When it comes to the 24 March crash of the Federal Express MD-11F in Tokyo, the expression “three strikes and you’re out” comes to mind. The MD-11 has an unfortunate history of landing accidents, and two previous wipe out’s featured the airplane striking hard on the tarmac, a wing breaking off and the rest of the airplane flipping onto its back.
And previous accidents have killed a few passengers, but now pilots, who have long known the MD-11 was unforgiving on landing, have paid with their lives. The pilot and copilot of the FedEx MD-11 that crashed at Tokyo’s Narita airport were killed.
They were a whole lot less fortunate than the crew of the FedEx MD-11 lost at Newark, NY, in July 1997. They were able to scramble out of the wrecked plane. After its investigation into the crash, the National Transportation Safety Board (NTSB) in August 2000 issued a brace of recommendation, calling for new flight control software to decrease the airplane’s pitch sensitivity and, more significantly, a study of the aircraft’s shaky landing characteristics with a view to improving certification criteria so that such a design would never be approved again for airline operation.
Wreckage of FedEx MD-11 at Tokyo’s Narita Airport. The airplane flipped upside down, the third such landing accident in 12 years. There have been other landing accidents where the airplane remained upright, but the overall safety record of the MD-11 is checkered.
This study was never done. It is likely that, say, an MD-12 (were one to be designed along the lines of the MD-11) would meet current certification standards, presenting the risk of further accidents down the road.
By closing its recommendations for the unaccomplished certification study, the NTSB has failed to advance the cause of aviation safety, and the Federal Aviation Administration (FAA) has been let off the hook, allowed to leave in place standards that have clearly led to deaths. Well, cynics don’t call the FAA the “tombstone agency” for nothing.
To understand the MD-11’s dicey handling qualities on landing, a bit of history is in order.
First, the MD-11 is essentially a stretched DC-10, with winglets, a 2-crew cockpit (one fewer than on the DC-10), and a few other nips and tucks. Significantly, the designers increased the operating weight and the length of the aircraft, without re-engineering the wing and rudder. As a result, the aircraft has rather sluggish roll and yaw responses to control input at low speeds, i.e., on short finals and during the landing flare.
The second significant factor that affects the MD-11s landing performance is speed. In order to compensate for the MD-11’s higher operating weight and reduced rudder authority, its approach speeds can be substantially higher than other comparable commercial jets. This is due to the higher Vmca of the MD-11. The term refers to the minimum airspeed at which an airborne multiengine aircraft is controllable with one engine inoperative (the MD-11 has three engines). For example, the approach speed (Vapp) for the MD-11 at maximum landing weight (213.8 tons) is around 168 knots, give or take 1-2 knots. This is increased further – by up to 20 knots – if allowances for high winds or gusts are factored into the landing calculations. As a consequence, the sink rate on a heavy approach is usually a few hundred feet per minute higher than on most other transport-category aircraft.
The third factor is the center-gear of the main landing gear. Quite unlike the center-body gear on the B747, which is quite forgiving, the center-line fuselage gear on the MD-11 is primarily designed to support the stretched airplane’s increased ramp weight. Its position at the fuselage center, slightly aft of the main gear, has a significant effect on the landing characteristics of the aircraft. For this reason, it is imperative that the pilot does not continue to flare the aircraft when the radio altimeter callout reaches 10 feet. There should be no continued flare even if the aircraft is not in the desired landing altitude. If the pilot continues to apply back pressure on the control column past this 10 foot radio altitude point, the flare has the effect of driving the center gear onto the runway, which in turn creates a large “up” force on the tail section of the aircraft – which in turn drives the nose gear onto the tarmac and creates a bounce.
Fourth, because of the airplane’s small horizontal stabilizer, a computerized longitudinal stability augmentation system (LSAS) was installed to improve handling qualities (and also the airplane’s shortfall in range). But this system does not operate when the autopilot must be disconnected and the aircraft hand-flown to touchdown. The observations of some MD-11 pilots are relevant:
“My question regarding the LSAS is that in the final with autopilot OFF, the big bird is unstable. That is, you must fly it at all times, it never maintained the pitch stable. Before the MD-11, I flew the B767 for five years and the stability on final of the twin is nice. You simply trim it and you can fly almost hands off. That is my big concern about the flying characteristics of the MD-11.”
