Recall that the Air Midwest accident aircraft was put back into revenue
service immediately after the tension was adjusted on its elevator cable (see
ASW, June 16). The notion of a flight test beforehand was raised before a return
to line service. That suggestion elicited the following response from Kevin
Horton, an engineering test pilot at Transport Canada:

"The discussion on the elevator cable rigging problem that was present on
the Jan. 8 crash of an Air Midwest Beech 1900D ends with the statement, 'There
is a very simple means of guarding against catastrophe. Any adjustment to flight
controls should involve a dedicated flight test to prove the controls are set up
correctly.' However, based on 15 years experience as an engineering test pilot,
I can assure you that the vast majority of flights will only require a small
portion of the total available travel on each axis. So the successful
completion of a test flight will not guarantee that full travel is available on
all control surfaces. The only way to ensure that full travel is available is
for the maintenance personnel to measure the actual travel at the control
surfaces as the cockpit control is moved from stop to stop."

ASW Contributing Editor John Sampson responded with these thoughts:
"There are possibly more than two schools of thought on this but the one
that used to keep me airborne more often than not as a unit maintenance test
pilot with Hawker de Havilland was the requirement for a functional check flight
on control adjustments as minimal as re-jigging the electric trim [system]. A
trims air-test would be a quick out and back again with CB's [circuit breakers]
pulled to freeze the elevator, aileron and rudder trims at the appropriate
position for nulling at the specified speed (normally 300kts). Sometimes a
stall-check would be required due to flap or flight-control re-work. I can
recall canopy removal/replacements where shims were inadvertently omitted and
the canopy sprang its locking and remained precariously held by a weak secondary
latch. I've experienced tip-tanks that came adrift and rotated around the
central spigot due to improper retensioning. I've seen similar post-maintenance
nasties in four engined aircraft that covered the whole spectrum from engines
and flight-controls to fuel-dump and handling qualities.

"By definition a post-maintenance functional check-flight (FCF) is
designed to avoid rude surprises for those persons not being paid to take those
risks. An FCF pilot is looking for (and unsurprised by) glitches, whereas line-
pilots tend to trust that the airplane is serviceable. The accident aircraft in
question was flown by a number of pilots in a variety of weight and balance
conditions. If that aircraft had been loaded to a standard max load and C of G
[center of gravity] and flown to a profile and the DFDR [digital flight data
recorder] then pulled and examined, the fault may have been disclosed. But as I
have said before this accident, air-tests nowadays, whether for engines or
airframe, are carried out with a full load of fuel and passengers.
"One sees such hazards as both engines in a twin-engine aircraft being
run out of oil because sump plugs were left out after a bore-scoping. Quite
sanguinely, it is a commercially dictated risk that is being taken.

"I agree with Horton that the prime responsibility is that the
maintenance task be done right in the first instance, but the conundrum is that
quite often it isn't - and thus there is no double-check for shift-change
misunderstandings, pencil-whipping or innocent error. That Emery Worldwide
Airlines DC-8F that crashed Feb. 16, 2000, due to a faultily connected elevator
could quite possibly have been full of passengers were it not for the fact that
it was a freighter (see ASW, May 27, 2002, and July 1, 2002). Research could
turn up a number of other instances. Indeed, a fatal 1978 accident involving a
Beech 99 seems a remarkably similar precursor to the Air Midwest crash of the
Beech 1900D earlier this year at Charlotte, N.C.

"Possibly it's the wording used in the ASW article that is the crux of
the matter. The article said 'a dedicated flight test' rather than 'a dedicated
test flight.' What's the difference? Well, simply this. If one were guarding
against the worst-case possible outcome of getting airborne with a full
passenger and freight load, i.e., the precise setting for that fateful Charlotte
flight, one would be testing for that exact condition by flying it with a
critical aft-biased loading.

"One of the cited advancements of aviation has been that BITE (built-in
test equipment) has allegedly taken much of the uncertainty out of black-box
swap-outs and allowed aircraft to be released with confidence about their
systems. But every now and again this arrangement proves itself to be eminently
fallible - and flight controls represent an area where there's little margin for
recovery from error. Recall the near-accident March 20, 2001, of a Lufthansa
A320. On its first flight following maintenance the airplane's wingtip came
within two feet of contacting the ground during takeoff at Munich. It turned out
the sidesticks had been cross-wired, which meant control inputs were reversed.
The near-crash was saved by an alert first officer (see ASW, June 4, 2001).

"The same type of glitch killed a very experienced Hawk instructor at RAF
Valley in February 1996 (ailerons disconnected).

"If I knew that the control cables of an aircraft had been totally
reworked, I'd be quite underconfident, with my background, about embarking as a
passenger on that accident aircraft. It may be useful to recall the recent June
7 Beech 1900D incident where the pilots got airborne into a Mayday situation
with the pilot flying only able to maintain wings level with 45 degrees of yoke
deflection (see ASW, June 16). How would the pilot have fared getting airborne
in a nasty crosswind? I wonder whether maintenance had just been carried out on
that bird? I'd guess so.

"I'll concede that Horton has a valid point but consider this: there is a
current tendency to overreact and recall airplanes when security appears to have
been compromised. But safety double-checking has been superseded and relegated
to a lesser priority because it isn't the cause celebre that it once was.
Security is now paramount."

Horton responded:
"I agree with John Sampson's basic premise that dedicated post-
maintenance functional check-flight could catch many discrepancies (I was a
maintenance test pilot with the Canadian Armed Forces for many years). However,
a post-maintenance functional check-flight can only be effective if there is a
detailed, standard block of tests with pass/fail criteria to be performed for a
particular system, and the flight crew is experienced at performing these tests.

I believe that most air forces make good use of post-maintenance functional
check-flights, but many airlines have neither standard post-maintenance
functional check-flight cards, nor crews who are experienced in conducting such
flights."

See examples of maintenance-related accidents at www.iasa.com.au/fcf.htm
>> Horton, e-mail hortonk@tc.gc.ca; sampson e-mail safety@iasa-intl.com <<

A Precursor Accident

At Richland, Wash., involving a Beech 99 in which all 17 aboard were
killed, 25 years before the Jan. 8 crash of a Beech 1900D at Charlotte. N.C.,
killing all 21 aboard:

"The aircraft lifted off ... and was then seen to begin a steep climb at
an angle of 20-45 to 300-400 feet height. The plane then [stalled] and
descended nose-down at a flightpath angle of about 45 until it struck the
ground ... and caught fire.

"The pitchup was induced by the combination of a mistrimmed horizontal
stabilizer and a center of gravity near the aircraft's aft limit.

"The mistrimmed condition resulted from discrepancies in the aircraft's
trim system and the flightcrew's probable preoccupation with making a timely
departure.

"Contributing to the accident were inadequate flightcrew training ...
inadequate maintenance procedures, and ineffective FAA [Federal Aviation
Administration] surveillance."
Source: NTSB accident investigation report NTSB-AAR-78-15