabilizer fully nose up to the stops during their travails and the Center of Gravity was well aft due to poor load distribution. Both of these factors would have been pro deep-stalling for a T-tail design that actually doesn't have a rogue history for that vice.
Because the crew was trying to get above the weather, the captain didn't want to turn on the performance robbing anti-ice compressor bleed air systems, so there was both increased weight and a loss of airfoil efficiency due to icing build-ups. When the crew lost speed (Mach 0.76 reducing to Mach 0.6), they kicked out the autopilot and forcefully attempted to maintain height manually. The aircraft's ice-roughened wings stalled, the deep-stall conditions were met and they crashed 50 seconds later. Passing FL140 the crew had radioed that they were "unable to control the plane".
The descent from FL330 took 210 seconds. The debris field was only 200m long and 110m wide, indicating a near vertical descent with no forward airspeed. The crew had radioed that "both engines had flamed out" but they were merely locked back at flight idle.
The problem with not turning on engine anti-ice when it's required is similar to that of the piston-engined pilot who elects not to use carburetor heat. By the time your engine quits and you're then of a mind to use that hot air, it's far too late. There's no hot air forthcoming from a jet stuck somewhere below idle or from a fuel-starved recip.
Most modern airliners operate with a "buffet margin" of around 1.3g. That is to say that they will limit their height not as a function of weight, thrust available or outside air temperature, but in accordance with a graph. That document will decree a margin of airspeed ensuring an increase of 0.3g (say, due to turbulence or maneuver) before encountering mach or pre-stall buffet. It's known that one airline, believed to be Ryanair, recently had a Chief Pilot's edict to temporarily increase that margin to 1.5g, some say due to some as yet unexplained high altitude incidents. The effect of that increment is to limit initial cruise levels to as low as FL330.
In an environment of RVSM (Reduced Vertical Separation Minima) with only 1000ft of planned separation between aircraft, it would be terribly frustrating to enter an energetic TCAS Resolution Alert (RA) avoidance maneuver at high level only to find oneself spinning down through the flight levels and initiating RAs for others as you plummet down to a certain fate.
VLJs are proliferating. Once they're clogging the airways, the incidence of TCAS RAs is likely to rise. That's about as certain as saying that somehow, somewhere, on some day, someone in a T-tail is going to yet again, get unlucky.