Investigation Into Fatal Crash Continues
Even though the Flight Data Recorder (FDR) was apparently more damaged than the cockpit voice recorder (CVR), only the FDR has yielded information on the events leading up to the Garuda 737-400 crash on March 7 that killed 22 people of the 140 aboard.
Australian Transport Safety Bureau (ATSB) Deputy Director Joe Hattley said: “We have tried every method we can to download the cockpit voice recorder, but without success and that includes in-depth consultation with the component manufacturer Honeywell in the U.S.”.
However, raw data yielded by the FDR had been sent to the Indonesian KNKT (NTSB equivalent) without any analysis. It is also understood that only a limited number of parameters were recoverable. The bureau’s executive director, Kym Bills, said it was up to Indonesian investigators to assess the data decoded so far. “It’s more of a matter of checking it against the physical evidence and the other evidence that they’ve gathered on the accident site and in relation to the whole investigation,” Bills said.
In that Garuda Flight GA200 accident, passenger and witness descriptions of their perceived sequence of events indicates that the nosewheel was “driven in” in a potentially catastrophic nosewheel first touchdown off a very fast (“hot”) approach, and that a destructive Pilot-Induced Oscillation (PIO) called “porpoising” then began. Porpoising can become a divergent phugoid in a jet that’s landed far too fast for its weight (and with insufficient drag flap).
What’s Porpoising?
After the nosewheel strikes first, its oleo rebounds upwards and the aircraft maingear “bounces” (courtesy of the MLG oleos compressing and then decompressing a short time later). The neophyte pilot’s natural tendency is to recover by lowering the nose and again to “spot the deck” (i.e., try to force the airplane onto the ground). It’s completely opposite to a normal flare and hold-off process. The inevitable result is another nosewheel first strike (and maingear rebound). The PIO is underway.
This apparently happened three times and on the third occasion the nosewheel oleo snapped and the wheels departed. Complicating matters in turboprops and piston-engined airplanes’ PIOs is the instant power response that’s available. Pilots can easily get “out of sync” by adding power in the bounce (i.e., on the rebound). That added power cycling tends to “eat up” runway remaining and stops the desired airspeed bleed-off.
However, to achieve the same porpoising effect in a jet, you just have to be “hot” (and high) over the threshold, have little or no flap (drag to kill off your float speed), and try to force the airplane onto the ground (resulting in a tricycle landing or even worse, striking nosewheel first). The “eat-up of runway remaining” complication can be provided in a jet by another system (see below) and by the lack of flaps and early spoiler and reverser deployment.
Adding to these self-inflicted woes is the fact that any power “adds” (or throttle manipulation) will inhibit spoiler extension and add to the overrun likelihood. Even without throttle jockeying, you still need the MLG squat switches to be depressed long enough for the spoiler panels to pop up (and then of course, the oleos to remain depressed).
During porpoising, that just won’t happen. Lacking that spoiler effect, the wing just keeps “flying”. Porpoising is a PIO cycle that’s destined to end up destructive and/or off the end, particularly if you’re hot and have landed too far in. We can expect to see many more of these types of “mishandled, confused and fixated” accidents the world over, as pilot experience levels drop and new pilots with Multi-crew Pilot Licences (MPLs) move up in the airline world.
This theory may explain the underlying cause for Flight GA200’s fate once they’d touched down. But why did they allow the aircraft to get so “hot” (i.e., fast) on finals? It may be that:
a. The RH seater just neglected to select flap (or misplaced the lever at <15� and then totally forgot about the flap lever because of the distraction of an overshoot trend) or,
b. The flap speed relief valve (aka load limiter) operated because of the high speed (regardless, flaps should extend normally to the selection once speed is reduced). Flaps can be operated hydraulically or via a backup system, electrically, as long as they are operable and not locked out, or
c. The aircraft suffered a flap lockout, either due to asymmetry (flaps extend a little but an asymmetry trip-switch cuts off the hydraulics to stop a greater asymmetry developing) or, for lockouts occurring in the flaps 0 to 25 range, suspect structural interference, drive-gearing failure, jammed transmissions, foreign object damage, or hydraulic system failure as being the most probable causes.
The EGPWS will give a call of “Too Low Flaps” at 250ft AGL if flaps aren’t at 30� or more. This is cancellable via the Ground Proximity Flap Override Switch. This is what likely happened: an unnoticed lockout.
What if they’d flown through a sharp gust or thermal and picked up a temporary tailwind and just lowered the nose to correct the overshoot trend? Well, that would just indicate a fairly inept handling of a common everyday problem. But unless there was some system failure, put it down to an inability to cope with the environmentals — and perhaps the relative inexperience of the RH seater.
The captain was quite experienced, but possibly inattentive and just let it all go too far. Porpoising is a PIO, and by definition it can be a self-sustaining destructive process. Pilots aren’t taught about porpoising in any syllabus, don’t know the “escape”, and so their first encounter with it will be an eye-opener. Task fixation (leading to task saturation) is by far the most common cause of mishandling accidents during the landing evolution. For some examples of porpoising, try tinyurl.com/33ektj.
From what is known to date, it is suspected that the first officer, flying the visual approach, allowed the aircraft to get too high on glideslope. Without flap, but not realizing it initially, and already using speedbrakes to get down, he likely flared high and then touched down hard about 685m’s into the 2200m runway. After bouncing again at the 885m mark and then losing the nosewheels on their third hit, it appears he latterly attempted to go round, but was thwarted by the other pilot simultaneously attempting manual wheel-braking.
