The investigation could not determine the origin of the right upper wing or when it was installed. Therefore, its history and airworthiness could not be determined. Despite the extent of damage incurred by the right upper wing during the 1993 accident, the fabric was not removed from the wing to conduct a thorough inspection of the wing structure. During the course of the repair, the wing structure was noted as appearing old. However, had a comparison with the logbook description of the wing been made at the time, it may have been evident that a deeper examination of the wing, to preclude the possibility of more extensive damage to the aged and unknown wing structure, would have been prudent.

It is possible that the compression failures and shakes found in the wooden wing structures subsequent to the 1998 accident, were caused by break-up forces. However, due to the degree of reported damage to the wings during the 1991 and 1993 accidents, it is likely that the damage was present prior to the accident flight.

The presence of either microscopic compression failures or visible shakes would have seriously reduced the load-bearing capacity of the wood. The 1993 inspection was conducted, as recommended by the maintenance manual, visually through holes cut in the fabric. The appropriateness of this type of inspection could be questionable considering the difficulty associated with visually detecting compression failures and shakes in wood components. If the compression failures and shakes existed prior to the accident, then once the wood was subjected to a load in excess of its reduced load-bearing capacity, it would have catastrophically failed without any warning.

The effect of any mis-rigging of the aircraft's upper right wing could not be determined. However, if the centre section was out-of-square, then the right upper wing may have carried extra loading for an extensive period. Although the right-wing rear spar had evidence of significant weakening around the fuselage attachment fitting, it was considered unlikely that this was the area that initiated the wing break-up because the spar was predominantly subjected to compression loads at that point.

Witness evidence and wreckage disposition indicated that the right upper wing failed while the pilot was pulling out from a loop. The wing failed in the area of the right upper wing spar where the inter-plane strut was attached. Evidence indicated that the upper right-wing spar was significantly weakened around the inter-plane strut attachment point by the effects of fungal decay and a partially de-bonded doubler.

Because the loop profile was described as being egg-shaped and the aircraft was possibly being flown at a lower height than normal, the pilot might have used more nose-up elevator control than usual during the pullout. The pullout from the loop may have induced a higher than usual g-loading on the wing structure, however, the loading could not be determined from the available evidence. In any case, the g-limits for the DH-82A were not published and the aircraft was not fitted with a g-meter. Therefore, the pilot was probably unaware of the aircraft's g-limit and of the g-loading he was putting on the aircraft structure just prior to the accident.

Despite the aircraft's flight manual requiring the wing slats to be locked before conducting aerobatics, it is possible that the pilot might not have locked them before commencing the looping manoeuvres. However, any opening of the wing slats should not have caused a serviceable wing to fail, although the upper right wing forward spar was already structurally weakened by fungal decay, delaminated doubler at the inter-plane strut area and possibly by pre-existing microscopic compression failures and shakes. A violent opening of the right slat may have applied some additional loading to the most critical area of the spar. If the slat deployed during the pull out from the loop, the additional loading may have contributed to the failure of the already weakened right upper wing spar.

CONCLUSIONS

Findings

The pilot may have conducted a more positive pullout from the loop than usual and in doing so probably applied a higher-than-normal g-loading to the aircraft. This, associated with the possible deployment of the wing slats, contributed to the in-flight failure of the already weakened upper right wing.

The pilot hired a de Havilland DH82A Tiger Moth VH-TMK to undertake a 30-minute local pleasure flight with a friend. Although the pilot was qualified to conduct aerobatics, he was not authorised by the aircraft's operator to do so during the accident flight. The aircraft departed Jandakot at approximately 1715 Western Standard Time (WST) and proceeded south to the training area. About 20 minutes later, witnesses saw the aircraft performing an egg-shaped loop at a lower altitude than usual. One witness reported that the aircraft had been conducting continuous loops. As the aircraft was pulling out at the bottom of a loop, witnesses heard a loud crack accompanied by a tearing sound. Two witnesses reported hearing three "bangs". A large, yellow object was seen to separate from the aircraft. The aircraft then appeared to stop and pitch nose-down before entering a spiral dive. The right wing was seen to fold back against the fuselage before the aircraft entered the dive. As the aircraft descended, the left wings folded back, shedding wing components. The aircraft impacted the ground in a near vertical attitude and caught fire. Bystanders were unable to assist the occupants.

