AAIB Bulletin No: 5/97 Ref: EW/C96/11/2 Category: 1.1
Aircraft Type and
Boeing 747-236B, G-BDXK
No & Type of Engines:
4 Rolls Royce RB211-524D4 turbofan engines
Year of Manufacture:
Date & Time (UTC):
2 November 1996 at 2330 hrs
Near Gatwick, Sussex
Type of Flight:
Persons on Board:
Crew - 18 - Passengers - 306
Crew - None - Passengers - None
Nature of Damage:
Airline Transport Pilot's Licence
Last 90 days - N/K
Last 28 days - N/K
AAIB Field Investigation
History of the flight
When the crew checked in for the flight on the evening of 2 November 1996, they were informed that there was a problem with door 4L, but
that the engineers were confident that the aircraft would be declared fit for service at or near the scheduled departure time. This proved to be
the case and the flight engineer checked the door for satisfactory operation. However, due to a recent history of problems with this door, the
cabin crew responsible for it were briefed to monitor it after take-off and keep the flight deck informed of anything untoward.
Immediately after take-off, the door handle was seen to rotate from its normally closed 3 o'clock position to 1 o'clock. Two cabin crew
immediately attended the door but experienced difficulty in keeping the handle in the closed position. The flight crew were informed, and
upon completion of the "after take-off" checklist items, the flight engineer went aft to the door, and confirmed that a considerable amount of
force was required to hold the handle in position. He noted that as soon as pressure was relaxed, the handle started to move towards the open
position. In addition, there was a loud noise of rushing air, which was apparent at the top and bottom of the door, as opposed to the sides. Two
cabin crew were instructed to keep pressure applied to the handle. The flight engineer became concerned not only for the integrity of the
door, but also about possible implications for the pressurisation of the aircraft. He therefore called the engineers at Gatwick and discussed the
problem with them. They suggested letting go of the handle to see where it would stop. The flight crew declined to do this, and, after further
discussion, elected to return to Gatwick. During this time, the aircraft had climbed to 21,000 ft to allow it to enter French airspace. The captain
had reassured the cabin crew that the door, being a "plug" type, could not come open. However, the cabin crew were understandably
becoming increasingly alarmed, and this was instrumental in the decision to return. The aircraft dumped 46 tonnes of fuel and carried out an
uneventful approach and landing. During the latter stages of the flight, the cabin crew looped together some extension seatbelts in an
attempt to secure the door handle in the closed position.
Basic description of door
All the main entry doors share the same principles of operation, with doors 2 and 4 having identical latch roller cranks and torque tubes, which
constitute the significant components in this incident. A diagram of the door is attached at Figure 1, which shows the principal features. The
operating handle is attached to a crank mechanism. Rotation of the handle causes linear motion in the two adjustable control rods emanating
from the crank mechanism, and in consequence, rotation of the upper and lower torque tubes. Latch rollers at each end of both tubes engage
with cam slot plates attached to the door frame. The "door open" microswitch is located on the lower aft cam slot plate. Additional adjustable
control rods on the torque tubes are connected to the upper and lower "gates', which are hinged sections of the door skin, located top and
bottom. These fold inwards thereby reducing the door dimensions to less than the door aperture, thus allowing the inward movement of the
door. When the door handle is moved towards the fully closed position, the door is pulled into the aperture by the reaction of the latch rollers
against the inboard edge of the cam slots. When the door is flush with the fuselage skin, the bellcranks on the upper and lower torque tubes
are just about to go into their over-centre positions and the upper gate is resting against a step on the inboard surface of the fuselage skin. The
final part of the handle rotation pulls the door into position so that the torque tube bellcranks roll over into an overcentre position. Eighteen
door stops (nine either side) are abutted against load bearing lugs on the inside of the door frame. Thus pressurisation forces push the door
firmly against the lugs, in accordance with the concept of a "plug" door. When the door is opened, the initial rotation of the handle disengages
the latch rollers. The handle mechanism is linked to the door hinge, and further handle rotation moves the door bodily a short distance
inwards, into the cabin. The door then turns through approximately 90° so that it can be swung outwards, trailing edge first through the
aperture, and forwards against the fuselage side.