The buffeting on finals (remarked upon by the Australian journalist onboard, Alessandro Bertellotti) is a known characteristic and could have been caused by a non-recommended use of speedbrakes with 30 degrees of flap, in an attempt to burn off the excess height. However, at a higher speed, speedbrake without flap could also have caused the buffeting noted by the passengers.
Elevator effectiveness in the flare is reputedly affected by speedbrake extension at any flap deflection. Speedbrakes are usually stowed by no later than 500ft AGL (at which height the approach should be stabilized or a go-round commenced). However, that journalist flies about 100 times a year. He is a credible witness and is reasonably sure that the trailing-edge flaps weren’t deployed, at least not as he’d been used to seeing them.
Additionally, he was quite certain that the speedbrakes were up throughout. Those double-slotted trailing edge flaps are quite unmistakable when extended, so that witness’s input tends to confirm that the flaps were “locked out” at only a few degrees of deflection — i.e., that the aircraft was flapless and thus prone to floating along for a delayed touchdown.
Note that the no-flap approach visual attitude and the normal approach attitude and flare for touchdown are chalk and cheese. Raising the nose to the normal flare attitude while both hot and flapless, but with speedbrakes, would have led to the next complication.
Per standard operating procedure, the auto-throttle (A/T) should be cancelled no lower than 100ft AGL, but we’re suggesting that it wasn’t; that this step also was forgotten due to the pace of events. After flaring too high (possibly because of the speedbrake and being a natural tendency because of the ground-rush impression at the higher speed) and with auto-throttle still selected, the auto-throttle’s thrust increase would have kicked in at any height above 27ft RA (radar altimeter).
Consequently, power would’ve been added to maintain the bug-speed (as set in the glareshield’s mode control panel) and resulted in the aircraft floating well down the runway (speedbrakes alone being quite ineffectual at lower speeds and anyways countered by the auto-throttle’s response). This would accord to the puzzling “push” referred to by Captain Komar in his interview with Captain Stephanus.
In fact, at 400 feet RA and below, autothrottle is placed in a rapid engine acceleration range. According to the journalist, the speedbrakes were left out throughout, even while the aircraft bounced after being forced down to the runway by lowering the nose (in a forward-stick yoke control input directly opposite to the usual backstick hold-off for the flare and sink onto the runway, as the speed dissipates naturally at idle throttle).
Disarming the A/T can easily be forgotten, and it’s not a checklist item. Since it won’t allow flight below a certain speed, landings become compromised. It doesn’t take a lot of thrust to destabilize an approach and cause a long landing, even without the bouncing. Note that Boeing approved procedures don’t include use of A/T in manual flight because of the possibility of unwanted thrust being applied and not detected.
From the 737 Flight Manual: “The characteristics of the Autothrottle Landing Flare Retard Mode are not intended, nor are they predictable enough, for landing with manual aircraft control inputs.”
During the third porpoise cycle, the nosewheel-first landing caused the nosewheels to detach and the photographic nose oleo-scrape evidence for that is to be clearly seen in the image at tinyurl.com/3x2qur (09 departure end /threshold 27). Also evident in that image are the maingear skidmarks that lead to the conclusion that one pilot was braking while the other was possibly attempting a very late go-round.
Note that the nosewheels may have snapped off much earlier and that the runway end scoring was caused by the oleo leg being driven down and into the runway by the belated commencement of main-wheel braking.
During that possible late attempted go-round, the aircraft struck the embankment just beyond the departure threshold. Damage to that berm (on the runway side of the road) was major. The center divider of the main-road that the aircraft crossed is reportedly intact and unmarked, and the opposite embankment (across the road) is minimally damaged. This means that the aircraft was airborne and floating by the time it crossed the road, but never made it into a positive climb-rate because of the berm-strike.
Evidence for the attempted late go-round is the wreckage distribution, but this explanation should come clearly from the flight data recorder. It has also been disclosed that the aircraft’s port engine reverser had been locked out due to sluggish operation the day before. It’s uncertain and unlikely, from what is known thus far, whether reverse was ever selected, although some witnesses say otherwise.
The accident appears to bear marked similarities to the indecisive events leading to an overrun in heavy rain at Bangkok by a QANTAS 747 in September 1999. However, Garuda Flt 200’s accident was in fine weather and very light winds. That things just “got away from them” is supported by cabin crew evidence that the cabin wasn’t ready just a mere 10 seconds prior to touchdown.
Winds were calm and there was no downdraft. “Normal procedures were not followed. The announcement to prepare for landing was issued about 10 seconds before the plane hit the tarmac, and some of the crew were not in their seats, he said.” (The Melbourne Age).
The ultimate answer may be as simple as both pilots not being in the habit of checking on the indicator that the flap they’ve selected is the flap that they’ve ended up with. If so, they’d understandably then become immediately preoccupied and fixated with the escalating problem of shedding energy and arriving at the threshold with the correct sight picture for a normal landing.
Because the “natural” solution of just pointing at the threshold while flapless will lead to becoming impossibly hot, that’s exactly where they then found themselves: very fast and with no hope of planting it and stopping. A destructive porpoising PIO resulted. The complication with a high round-out and the auto-throttle then adding thrust? That was just the unwanted icing on a fast-crumbling cake.
It’s amazing how quickly a scenario can unravel when you’re a creature of habit and caught unawares by an undetected simple malfunction. Not pointing the finger here. As pilots, we’ve all been there. The solution is in the assertiveness and standards that are required of a commander. Sometimes it’s not an accident, but consequence, that happens.