Aircraft characteristics

TMK was a single-bay biplane with a wood and metal structure covered by fabric. Metal, aerofoil shaped, flying and landing wires braced the wings. Wing slats were mounted on the outboard leading edge of the upper wings above the inter-plane strut attachment points. British Aerospace, the type certificate holder for the DH-82A, reported that this area underwent the greatest bending stresses when the wing was placed under aerodynamic load and, therefore, determined the ultimate load limit of the wing. The ultimate load limit was determined to be 7.5g, although the aircraft's structural g-limit was not published in the DH-82A flight manual. TMK was certified in the normal and acrobatic categories.

The wing slats were lift augmentation devices that reduced the aircraft's stalling speed by about 2 kts. As the wing approached the stall, the aerodynamic centre of pressure moved forward, causing the slat to automatically deploy. Many pilots used the initial stages of this deployment as an indication that the wing was approaching the stall. The slat assembly was attached to nose ribs which, in turn, were glued to the upper and lower surfaces of the spar. The slats were locked in the closed position by a slat-locking lever located in the cockpit. The slat-locking lever had a positive locking mechanism to preclude inadvertent movement; however, it was known that if the locking mechanism was worn, the lever might disengage when the aircraft experienced g-forces during aerobatics.

The aircraft's flight manual required the slat-locking lever to be locked before the conduct of aerobatics. Engineering analysis by British Aerospace determined that slats deploying during a looping manoeuvre would not cause a serviceable wing to fail.

Discussions with experienced DH-82A pilots indicated that they had, on occasion, inadvertently conducted aerobatics with the slats unlocked or when the slats had opened due to the slat-locking lever disengaging from the locking detent. One pilot reported that although he had experienced a violent opening of the slats under heavy g-loadings, there had been no damage to the wing. British Aerospace reported that their archives indicated that there was no previous record of slat separation being implicated in the structural failure of a DH-82 wing.

Wreckage examination

Pieces of wood and fabric were strewn along the flight path from north to south, over a distance of approximately 750 m. Numerous small pieces of the wooden wing internal structure were found early in the wreckage trail along with the two yellow painted aluminium wing slats. A fierce fire consumed the aircraft after it struck the ground, destroying most of the remaining wood and fabric components. The engine was operating when the aircraft hit the ground. The large, yellow object seen separating from the aircraft could not be positively identified.

The fire had extensively damaged the slat-locking mechanism and the investigation could not determine whether the mechanism was previously worn or damaged. However, the slat-locking lever was found to be in the unlocked position. The investigation could not determine whether the slat-locking lever was unlocked during the aerobatics or became unlocked during the subsequent in-flight break-up or ground impact. Both slats were bent upwards in a V-shape around the centre attachment. The outboard part of the right-wing slat had additional deformation and contained a deep cut. The cut was consistent with the slat impacting either the right wing's flying or landing wires. Within the wreckage trail, the slats were found beyond the separated pieces of wing spar and internal structure.

Detailed analysis of the recovered pieces of the wing structure found that the upper half of the right upper wing front spar was affected by fungal decay. The centre of the affected area was just above the hole where the slat-locking cable passed through the spar.

A wooden doubler made from ash timber, attached to the front side of the spar at the inter-plane strut attachment, was delaminated from the spar. Engineering advice was that delamination of the doubler would increase the bending stress in the spar at the inter-plane strut attachment point by about 33 per cent. The grain slope of the doubler was also found to be extremely steep, which significantly reduced its load carrying ability.

The upper right wing rear spar root end had fractured along a line that ran through the centre of the outer row of fuselage attachment fitting bolts. The fracture surface was discoloured and the wood weakened by reacting with corroded attachment fitting bolts. Pieces of the front and rear spars of the upper right wing had numerous slip planes and creases in the wood cell walls, which indicated that the pieces had experienced compression overload.

When wood has been subjected to compression overload along the wood grains, the grains exhibit microscopic slip planes and creases (also known as failures). If a large section of wood is subjected to compression overload a well-defined visible wrinkle across the face of the wood, known as a compression shake, may be present. The presence of either microscopic compression failures or visible shakes seriously reduces the load bearing capacity of the wood. However, experience indicates that even the visible compression shakes may be difficult to detect. The evidence of a shake is usually associated with a sudden change of the spar's cross-section, which is often directly at the side of a doubler. Shakes could be extremely subtle and hidden by paintwork or other surface features that hinder their detection. Despite the difficulty associated with detecting compression shakes in wood, the aircraft's maintenance manual recommended that such inspections be visual, conducted through inspection holes in the wing's fabric.