Recent door problems
On 26 October, the 4L door warning light illuminated on the flight deck shortly after take-off, and with approximately 1.5 psi pressure
differential applied. The cabin pressure remained normal, but the door handle was not fully closed. The handle was pushed to the fully closed
position, but popped up again immediately. However, a further attempt at reselecting fully closed appeared to be successful.
After landing, it was found that the forward end of the upper torque tube had failed, thus releasing the forward latch crank. The aircraft was
taken out of service and a new torque tube and latch crank were fitted by the night shift team. The door was rigged during the following day
but difficulties were encountered in that the pull-off load on the door handle (i.e. the force required to move the handle away from its closed
position) was too low. A number of adjustments were made which brought the release load into limits and the aircraft was released for service
on the evening of 27 October.
On 27 October (the first flight following the rectification), the 4L door light again illuminated during the climb. Pushing down on the door
handle extinguished the light. The door was checked when the aircraft landed, whereupon it was confirmed that it was difficult to close unless
it was pushed on "the top right hand corner" An acceptable deferred defect (ADD) was raised to re-rig the door during the next maintenance
input. During the next sector, on 28 October, the door warning light again illuminated, together with associated movement of the door handle.
When the aircraft returned to Gatwick, the door was rigged in accordance with Maintenance Manual requirements. and the door warning
microswitch was also adjusted. These actions cleared the ADD.
On 30 October the same symptoms, i.e. warning light and handle movement recurred, this time in the cruise. On arrival, a "vast quantity of
water was found around the floor/girt bar...", but no other defects were noted. Another ADD was raised for further investigation. The same
problems with door 4L latching continued to occur during the three sectors flown on 31 October. The maintenance activity on that day
included a further check of the door handle release loads, plus lubrication of the microswitch. A further occurrence on the 1st November
resulted in the door handle being tied down for the rest of the flight. An "inoperative" label was subsequently attached to the door and a
further ADD raised. This resulted in the additional maintenance at Gatwick, and preceded the subject incident on 2 November which was
notified to AAIB.
The entire sequence of events subsequently formed the subject of a Quality Assurance investigation by the airline. In addition, the aircraft
manufacturer contributed to the technical investigation.
Examination of door components
On arrival back at Gatwick on 2 November, the aircraft was removed from service for investigation. Following removal of trim components, the
aircraft was pressurised in an attempt to reproduce the defect. At 0.6 psi it was found that the door handle moved towards the unlock position.
After comparison with another door, the engineers discovered that the upper torque tube that had been installed on 26 October had been
drilled such that there was an incorrect angular relationship between two sets of holes. Each latch roller crank is retained in the end of the
torque tube by a pair of bolts, one at 90° to the other, located across the tube diameter. Another pair of holes is drilled in the centre of the tube
for the purpose of attaching the operating bellcrank. A tooling hole is drilled at either end of the tube, inboard of the latch crank bolt holes.
When the replacement torque tube was compared with the failed one, it was found that the axes of the bellcrank bolt holes had been drilled
with approximately 18° of circumferential displacement from their correct position. As a result it would have been impossible to rig the door
correctly, although the angular error could be largely compensated by extreme adjustment of the control rods attaching to the torque tube.
Figure 2 shows a sketch of the tube, together with the relative positions of the various holes.
Metallurgical examination of the torque tubes
Both the failed tube and the wrongly drilled replacement, together with the failed latch crank, were subjected to metallurgical examinations.
The latch crank fitting, which had been machined from an aluminium alloy forging, had suffered a single torsional overload as a result of a
segment breaking out of the tube. The fracture in the tube resulted from a low cycle, high stress tension fatigue mechanism which split the
tube at its end and progressed in a longitudinal direction. After approximately 1inch the crack branched into two, with one arm running
diagonally back to the end of the tube by way of one of the outer latch crank bolt holes, thereby releasing the segment of tube noted above.
The other branch went to one of the inner bolt holes. An additional crack extended from these branches, following a helical line, to the
adjacent tooling hole. The tube was also longitudinally cracked in the diametrically opposite position, with the crack running into an inner
latch crank bolt hole.
The tube was sectioned and it was apparent that some of the fracture surfaces were stained, suggesting that the cracks had existed for some
time. There was no plastic deformation associated with the fractures, indicating that the material was of low ductility. The overall conclusion
was that the failure had resulted from a low cycle fatigue mechanism, with crack development being accelerated by stress corrosion under the
static loads imposed when the door was in a closed and latched position.
