The final investigation report on the crash of Alaska by rub18840



         The final investigation report on the crash of Alaska Airlines Flight 261
         finds fault with maintenance procedures and design.

         Martin Aubury                                  failure of the gimbal nut threads of the      attachment configuration that was not fail
                        h here we go,” the captain      horizontal stabiliser trim system jackscrew   safe – a deficiency that remained hidden

         “A             said immediately before the
                        McDonnell Douglas MD-83
         crashed into the Pacific Ocean just north
                                                          “The thread failure was caused by exces-
                                                        sive wear resulting from Alaska Airlines’
                                                                                                      for almost 40 years. On the stabiliser trim
                                                                                                      jackscrew assembly, approved extensions
                                                                                                      of lubrication intervals increased the like-
         of Anacapa Island, California. Two pilots,     insufficient lubrication of the jackscrew     lihood that an unperformed or inadequate
         three cabin crew members and 83 passen-        assembly,” the report said.                   lubrication would result in excessive
         gers on board died in the crash, on 31           However a single maintenance failure        thread wear.
         January 2000.                                  should not lead to the loss of an aircraft       The excessive wear was at the maximum
            The US National Transportation Safety       and all people on board.                      limit, and went unaddressed, according to
         Board’s final investigation report found         One of the key contributing factors         the Alaska Airlines lead mechanic,
         that the probable cause of the accident was    identified by the NTSB was the certifica-     “because it was within the allowable limits
         a loss of aeroplane pitch control due to the   tion standards that allowed a stabiliser      on the work card”.


                                               Mexico, and was bound for Seattle via San        from the ocean were the horizontal
                                               Francisco. During climb, pitch trim              stabiliser and its attachments that join it
                                               jammed. The crew did not know why but            to the top of the vertical tail.
                                               continued the flight. About 70 minutes             There are only three attachments, one
                                               later, at 1549 (local time), the crew            of which is the trim jackscrew at the front
                                               contacted its maintenance control in             spar. The jackscrew is about 60cm long
                                               Seattle to discuss diverting to Los Angeles.     and runs in a nut, mounted in gimbals
                                                  A Los Angeles based maintenance               that are attached to the vertical tail.
                                               employee called back and asked the crew            Threads in the gimbal nut had sheared
                                               if they had tried to rectify the problem by      off and remnants were found as spirals
                                               using the “pickle switches and the suitcase      wrapped around the jackscrew. There was
                                               handles”, (switches that control trim by         no lubricant on the jackscrew. An end stop
                                               tilting the horizontal stabiliser).              on the jackscrew assembly did not prevent
                                                  Following a discussion of primary and         the jackscrew from sliding out of the
                                               alternate trim indications the pilot             gimbal nut, nor was it designed to do so
                                               reported that the stabiliser still did not       once the threads had stripped. There was
                                               seem to move.                                    no mechanical back-up restraint or stop.
                                                  After the exchange, the cockpit voice           Investigation determined that threads
                                               recording suggests that the crew again           in the gimbal nut were worn badly before
                                               tried to free the trim. The pilot said, “let’s   the accident. During the accident
                                               do that”. There was a click, then the state-     sequence the threads initially jammed in
                                               ment, “this’ll click it off ”, followed by       the screw. It is deduced that the crew’s
                                               another click and two faint thumps.              efforts to free the jam caused the threads
                                                  The aircraft’s aural warning system           to shear off completely.
                                               began indicating “stabiliser motion” and           The jackscrew pulled partly out of the
                                               the CVR and flight data recorder show            nut, causing the first dive. Later, the
                                               that the aircraft dived from FL310 until         jackscrew assembly failed completely,
                                               the flight crew regained control at FL240.       releasing the only attachment at the front
                                               About 1612, the crew called Alaska main-         of the horizontal stabiliser. The tail
                                               tenance in Los Angeles to report:                pivoted nose up and the aircraft entered a
                                                  “We did both the pickle switch and the        dive from which recovery was impossible.
                                               suitcase handles and it ran away full nose       Design and certification: The accident
                                               trim down.”                                      aircraft was a successor to the Douglas
An MD-83, similar to the Alaska Airlines
                                                  The exchange ended optimistically with,       DC-9 that entered production in 1963 and
aircraft that plunged into the sea off the
                                               “… see you at the gate”.                         has been further developed as the Boeing
California coast in 2000.
                                                  A minute later, before descent the crew       B-717. On all of these aircraft, the hori-
  Alaska Airlines officials told the NTSB      again commanded a nose-up trim. The              zontal stabiliser is mounted on top of the
that the jackscrew assemblies were not         aircraft again responded in the opposite         vertical stabiliser by two pivot points
stocked as inventory items at the time of      direction, nose-down and the pilot told          about mid-chord and by an electrically
the C-check and were only available            ATC: “I need to uh get down about 10,            driven jackscrew assembly connecting the
through third-party vendors.                   change my configuration make sure I can          front of the horizontal stabiliser to the
  More than two years later, when the          control the jet and I’d like to do that out      vertical stabiliser.
jackscrew jammed in flight, the aircraft       here over the bay if I may.”                       When the flight crew adjusts pitch trim
was still controllable. However, attempts         At 1617 while trying to descend and           or when the autopilot commands an
to free the jammed screw led to complete       configure the aircraft for landing the crew      adjustment, the assembly’s motors, oper-
failure. The front of the stabiliser came      spoke of “like a big bang back there”. Two       ating through a gearbox, rotate a titanium
loose and pivoted up causing the aircraft      minutes later the CVR recorded a faint           torque tube that is inside and splined to a
to dive.                                       thump and, at 1619:36, an extremely loud         hardened steel acme screw. In turn, the
  How did the accident happen, and what        noise followed by increased background           screw rotates in a gimbal nut attached to
can we learn from it?                          noise until the recording ended at               the vertical stabiliser, raising or lowering
  The aircraft, operating as Alaska Airlines   1620:57.                                         the leading edge of the horizontal
Flight 261, had departed Puerto Vallarta,      Wreckage: Among the debris retrieved             stabiliser.