A compression shake may result from abnormal bending overloads often experienced during relatively innocuous situations such as a heavy landing or a landing gear collapse. Unlike a crack in metal, a compression shake in wood does not progress during the aircraft's normal utilisation. However, once the wood is subjected to a load in excess of its reduced load bearing capacity, it may catastrophically fail without any warning.

Aircraft history

The history of the aircraft prior to a rebuild in 1980 could not be established. Since the rebuild it had been used for training, private flying, and commercial operations, including the carrying of fare-paying passengers on joy flights.

Prior to this accident, TMK had been involved in two other accidents. In 1991 the aircraft had a heavy landing and in addition to other damage, it was recorded that only the lower right wing required repair. The second accident occurred approximately 2 years later, when an engine failure resulted in a forced landing. In addition to other repairs, structural damage was such that the lower left-wing spars and the lower right wing required replacement. The right upper wing front and rear spar inter-plane strut attachment fittings were extensively damaged and required replacement. The inspection of the right upper wing critical points was conducted through inspection holes in the wing fabric. The licensed aircraft maintenance engineer (LAME) who conducted this inspection reported that the wing structure appeared old.

During the 1980 rebuild, the right upper wing front spar was recorded in the aircraft's logbook as being replaced with one manufactured by Perfectus Airscrew Pty Ltd. However, after the 1998 accident, the recovered right upper wing front spar components were identified as not being of Perfectus origin. There was no entry in the logbook to indicate that the right upper wing or spar was replaced during the period between the 1980 rebuild and the 1998 accident.

It was established that following the 1993 accident, the aircraft was difficult to rig and there were problems with the aircraft's flying characteristics. Despite minor wing rigging adjustments, no attempt was made at this time to review the aircraft's rigging in accordance with the aircraft's maintenance manual. The right rear centre section spar attachment point was, reportedly, three-eighths of an inch lower than the left.

In November 1997, the wings were re-rigged as a result of pilots reporting that the aircraft had a tendency to roll to the right when it was flown hands-free. The logbook showed that the aircraft wings were re-rigged in accordance with the aircraft's maintenance manual. However, the new maintenance organisation reported that the wings were not removed during the procedure. British Aerospace confirmed that the DH-82A maintenance manual required the removal of the wings to effect accurate rigging of the centre section. When the maintenance organisation re-rigged the aircraft, it did not find the reported misalignment of the centre section. Failure to correctly rig the centre section and wings could induce additional stresses into the wing structure.

Repair and maintenance aspects

During the course of the investigation, it was reported that significant structural defects and deterioration in other Tiger Moth aircraft had been discovered. Although these aircraft had been previously inspected, the defects found included: incorrectly manufactured wing spars; wing tie rods made from incorrect, and much weaker, material; cracked and deteriorated spars; corroded fuselage frames; incorrect materials used in the wings; and incorrect repair and construction techniques. These defects and the evidence found on the accident aircraft, appeared to indicate that periodic inspections on some aircraft were being conducted inadequately, and that some LAMEs were approving materials and work that were deficient.

Personal information

The pilot held a private pilot's license and had a valid class 2 medical category.

Video evidence indicated that the approach to the water drop was normal, although the airspeed approaching the target area was about 10 kts higher than the maximum recommended in the AFM for load release. The aircraft nose would have suddenly pitched-up during the release of the load and there was no evidence of any significant elevator input to counteract the change in nose attitude.

Although the pilot had only logged 7.6 hours operating aircraft equipped with the Fire Retardant Dispersal System, he was sufficiently experienced on the aircraft type to appreciate the magnitude of pitching moment likely to be encountered. Insufficient forward elevator had been applied during the release of the load to counter the tendency of the aircraft nose to pitch upwards.

There was no evidence to indicate that a mechanical defect or medical incapacitation had contributed to the lack of elevator input during the load release and subsequent climb. It was possible that the pilot had intended to climb the aircraft steeply after releasing the load in an attempt to increase the visual impact of the display, which was consistent with comments attributed to the pilot about his stated intention to put on a good display. The observed yawing and rolling to the left during the climb may have been an attempt by the pilot to turn the aircraft for another pass of the crowd.

The extension of flap during the climb would have created a significant amount of additional drag. Consequently, for the aircraft to reach a height of 450 ft with the amount of flap being applied, it was likely that the engine was operating at a high power setting and the propeller producing a significant amount of thrust and resultant torque.