Hardness and flattening tests were conducted on both the failed tube and its replacement, which revealed marked differences in the material
properties. The hardness values indicated that the tensile strengths of the failed and the new tube materials were 208,000 and 155,000 lbf/in2
respectively. The end of the new tube could be flattened to half-diameter without cracking, whereas the failed tube cracked longitudinally
without plastic deformation.
The manufacturer is aware of a history of cracking of the latch torque tubes, and has addressed the problem by means of Boeing Service
Letter 747-SL-52-25, with Revision 'B' dated November 1984. The document emphasises that there has been no operational difficulty
associated with cracked or failed torque tubes, but suggested that operators take action by reworking or replacing the tubes with components
made from improved ductility material, which is achieved by a lower temperature heat treatment process. The airline has stated that torque
tubes are being replaced on an "as and when" basis, with the stores inventory records showing that four units have been drawn up to the time
of this incident. The tests indicated that the incorrectly drilled item was made from the improved material.
Effect of mis-drillled holes on door operation
The design of the door is such that unlike the upper gate, the lower gate is not in a "hard" contact with the sill during the final part of the
latching sequence thus giving rise to a non-symmetrical distribution of loads in the upper and lower torque tube linkages. The aircraft
manufacture stated that 80% of the overcentre lock loads are generated by the correct relationship of the upper torque tube, its associated
door gate and control rods. The remaining 20% is created by the lower latching mechanism and the latch rollers locating overcentre. Unless all
the components are installed in the correct relationship to each other, the latching sequence will not operate correctly. The angular position of
the centre bellcrank on the torque tube, with the 18° error, meant that it would not have been possible for the linkage to achieve the correct
overcentre lock required to keep the door latched, even by utilising the full extent of the adjustment available in the control rods. The
manufacturer additionally stated that without the overcentre lock, but with the correct rigging adjustments, it is possible that the door would
remain latched during normal operations. However the latching mechanism would have been susceptible to movement with the slightest
disturbance. It is possible that such a disturbance could result from a small amount of door movement arising from increasing pressurisation
loads as the aircraft left the ground.
In the event that there was some movement of the top of the door relative to the frame (and hence the cam slot plate), it would result in a
small amount of rotation of the latch cranks and the torque tube. This would back-drive the door operating linkage, overcoming any marginal
over-centre condition, thus causing the handle to move. Remembering that the torque tubes are connected to the gates, it is apparent that the
latter would also have moved, thus accounting for the wind noise at the top and bottom of the door.
Effect on flight safety
The door is essentially a "plug" design, such that loads holding the door in the fuselage frame are a function of cabin differential pressure.
Boeing have indicated that with the aircraft at an altitude of 1,500 ft, the cabin differential pressure is typically 0.3 psi, leading to a closing force
on the door of 900 lbs. They additionally indicated that the handle mechanism is not capable of moving the door when the differential
pressure exceeds 0.18 psi. Above this value, there is no possibility of the door moving from its position in the fuselage regardless of the handle
position. Thus any risk of the door moving is confined to such times as when the differential pressure is close to zero. Boeing stated that
aerodynamic studies showed that in the event that the latches became disengaged, thereby making the door susceptible to moving inwards to
the "cocked" position under the action of vibration, then airflow forces would not cause the door to fully open.
Note: The standard take-off configuration for the Boeing 747 is with the air conditioning packs to be turned OFF. This results in a slight
negative pressure in the cabin until approximately 700 ft, when the first pack comes on line. For landing, the standard procedure is to keep the
packs ON. This maintains a slight but decreasing positive pressure in the cabin until touchdown.
There is in fact an Emergency Procedure in both the Flight and Operations Manuals for "Upper and Main Deck Smoke Evacuation", which
details the procedures for opening the main entry doors in flight for the purpose of clearing smoke from the cabin. Airflow forces will cause the
door to be retained in the "cocked" position, such that a gap of approximately 2 inches can be expected around the periphery of the door. The
door can be left unattended in this position, with no danger of persons falling from the aircraft through such a small opening.
