Two Takeoff Overruns Highlight Safety Issues On The Ground

Air Safety Week, Jan 31, 2005

When MK Airlines Flight 1602 hit the 22-foot high raised earth and concrete berm off the end of Runway 24 at Halifax, Nova Scotia, last Oct. 14, the 747 freighter was still struggling to get airborne - but its nose-high attitude ensured that the aft fuselage could not clear any significant obstacle in the overrun. It's no certainty that the aircraft would have gotten airborne had the large raised mound of earth not been across its flight-path. But its presence ensured that flight beyond the overrun stopway had no chance of success.

What was the berm doing there? Quite simply it was supporting lighting stanchions and the localizer antenna. Many are now asking why Halifax Airport (HFX) used a berm instead of the much more commonplace frangible pole-mount gantries that simply break off in the event of an overrun or short landing. A clearway should be as clear as possible, but 1950s clearway standards may be inappropriate for the inertia of a 747 or A380.

Speaking at a news conference in Halifax following the crash, investigator Bill Fowler of the Transportation Safety Board (TSB) of Canada said, based on his initial survey of the crash site, "it appears the jet's tail struck the ground twice, before tearing loose on contact with an earthen berm 300 meters beyond the end of the runway."

"There is an indication that the aircraft was slightly airborne; in other words, the scrape trail disappears just a few hundred feet before the berm. At the antenna-topped mound, the plane broke in two," he said. With its tail torn off, the rest of the fuselage flew into the bush, cutting a half-mile-long swathe before coming to a stop.

Runway overruns and compacted stopways are of little interest until you suddenly find yourself on top of one. If you are in the throes of a rejected take-off and trying to stop, then a gravel arrester bed just beyond the stopway is welcome. If that's not there (because they're normally found only where an overrun might otherwise invade a built-up area, cliff or beach promenade), then dirt, mud or undergrowth is an acceptable alternative. But a large mound of earth? That design can be lethal.

The TSB is steadily eliminating possible causes and appear to have concluded that:

* The crew used all the available runway length;

* The aircraft was not overweight (although the "averaged" weighing protocols at Halifax are being questioned);

* The four engines were punching out the power requested of them (although two had recently been worked on) and the aircraft was correctly configured; and

* Added power (the final increment of throttle advancement) was noisily applied quite late in the takeoff roll, apparently beyond the required EPR (computed engine pressure ratio) and, in fact, according to the flight data recorder (FDR), all the way up to MAX.

The bottom line would appear to be that the pilot applied the initial coarse power throttle-up upon rolling (setting > circa 1.3 EPR) and that then the flight engineer (FE) may have been distracted by a discrepancy or left in limbo by the lack of a standard pilot call-out to finely "set thrust" (1.5 EPR). Consequently, the required power may have been set too late or incorrectly set. The FE may have mistakenly used the runway length from the last port of call (Bradley International Airport near Hartford, Conn., at 10,000 feet) to calculate his required power. The airspeed indicator bugs (if not destroyed) would show whether the Bradley V-speeds were used. If the old takeoff data card was used accidentally (instead of the newly calculated numbers) this would also explain a lot.

Equally, the FE might have entered ZFW (zero fuel weight) in lieu of "all up weight" (AUW) into his laptop software or been sucked in by a software "entry trap". Whatever the error's origin, the "rolling down the runway" result of any such fatigue-induced train of events initially would be confusion. Then, first officer (F/O) vocal or physical intervention might result, and later an involuntary pitch overcontrol would stem from pilot alarm. One factor on Runway 24 in HFX is that the far end of the runway is lower than the mid point, and it may be that the pilot did not realize his proximity to the end until it was too late (i.e., cresting the high point and suddenly seeing the red lights marking the last 1,000 feet of runway materialize - looming large).

Once it's too late to stop, you are go-oriented, so that overcontrol would take the form of a desperate over-rotation - as clearly observed by the ground-crew witnesses. Marks on the runway caused by a lengthy tail-drag signify that the late application of high power, plus a simultaneous "go for it" overrotation, indicated the crew's late sudden realization of their predicament. Then, once rolling lightly on the main-gear, it would be hard for the pilot to tell at night whether or not the aircraft was airborne. But within seconds the aft fuselage had smashed into the steeply sloping berm, the tail broke off and an accident was under way.

Reduced power "balanced field" (aka flex) takeoffs are used routinely to save aircraft engines. Using Flex power, the takeoff distance required equals the accelerate/stop distance for an abort initiated at airspeed V1 (takeoff decision speed). There are various entry arguments for finding that power, but the main ones are mass and runway length. Pilots have argued for ages about whether a rolling takeoff or a standing start affects takeoff roll distances. Boeing's advice is that it does not. However, it is incontestable that in either case, any delay in achieving the computed power will lengthen the takeoff roll, whatever the other environmental factors of temperature, runway slope and wind component. Rarely are we reminded of that truism. Why? Because it's a "given." But could a chronically fatigued crew in the lethargy of an early morning routine departure make that fundamental error? Conceivably.