It was not possible to determine whether the extension of flap beyond 10 degrees during the climb was intentional. The extension of the flaps probably reduced the likelihood of the manoeuvre being safely completed. Although the extension of flap during the climb may have caused the aircraft nose to pitch-up further than the pilot had originally anticipated, there was no evidence that the pilot had made an elevator input to reduce the steepness of the climb.

The ailerons would have become less effective as the airspeed of the aircraft reduced during the climb. The low airspeed combined with the apparent turning manoeuvre, reduced aileron effectiveness and high torque being produced by the propeller probably contributed to the aircraft's roll inverted and entry to the incipient inverted spin. Once the aircraft had entered the spin, it was unlikely that there was sufficient height available for the pilot to effect a recovery.

The pilot of the Air Tractor 802A (AT-802A) was scheduled to demonstrate the fire-fighting capabilities of the aircraft at the Mount Gambier airshow. After becoming airborne the pilot positioned the aircraft for the first pass of the crowd. This pass was made at a height of approximately 100 ft in a north easterly direction and overhead the runway that was being used as the display axis for the airshow.

The pilot then confirmed by radio to the airshow coordinator that he was starting his "drop run". The aircraft was observed to fly in a gentle descent towards the designated target area, and at a height of about 40 ft the load release commenced at, or close to, the maximum rate. During the load release the nose of the aircraft pitched up and the aircraft entered a climb. On completion of the load release the aircraft nose continued to pitch up and the climb angle increased.

The aircraft climbed straight ahead for a short distance before commencing to yaw and roll to the left. The bank angle increased to a maximum of about 90 degrees while the nose attitude dropped to almost the horizontal. At a height of about 450 ft and while at very low airspeed, the aircraft rolled inverted and entered the incipient stages of an inverted spin. Recovery to controlled flight was not achieved and the aircraft impacted the ground inverted, in a wings level attitude at a nose-down angle of approximately 45 degrees.

The aircraft caught fire immediately after it struck the ground. The fire was fed by aviation turbine fuel from the ruptured fuel tanks and was quickly brought under control by local firefighting services which had been on stand-by at the aerodrome. The pilot sustained fatal injuries. Impact forces and the ensuing fire destroyed the aircraft.

Wreckage and impact information

Fire had affected the forward fuselage, consumed most of the right wing and the inboard portion of the left wing. The left wing flap was at an almost fully extended position and the right wing flap was destroyed by fire. Examination of the wreckage did not reveal any mechanical defect that may have contributed to the loss of control.

An examination of the aircraft's propeller revealed that the blades remaining within the propeller hub were in an approximate coarse pitch setting. One of the blades had dislodged from the hub on impact with the ground and an adjacent blade had fractured in close proximity to the hub.

The engine and propeller were dispatched overseas to the engine manufacturer for further examination. Examination of the engine revealed no evidence of pre-impact distress or operational dysfunction. The engine damage was consistent with it producing high power at impact.

The engine manufacturer subsequently reported that they could find no record of receiving the propeller. Despite additional inquiries, the propeller could not be located and consequently it was not possible to conduct a detailed examination of this component. Therefore, it was not possible to establish if the propeller blade angle observed at the accident site was due to impact forces, a result of a malfunction, or because of a control input by the pilot.

Pilot information

The pilot in command was appropriately licensed and qualified to undertake the flight. He held a valid Commercial Pilot Licence and Grade 1 Agricultural Rating and had accumulated a total of approximately 11,354 hours aeronautical experience, including 182.5 hours logged in AT-802A type aircraft. The pilot was experienced in airborne fire-fighting operations and was professionally employed in that capacity.

The Civil Aviation Safety Authority (CASA) had issued the pilot with a class one medical certificate. The CASA Acting Director of Aviation Medicine reviewed the pilot's medical history file together with the pathologist's report of the post-mortem examination. He reported that it was unlikely there was a direct medical factor involved in the apparent loss of aircraft control.

The pilot had previously completed demonstration flights of the AT-802 aircraft. One of the owners of the aircraft reported that he had spoken to the pilot about flying at the airshow a few days before the accident. They discussed some aspects of this type of event, in particular the potential for a pilot to impulsively initiate an impromptu routine. The aircraft owner reported that the pilot said that this was not going to be a problem and appeared quite subdued about his participation in the event.