                                                                            FLIGHT SAFETY AUSTRALIA, JANUARY- FEBRUARY 2003 < 41

The captain of a private diving boat passes an Alaska Airlines seat cushion to US Coast Guard workers collecting debris from Flight 261
in the Pacific Ocean near Port Hueneme, California on Tuesday, 1 February 2000.

  The configuration of the horizontal            Boeing contends that the concentric            Because the MD-80 series was a deriva-
stabiliser, its attachments and control        torque tube and acme screw achieve fail-      tive of the much older DC-9, its mainte-
system, were originally certificated to US     safety by providing dual load paths strong    nance requirements and procedures had
Civil Aeronautics Rule (CAR) 4b effective      enough should either fail. Within the acme    developed through many stages. Some
March 1962 and have remained essentially       nut, structural redundancy takes the form     underlying concepts of maintenance had
the same.                                      of dual independent threads that provide      changed.
  The rules required that “all control         protection if either thread fails or cracks      The original On Aircraft Maintenance
systems shall be provided with stops           along its root. What actually happened        Program (OAMP) recommends that the
which positively limit the range of motion     was that both threads failed because they     jackscrew be lubricated at between 600
of the control surface” and that “control      were so badly worn. There was no truly        and 900 flight hours. A later one recom-
system stops shall be capable of with-         independent load path.                        mends lubrication at every “C” check,
standing loads corresponding with the          Maintenance: From the outset, safety of       nominally 3,600 flight hours.
design conditions for the control systems”.    the stabiliser attachment depended on            Over the years, the manufacturer
  Also “...catastrophic failure…[is] not       maintenance of the screw assembly, espe-      recommended several changes to proce-
probable after fatigue failure or obvious      cially its correct lubrication.               dures for lubrication and the end-play
partial failure of a single principal struc-     The nut was made from relatively soft       check.
tural element”. However, none of the           aluminium bronze alloy, so it was                Alaska Airlines scheduled jackscrew
rules exactly addressed the failure that       expected to wear and was inspected peri-      lubrication every eight months, about
happened on Flight 261.                        odically for excessive wear. That was done    2,500 flight hours, more often than the
  Apparently the design not only complies      by a “go/no-go” check of end play in the      industry norm. End-play was checked at
with the original rules, but with current      assembly: wear rate trends were not           alternate “C” checks.
Federal Aviation Regulations Part 25.          monitored.                                       “C” check escalation was not subject to


