Mr James Donnelly
Manager, Product Safety
Bombardier Aerospace, Regional Aircraft
2001 Australasian Air Safety Seminar
June 2 - 3, 2001
.....
The final example I'd like to bring before you today also deals with a modified aircraft.
Like the first event I described, it's unusual because virtually the entire accident sequence
is captured on videotape.
The accident occurred in August 1992 at the Gimli Industrial Park in Manitoba - a
famous site in Canadian aviation history, where an Air Canada Boeing 767 known as the
Gimli Glider was dead-sticked onto a drag racing strip, following fuel exhaustion.
The accident aircraft is a highly modified de Havilland aircraft. In this case a Caribou had
been converted to turbine power and was operated under the EXPERIMENTAL category
of CAR 4b.
The conversion was accomplished at Gimli, and the aircraft first flew in mid-November
1991, before accumulating about 23 hours on 12 flights by month end.
These preliminary tests revealed the need for the replacement of the aircraft's mechanical
vacuum pumps with a Bendix suction system, the addition of in-line fuel boost pumps
and the installation of a newly designed hydraulic pump.
The accident occurred on August 27, 1992 on the first of several planned trips to flight-
check the fuel and hydraulic systems. The aircraft had been hangared in a partially dis-
assembled state over the winter, and had only recently been re-assembled, including the
re-installation of the complete tail section.
I'll let the video show you what happened - and I will caution you right now that this
footage is extremely graphic -
[Video]
The accident investigation used this videotape and some 35mm photographs as a key
resource in determining what went wrong at Gimli.
With the exception of a slightly higher-than-normal nose attitude at lift-off, the aircraft's
initial climb appeared normal. At about 35 feet AGL, the aircraft made a noticeable pitch-
up movement.
When I tell you that the photography revealed that the elevator control surfaces were
observed to pitch trailing-edge-up for rotation, neutralize and then remain in the neutral
position through the balance of that short flight, I expect most of you will come to the
same conclusion as the Transportation Safety Board of Canada. The aircraft's control
gust locks were at least partly engaged.
A very close examination of the video does indicate rudder movement and minimal
elevator movement, during the start of the takeoff roll.
On the standard Caribou, the gust lock control handle is located forward of the power
quadrant, and it has two positions - forward for Unlocked, and aft for Locked. If the
control surfaces are not in the neutral position when the lock is engaged, any movement
of the surfaces through the neutral position will cause the lock to engage.
In addition, on the factory-standard Caribou, the control handle is designed so that when
it is in the aft-Locked position, the power levers cannot be fully advanced. This is
intended to prevent power application and takeoff when the gust lock system is engaged.
The accident investigation further revealed that the aircraftÃŒs takeoff distance was
approximately 20 per cent longer than anticipated for the conditions. This may provide
further evidence that the gust locks played a part in this event.
Analysis of the recovered debris indicated that, although the aileron and elevator locking
mechanisms were in their respective Disengaged positions, the rudder locking
mechanism was found to have been in the fully engaged position at impact.
Further investigation revealed that in fact, it had been jammed there by the forces of the
impact. In addition, the analysis determined from the damage evidence that the aileron
control lock had been dis-engaged at the time of impact.
In its synopsis of the accident, the Transportation Safety Board concluded that the control
gust lock system had not been fully disengaged prior to flight and that one or more of the
locking pins had become re-engaged after lift-off.
What could have prevented this accident? The most obvious solution was that a complete
six-point control check prior to takeoff would have revealed that free and proper
movement of the control system was compromised.
No control check was seen by witnesses on the ground, nor was one recorded on video or
still photography. As noted earlier, some rudder and elevator movement was observed, at
the end of the runway at the start of the takeoff roll.
The CaribouÃŒs standard procedures do allow for locking the control surfaces for ground
operation, but the aircraft flight manual also requires a six-point control check prior to
takeoff.
Another pointÛalthough not one addressed by the TSB in its reviewÛconcerns the crew.
We understand that shortly before the flight, the scheduled co-pilot - a very experienced
piston-Caribou captain - was replaced by another pilot with considerably less total time
and experience on type. He was, in fact, the aircraft owner's son.
We therefore speculate whether a more experienced co-pilot might have caught the
missed six-point control check, or might have been more aware that the aircraft was not
responding as it should have.
During the post-accident autopsy, a knob from the gust lock handle was found embedded
in the captain's right wrist. The TSB concluded that the captain was attempting to operate
the gust lock handle when the aircraft hit the ground.
Our expectation was that the pilot flying would have had his hand on the power lever
quadrant, which is located immediately aft of the gust lock handle. It is therefore
conceivable that, during the impact sequence, his hand might have moved forward, and
that this might account for the autopsy finding.
This accident investigation was problematic for us as the aircraft's original manufacturer,
as we had not been involved in the turbine conversion, system modifications, or
subsequent flight testing.
As I noted earlier, the conversion required extensive modification of a large number of
the aircraft's systems, and we know from the investigation that these included a re-
designed throttle quadrant.
The TSB report concludes that the newly designed system did not interfere with the
positional relationship between the throttle levers and the gust lock control handle, as full
power could not be obtained with the lock handle in the engaged position.
However, in our minds, since we did not design or participate in the modification
process, we cannot conclusively rule-out interference with normal operation of the
aircraft's original systems.
In this accident investigation our contribution was therefore essentially limited to the
identification of components familiar to us, confirming the operation of the original gust
lock system, and confirming the deflection of the control surfaces' spring tabs when
operation is attempted against the locks.
We had not reviewed the turbine aircraft flight manual but here again, we suspect it was
substantially similar to the original aircraft flight manual.
... full paper: http://www.asasi.org/papers/2001/Four%20Unrelated%20Accidents.pdf