A number of people reported that they had spoken to the pilot on the day of the accident. He had given the impression that he intended putting on a "good display" and that he thought a high-speed load release would look spectacular. The pilot reportedly also commented about the high standard of some of the other display routines and that he would "pull something out of the box" to impress the crowd. Aircraft information

The AT-802A had a five-blade constant-speed propeller that was powered by a PT-6 turbine engine. The accident aircraft was specifically equipped to conduct airborne fire-fighting operations and was fitted with a computer controlled Fire Retardant Dispersal System. The system had the capacity to deliver high volumes of water through a pair of hydraulically operated, computer controlled doors at the base of the hopper at rates well in excess of conventional delivery systems. The pilot used a control panel in the cockpit to select the ground coverage rate and the quantity of hopper contents to be delivered. The hopper capacity was 3,104 litres.

The pilot had logged approximately 7.6 hours flying aircraft that were equipped with this type of dispersal system. The investigation could not determine the number of times the pilot had used the system or the types of delivery he had made.

The AT-802A wing was equipped with fowler type flaps, which extended to a maximum setting of 30 degrees. The approved Aircraft Flight Manual (AFM) recommended that during fire control operations the flaps be set to 10 degrees for approach and load release and that flaps may be used as an aid in turning when extended to a maximum of 8 degrees.

Extending the flaps beyond 10 degrees resulted in a significant amount of additional drag and flap extensions greater than 10 degrees was normally used only for landing. The flaps could be selected by the pilot to any position between 0 and 30 degrees using a switch, mounted just below the throttle quadrant, or by a toggle switch mounted on the control stick. Experienced AT-802A pilots reported that it was possible to inadvertently extend the flaps by unintentionally activating the switch mounted on the control stick. Extending the wing flaps resulted in a conventional nose-up pitching moment.

The AT-802A type aircraft was certified by the US Federal Aviation Administration as an aircraft for "special purpose operations". Flight-testing during the certification process assessed the aircraft as being compliant with the Federal Aviation Regulations (FAR) which required an aircraft to demonstrate satisfactory aerodynamic stalling characteristics. Because low altitude agricultural type operations were considered to significantly reduce the probability of recovery from a spin, the aircraft's compliance with the FAR relating to satisfactory spin recovery characteristics was not required to be assessed.

The AFM for the AT-802A prohibited acrobatic flight manoeuvres, including spins. The manual also noted that during fire control operations the load release should be conducted at an airspeed between 109 and 113 kts and recommended that 10 degrees of flap be used to approach the target area and for the load release. In addition, the AFM advised pilots to "be aware that during the load release there will be a sudden pitch-up of the nose of the aircraft" and to "begin forward motion on the control stick as soon as the drop button has been activated".

Pilots experienced on the AT-802A reported that the intensity of the pitching moment depended on the aircraft's speed and the rate at which the hopper was emptied. The most significant pitching moment occurred when the full hopper contents were released at the maximum rate, at an airspeed exceeding 125 kts. It was also reported that a pilot experienced on the AT-802A should be able to anticipate the intensity of the nose pitch and accordingly, could be expected to safely control the climb profile of the aircraft.

Weather conditions

The weather conditions at the time of the accident were generally fine with a light to moderate south easterly wind. The temperature was about 23 degrees C and there was scattered cloud at 3,000 ft. The prevailing weather conditions were not considered to have been a factor in the accident.

Video & photographic information

Analysis of video and photographic evidence revealed that the aircraft approached the designated target area with about 10 degrees of flap extended and at an airspeed of about 125 kts. The elevator remained approximately in a neutral position during the release of the load and the aircraft nose commenced to pitch up, reaching an angle of approximately 45 degrees on completion of the delivery. The wing flaps extended to at least 25 degrees during the first part of the climb with the elevator remaining close to a neutral position. The climb angle then progressively steepened to about 70 degrees. Pilots experienced on the AT-802A assessed the initial delivery of water and foam to be normal, however the subsequent aircraft climb profile was abnormally steep.

After the aircraft had rolled inverted, it adopted an almost flat attitude, consistent with the incipient stages of an inverted spin. Movement of the elevator control was evident during the initial stages of the spin, however due to the resolution of the video recordings, it was not possible to conclusively assess any other movements of the control surfaces.

The two eyewitnesses gave conflicting views on whether both skids were off the ground. The witness who reported that the right skid did not leave the ground during the attempted lift-off was a helicopter flying instructor. His evidence was given greater weight because he was an experienced observer of helicopter operations, having daily monitored solo student helicopter pilots over several months.

In the absence of any evidence of flight control malfunction, strong winds, or intentional pilot input, the circumstances of the accident were consistent with dynamic rollover.