Quality assurance investigation
During the airline's investigation, all the personnel involved were interviewed with the aim of detailing the events that led up to the fitment of
the wrongly drilled torque tube and the subsequent release to service of the aircraft. The first report of problems with the door was on October
26, and resulted in the discovery, at Gatwick of the broken torque tube and latch crank. A new tube and crank assembly were ordered from
Heathrow. The rectification requirements were considered outside the scope of ramp maintenance, with the result that at 1900 hrs (i.e. the
start of the night shift), the aircraft was handed over to the Gatwick Support Unit for investigation and repair. A verbal handover was given to
the shift Fleet Technical Liaison Engineer (FTLE). There was a shortage of certifying staff and a Licensed Aircraft Engineer (LAE) had to be
called in on overtime. Pending his arrival, two hangar engineers were given the task of removing the door trim.
The LAE arrived at 2100 hrs and was briefed by the FTLE. He had replaced a door torque tube on a previous occasion, at Heathrow, and was
happy to accept the task. He was also asked to investigate an APU fault and a toilet problem on the aircraft.
Whilst awaiting the arrival of the parts ordered from Heathrow, the unserviceable tube and latch assembly was removed from the door, after
the door-to-tube relationship had been marked on the adjacent door frame. An inspection of the door was also carried out at this time and it
was noted that there were witness marks on the external surface of the forward upper cam slot plate, indicating that the associated latch roller
had not been correctly located. This in turn suggested that the door had not been rigged correctly.
The new torque tube arrived at 0030 hrs and was found to be undrilled, apart from the tooling hole at each end. The Maintenance Manual
called up a drill jig, which is designed to be attached to the tube via the tooling holes, thus allowing the latch crank and the central bellcrank
holes to be drilled off. However, the drill jig was not available and the main workshops were closed for the night. The decision was made by the
LAE and the FTLE to drill the tube in the area workshop. Consideration was given to calling out the workshop engineer and to leaving the task
until the morning, but these options were discounted due to time constraints and the operational requirement for the aircraft.
Note: Once the non-availability of the drill jig became apparent, it would not have been possible to comply with the Maintenance Manual
requirements. The correct action would have been to call the Fleet Technical Office who have the necessary CAA approval to issue
Engineering Orders covering work that falls outside the Maintenance Manual. It was subsequently established that the airline had never held
the drill jig. The LAE had of course faced a similar situation on the previous occasion he had replaced a torque tube. However, at that time he
had had the facilities of the main workshops available to him.
After bolting the new tube to the damaged tube through the tooling holes, and using the existing latch crank holes in the damaged tube as
guides, it was possible to drill two of the latch crank holes in the new tube with the aid of a pillar drill. The second set of holes were at 90° to the
first, and these were drilled, after their positions had been marked, by means of a rechargeable hand drill. This part of the operation was
witnessed by the FTLE and other engineers, but was not overseen by the LAE, who was away investigating the other two defects on the aircraft.
To drill the centre bellcrank mounting holes in the torque tube, the LAE marked their positions after obtaining the necessary angular
relationships from the Maintenance Manual, and also by comparison with the old tube. Once satisfied, the new tube was installed in the door
and the latch cranks bolted into position. After alignment with the marks made on the door frame during disassembly, the LAE instructed the
engineers to drill the centre bellcrank holes. It must be presumed that the angular error arose at some point in the marking/alignment
process, possibly as a result of mis-reading the drawings in the Maintenance Manual.
The Maintenance Manual required the tube to be painted with primer prior to assembly, and also called up a corrosion preventive
compound. However, the tube is an interference fit through three bearings and unless the primer was allowed to dry, it would have been
scraped off on assembly. The components were not wet-assembled as it was intended to apply a corrosion inhibitor to the area on completion
of the work. However, this instruction was omitted from the subsequent handover to the day shift.
Rigging of the door commenced at approximately 0230 hrs and continued for three hours. By 0530 hrs, the door was fairing correctly but the
"pull-off" load on the door handle (i.e. the force required to move the handle away from its closed position) was below the 20 to 25 lbs required
by the Maintenance Manual. The team were very tired by this time and so it was decided to hand the task over to the incoming day shift. The
work that had been carried out so far, together with the outstanding tasks, was documented in the aircraft Technical Log. No stage sheets were
raised (see the Note below) as the job was considered straightforward enough to be contained within the Technical Log. However, the LAE did
produce a written handover for the next team, although this was never actually seen by anyone.