                                                                                               The jackscrew from the horizontal
Gimbal nut from Flight 261, shown attached to the front spar of the fin.
                                                                                               stabiliser of Alaska Airlines flight 261,
                                                                                               recovered from main wreckage site.

task-by-task evaluation. In retrospect, this    metres up. Procedures for both tasks were      • Modify certification regulations to
led to Alaska and other airlines servicing      imprecise before the accident.                 ensure that new horizontal stabiliser trim
the jackscrew far less often than originally       NTSB does not investigate accidents to      control system designs do not have a
envisaged and contributed to the risk of        lay blame, only to identify factors that       single-point catastrophic failure mode.
inadequate lubrication, undetected wear         jeopardise safety.                             • Revise certification regulations to ensure
and eventual failure.                              From Flight 261 the NTSB has made 16        that wear-related failures are fully consid-
NTSB findings: The NTSB attributes the          recommendations including:                     ered and will not be catastrophic.
loss of Flight 261 to insufficient lubrica-     • Instruct pilots facing a faulty flight       Lessons for Australia: The NTSB prima-
tion of the stabiliser jackscrew. However       control system to complete only the appli-     rily addresses its recommendations to the
there      were     mitigating      factors.    cable checklist procedures, and not to         US Federal Aviation Administration for
“Contributing to the accident was (1)           attempt any other corrective actions. If       action. In time, changes to FAA design
Alaska Airlines’ extended lubrication           checklist procedures are not effective, they   rules and maintenance practices enhance
interval, and the FAA's approval of that        should land at the nearest suitable airport.   safety worldwide.
extension, which increased the likelihood       • Require an inspector’s sign-off for (DC-        Meanwhile some lessons are clear and
that an unperformed or inadequate lubri-        9, MD-80/90, and Boeing 717) jackscrew         immediate:
cation would result in excessive wear of the    assembly lubrication.                          • Do every maintenance task meticulously
gimbal nut threads; and (2) Alaska              • Review intervals for critical maintenance    and on time; carelessness kills.
Airlines’ extended end-play check interval,     tasks that have been extended without          • Beware extending maintenance intervals
and FAA approval of that extension, which       adequate engineering justification.            without sound engineering justification.
allowed the excessive wear of the gimbal        • Pending the incorporation of a fail-safe     • When a failure happens in flight, follow
nut threads to progress to failure without      mechanism in the design of DC-9, MD-           the checklist. If that does not solve the
the opportunity for detection.                  80/90, and Boeing 717 horizontal stabiliser    problem, land. It is always safer to fix a
   “Contributing also to the accident was       jackscrew assembly, establish an end-play      defect on the ground.
the absence on the MD-80 of a fail-safe         check interval that (1) accounts for the             Adapted from NTSB aircraft accident
mechanism to prevent the catastrophic           possibility of higher-than-expected wear          report - 02/01. Martin Aubury, formerly
effects of total gimbal nut thread loss.”       rates and measurement error and (2)             head of Structures at CAA now teaches at
   Moreover, lubrication and the end-play       provides for at least two opportunities to            the UNSW School of Aerospace and
check were difficult. Access was through        detect excessive wear before wear becomes       Mechanical Engineering at the Australian
a small panel near the top of the tail, eight   catastrophic.                                                      Defence Force Academy.

                                                                            FLIGHT SAFETY AUSTRALIA, JANUARY- FEBRUARY 2003 < 43

To top