“LSAS is a ‘yaw damper’ in the pitch axis intended to provide the requisite longitudinal stability after the horizontal stabilizer shrank (for drag count) 40% from that of a DC-10 … In 4+ years on the MD-11, it seemed that LSAS rewarded a smooth pilot but penalized an average/mediocre/’having a bad day’ fellow horribly by opposing every pitch input – particularly in descent and approach.”
“I would agree that the airplane is unforgiving of bad technique, but I believe that is [a consequence of] the inherent lack of pitch stability due to the fact that LSAS is not providing that at lower altitudes where most hand-flying is done … as it washes out below 16,500 feet.”
“If, in an MD-11 simulator, you do not flare and do not decrab, you will certainly make a bone crushing landing, but it will not bounce (remember, it’s just a simulator). However, if you make a last-ditch flare input at 10 ft or 5 ft … you will put the mains onto the runway hard and after that become airborne again. If you pulled hard enough, you will easily zoom up to 30, 40, 50 feet. If you then do nothing for just a few seconds, the aircraft will nose over and start to roll … It is a sickening thing to do in the sim, and it unfolds quickly. The only way out (again, in the sim) appears to be … a go-around. After a bounce like this, full thrust ASAP and wiggle your way out with some delicate pitch-control until it starts flying again.
“Regarding the tendency of the nose going place – it used to be a pronounced nose up tendency, but this has been (mostly) suppressed by the Positive Nose Lowering feature of the LSAS.”
Not to mention that the autothrottles go to idle at 35 feet radio altitude, per design. For this reason, it was recommended that the MD-11 not be flown on autothrust while descending on final in possible windshear situations.
These are some of the factors that come into play when trying to land the MD-11. In the Tokyo crash, with gusty crosswinds and possibly fatigue resulting from the schedule and the flight from Guangzhou, China, the pilots found themselves seriously and abruptly out of phase with the divergent phugoid of the MD-11’s porpoising proclivities.
In video footage of the crash, the plane appeared to bounce on its landing, slamming down, veering sharply to the left and then exploding into a fireball as it flipped over and skidded to a halt in the grassy strip next to the runway.
Three MD-11s have now crashed on landing and gone belly up. There have been other landing crashes and incidents, but let’s focus on the three where the plane ended upside down.
The first time was at Newark in 1997, where a FedEx MD-11 freighter landed in good weather, but bounced on landing, came down hard on the main landing gear, broke off the right wing and was wrecked. The two-man crew got out alive and unharmed. The NTSB determined that the probable cause of the accident was “the captain’s overcontrol of the airplane during the landing and his failure to execute a go-around from a destabilized flare.”
The burned out wreckage of the FedEx MD-11 at Newark, NJ. Although in fair weather, the airplane flipped onto its backside after bouncing.
What resulted was pilot-induced oscillation (PIO). As the NTSB explained:
“Considering the captain’s three significant elevator control inputs in sequence, it is apparent that after the first destabilization of the landing flare (from the captain’s nose-down input at 17 feet above ground level), each of the succeeding nose-up/nose-down elevator inputs resulted from the captain’s attempt to correct for the immediately preceding control input. His perception of a short runway and the need to constrain the pitch attitude within a very limited range (to avoid a tailstrike) would have motivated the captain to rapidly return the airplane to a stable attitude. He attempted to accomplish this goal with a quick application of large elevator inputs; however, this succession of elevator inputs and pitch oscillations rendered the landing attempt increasingly unstable.
