February 3, 2003

The crash of an Air Midwest twin-turboprop has pushed the cumulative toll
of those killed from maintenance errors to more than 3,040 over the last three
decades, raising new questions about maintenance oversight as well as concerns
about passenger and baggage weights used in determining takeoff weight.

The Jan. 8 crash of the Beech 1900D twin-turboprop during takeoff at
Charlotte, N.C., already has spurred the Federal Aviation Administration (FAA)
to issue an emergency airworthiness directive on elevator rigging and to order
operators of aircraft carrying up to 19 passengers to validate passenger and
baggage weights by March 13. The case is under investigation by the National
Transportation Safety Board (NTSB).

Maintenance work was performed on the airplane's elevator control system
two days before the crash. Investigators have recovered the cables, control
rods, bell cranks and such from the crash site. According to a Jan. 27 NTSB
situation report:

"Examination of the elevator control cables revealed that the turnbuckle
on the 'down' elevator cable was offset to nearly full extension and the
turnbuckle on the 'up' elevator cable was near the fully retracted position, a
difference of 1.8 inches. The turnbuckles are used to set cable tension and are
typically adjusted to about the same length."

Unconfirmed reports indicate that the contract technician who performed
the work was doing so for the first time, and for the first time on this
particular model of airplane. The crash occurred on the ninth flight two days
later. The preliminary evidence suggests that the cables might have been routed
such that there was tension in the "take up" direction (i.e., back-stick or
rearward movement of the control column), but possibly a lack of tension in the
"stick forward" direction. For reasons of passenger comfort, pilots achieve
descent attitudes airborne by reducing power (nose drops naturally) and nose-
down trimming; full stick forward is only checked on the ground (i.e., not under
air loads).

In the case of this particular aircraft, the rigging was such that the
pitch control position sensor showed a 10 nose-down shift. In other words, a
yoke or stick forward control input would have hit the stops sooner. This likely
would not have been noticed during the flights between the maintenance action
and the accident when the airplane was more lightly loaded.

Because the aircraft was heavily laden for the accident flight, it is
likely that the pilot would have needed - or would have required - full back
column for rotation. It is possible that at full up elevator deflection a cable
may have come off a pulley and jammed, leaving the elevator itself in the near
full nose-up position.

At a press conference NTSB member John Goglia expressed his "personal
opinion" that the yoke likely was "stuck" at the near full back position on
rotation.

A pulley-jumped cable-jam would have caused the yoke to seize in the
rearward position and the aircraft would continue to pitch up, which it did to
the point of stall at about 52 above the horizontal.

The FAA issued an emergency airworthiness directive Jan. 27 requiring
Beech 1900, 1900C and 1900D operators to conduct control column sweeps and other
rigging checks by Jan. 31 (AD No. 2003-03-18):

"If maintenance procedures are not properly followed ... it may appear to
the crew that they have full elevator control column movement," the AD said. It
went on to caution, "However, the elevator may not have full travel. Such
restricted travel may remain undetected until the airplane is operated in a
loading condition that requires full elevator authority to control the pitch."
(Emphasis added)

Regarding the weight issue, preliminary accounts indicate that the
airplane was within 100 pounds of its 17,000-lb. maximum takeoff weight and
within one percent of its rearward center of gravity. However, the weight is
based on average weights applied to passengers and baggage.

On the same day the emergency AD was issued, the FAA ordered operators to
conduct spot checks for purposes of validating average passenger and bag weights
used in determining payload weight in 10-19 seat passenger aircraft.

Meanwhile, as the investigation grinds forward, the crash has all the
earmarks of maintenance error as the precipitating event. If so, the 21 killed
in the crash bring to 3,048 the number of people killed in maintenance-related
airliner crashes. That is the toll amassed since 1974, when incomplete
application of a service bulletin caused the cargo door to blow open on a
Turkish DC-10, resulting in the deaths of all 346 aboard, according to a
compilation by this publication (see ASW, Aug. 12, 2002).

Already litigation is in the air. A lawyer formerly with the FAA declared
his intent to file suit soon on behalf of bereaved families.