Note: In some maintenance situations, operators use pre-planned stage sheets which briefly describe each action required, giving the
appropriate Maintenance Manual reference and providing boxes for the authorising signatures. Whilst these documents should not be used
in place of the Maintenance Manual, they nevertheless serve as an aide-memoir and provide confirmation that all critical tasks have been
completed. Thus stage sheets can be of considerable benefit in the non-scheduled maintenance environment.
A day shift LAE was not available, and so a verbal handover was made to one of the two hangar engineers assigned to complete the task. This
engineer was unaware of any written information other than that in the Technical Log and did not see the handover notes written by the night
shift LAE. Later on, another LAE was tasked with co-ordinating the activity on the door in addition to his other work. He therefore made
frequent visits to the aircraft as work progressed. The work was mainly directed to obtaining the correct pull-off load on the door handle, and
included adjustment of the operating rods, the torque tube and the forward latch guide. A considerable amount of free play was also identified
within the handle mechanism, and was eliminated by re-tightening the central mounting bolt. Eventually, a pull-off load of 20 lbs was
achieved, which was considered satisfactory. The door was inspected by the LAE, who considered that all the Maintenance Manual
requirements had been met, and was then reassembled. The aircraft was released for service at 1700 hrs on October 27.
The aircraft continued to experience problems with the door, as described earlier, culminating in the incident on November 2. When the
drilling error became apparent, another torque tube was obtained and drilled off in the main workshops using a vertical milling machine
incorporating a dividing head and a digital dial test indicator. The door was subsequently reassembled and rigged with no problems, the
handle pull-off load now being 25 lbs, as required by the Maintenance Manual. The aircraft was returned to service with no further door
problems since being reported.
Conclusions and follow-up action
Following its investigation, the airline made the following observations:
1. The Maintenance Manual requirements were not followed in respect of; a) Not using a drill jig when drilling the torque tube; b) Failure to
call the Fleet Technical Office when it was apparent that the work now fell outside the Maintenance Manual, c) Failing to protect the tube with
primer and to "wet assemble" during installation.
2. The work should not have proceeded without the jig, workshop or technical support.
3. The LAE in charge was on overtime and working three defects. He made the decision to proceed with the work without the jig, due to
commercial pressures, and was not supervising the work when the tube was being drilled.
4. No stage sheets were used following the initial tube replacement, and no written handover could be found.
5. No consideration was given at the time that something could be wrong when the handle pull-off load requirement could not be met, or when
the door could not be rigged easily.
6. When the aircraft returned from its first trip the full extent of the problem was not appreciated by the engineering organisation. Also, the
problem was not detected by the appropriate Reliability section after the subsequent flights and the numerous failed attempts to rectify the
7. No entries were made in the Technical Log by the flight crews to emphasise the magnitude of the problem.
Measures adopted since the incident have included ordering a torque tube drill jig as specified in the Maintenance Manual. However, in
order to minimise dependency on the jig, it is intended to order pre-drilled torque tubes from the manufacturer. It is also intended to establish
a more effective control over the shift handover files. In addition, stage sheets are to be used whenever work is carried out in the hangar which
is considered to be beyond the scope of ramp maintenance. Furthermore, the airline intend to publish the findings of the investigation,
together with a statement, in an in-house publication, of the requirement to contact the Fleet Technical Office in the event that specialised
equipment is unavailable.
Trans World Airlines Flight 800
Air India Flight 182
Pan Am Flight 103
United Airlines Flight 811
The Type Airplane
The Damage Starts
The Radar Blips
The Sudden Loud Sounds
The Abrupt Power Cuts
The Fodded Engines
The Inflight Damage
The Missing Bodies
The Torn Off Noses
The Wreckage Plots
The Red Herring: Bomb!
Inadvertent Opening of the Forward Cargo Door in Flight
Forward Cargo Door Section
(larger picture with DC-10 door also)
747cargo door and nose
Airworthiness Directive 79-17-02.html
Airworthiness Directive 88-12-04
Airworthiness Directive 90-09-06
Boeing 747 nose picts right side cargo door
More pictures of UAL 811 cargo door hole