“Throughout the sequence of extreme nose-down and nose-up elevator inputs, which were consistent with a ‘classic’ pilot-induced oscillation (PIO), the captain continued to attempt to salvage the landing …”
The second such landing accident – this time involving the airplane skidding off the runway in near-typhoon conditions of rain and wind and flipping upside down – occurred in August 1999 at Hong Kong involving a China Airlines passenger MD-11. Three passengers of the 315 aboard were killed.
“China Airlines crash wreckage at Hong Kong. Just before touchdown, while landing with a 24 knot crosswind, the right wing struck the runway, causing the aircraft to flip into its back.”
China Airlines crash wreckage at Hong Kong. Just before touchdown, while landing with a 24 knot crosswind, the right wing struck the runway, causing the aircraft to flip into its back.
The Civil Aviation Department of Hong Kong (HKCAD) investigated the accident. This passage in its accident report is interesting to note regarding trials run in the simulator after the accident:
“During these approaches, any ability to flare the simulator below 50 ft using the technique recommended in the China Airline Operations Manual and achieve a normal touchdown t a low rate of descent proved unsuccessful on the majority of approaches flown; if power was manually applied late in the flare, the rate of descent could be reduced but was still high at touchdown.”
As a simulator instructor is wont to say, “It’s not hard to land, it just lands hard!”
Actually, the airplane is tricky to land. The HKCAD report in fact referenced the FedEx crash in Newark:
“In that accident, the aircraft which occurred in good weather conditions, the aircraft also suffered structural failure of the [right main landing gear] and right wing rear spar, and came to rest inverted.”
And the HKCAD report noted limitations of the LSAS in providing pitch attitude hold: “LSAS is not provided when the autopilot is engaged. Below 100 ft radio altitude, and transparent to the pilot, LSAS is progressively removed from the pitch control system.”
One would think LSAS would function all the way to touchdown.
The latest FedEx accident at Tokyo brings home how lucky the crew and passengers were in the China Airlines crash, surviving with only 3 deaths out of the over 300 on board.
One is led to the inescapable conclusion that the MD-11 is really a handful in gusty crosswind landings. It might even be called “unsafe” in such a situation, given this latest FedEx crash and that of the China Airlines crash. The 1997 FedEx crash was in good weather and yet it still bounced during landing. It would appear that the MD-11 takes way better than average ability to land safely in high winds or off a bounce.
The characteristics of this airplane ought not be permitted in future designs, as stated in the outset.
Which raises questions about the characterization by the NTSB of FAA responses to its post-Newark accident recommendations.
The NTSB urged that the FAA to require the installation of new LSAS software, known as the Flight Control Computer upgrade (FCC-908) to render the airplane less susceptible to over control. The FAA never required this upgrade, but it was installed in all MD-11s worldwide nonetheless. The NTSB closed this recommendation as “Acceptable Alternate Action.” Well, the change was made to all U.S.-registered MD-11s, including the one that just crashed at Tokyo. Clearly, upgraded LSAS was not sufficient to cure the MD-11’s shaky and demanding landing characteristics.
The NTSB also recommended a study assessing “qualitative and quantitative stability and control characteristics” for purposes of implementing “improved certification criteria for transport category airplane designs that will reduce the incidence of landing accidents.” This recommendation was not implemented.
In fact, the FAA blew off the need for any such review, saying:
“The FAA shares the [Safety] Board’s concern that certain complex system interactions in the flight control system, piloting characteristics, and other factors can, on rare occasions, combine in undesirable ways. However … the FAA does not believe that basic research based on past accident reports will identify any undesirable landing combinations that are directly related to stability and control characteristics … causal and circumstantial factors are more directly related to operational procedures, piloting issues, and environmental conditions.”
The NTSB concluded that “review of the studies described by the FAA represents an acceptable alternate method to address Safety Recommendation A-00-100.” On this basis, the recommendation was classified “Closed – Acceptable Alternate Action.”
Two dead MD-11 pilots at Tokyo indicate that the airplane’s handling characteristics still do not constitute an acceptable design now, or in any future transport category aircraft. Any design so exacting in its requirements and so very vulnerable to any touchdown asymmetry ought not be permitted in future, period.