Dynamic rollover can occur when a helicopter is in the hover with one skid touching the ground. If the helicopter is allowed to roll to one side, pivoting around the skid on the ground, the thrust from the tilted rotor will apply a sideways force to the top of the helicopter, causing it to continue rolling. Rapid pilot input is then necessary to prevent the main rotor striking the ground.

For dynamic rollover to occur, one wheel or skid must be touching the ground or a fixed object. At the accident site, there were no marks on the tarmac to suggest that either skid had been restricted in its movement when the helicopter attempted to lift off. However, eyewitness evidence suggested that the right skid remained in contact with the ground during the accident sequence.

Due to the pilot's low number of flying hours on helicopters and his lack of recent helicopter flying, he may not have recognised that a dynamic rollover situation was developing or may have been slow to apply appropriate corrective action.

The rescuers reported that although the pilot's shoulder harness was not secured, the lap belt was when they attempted to move his body from the aircraft. The pilot had either not secured his shoulder harness before starting the engine, contrary to the flying school's standard operating procedures and aircraft checklist or, for reasons unknown, had unsecured the shoulder harness prior to the lift-off sequence.

The right side of the cockpit occupied by the pilot was largely undamaged; therefore, the accident was most probably survivable. The evidence suggested that the pilot's head injuries were caused when his head struck the cockpit structure near the cockpit roof. Although a secured shoulder harness would not have prevented the pilot's head from contacting the right door, it would have reduced the upper body movement and therefore may have reduced the severity of the pilot's injuries.

The available evidence indicates that the aircraft arrived in the area of the mine later than planned, but not having encountered any difficulties. When flying over the mine the pilot was probably flying with reference to the ground lighting and then the runway lighting. By flying to the right of the runway lights he would have been in a good position to examine the windsock in order to decide which direction to make a landing and to determine the wind velocity.

Having made this assessment, it is likely that he then commenced a right turn to track from the runway to join on the downwind leg for a landing toward the south-east. The impact location and direction are consistent with such a manoeuvre. As the pilot commenced the turn, he would have lost visual reference with the runway and other lights. This would have required him to fly the aircraft solely with reference to the cockpit instruments. The attitude of the aircraft at impact indicated that he did not maintain control of the aircraft sufficiently to prevent it entering a steep descending turn.

The pilot submitted a flight plan indicating a planned departure from Cobar at 1500 ESuT (1400 EST). A refuelling stop was to be conducted at Windorah. The planned arrival time at Osborne Mine was 1830 EST. Last light in the area was 1931.

The aircraft left Cobar at about 1455 EST. (The reason for the late departure was not established.) The planned flight time to Windorah was 2 hours 40 minutes. The aircraft was on the ground at Windorah for about 1 hour 30 minutes, apparently because the passengers walked to the township. Refuelling was completed at Windorah at about 1845. The planned flight time from Windorah to Osborne Mine was 1 hour 30 minutes. The pilot contacted the mine by radio and reported that he would be arriving at 2030. The runway lights were then activated by the mine staff.

A witness at the mine saw the aircraft, with navigation lights operating, fly overhead at an estimated height of 300 ft above ground level, considerably lower than the normal aircraft altitude. The aircraft was visible in the glow of the lights at the mine. A short time later, the witness was in a position to see the runway lights, and noted that the aircraft was to the north of the runway. He then lost sight of the aircraft as he drove the remaining distance to the strip. Later, he reported to the mine's communication centre that the aircraft had not landed. A formal search was commenced at 2100 when the pilot failed to cancel his search and rescue watch. A satellite which monitors transmissions from emergency locator beacons detected a beacon signal at 2132, when it passed over the accident area. The aircraft wreckage was subsequently located about 400 metres north of the airstrip.

Wreckage examination

Examination of the wreckage indicated that the aircraft struck the ground at a high rate of descent, and banked about 50 degrees right. Aircraft speed at impact was estimated at about 100 kts, and the engine was developing moderate power. No fault was found in any aircraft system which might have contributed to the accident. The impact was not survivable.

Pilot experience

The pilot was the holder of a private pilot's licence and a current medical certificate. He held a Night Visual Flight Rules rating and had accrued 30.4 hours of night flying experience. To act as pilot in command of an aircraft under these rules it was necessary for the pilot to satisfy a number of recent experience requirements. These included one hour flight time at night in the previous 12 months; one take-off and landing at night in the previous six months to fly without passengers; and three take-offs and three landings at night within the previous 90 days in order to carry passengers in the aircraft. According to the pilot's logbook he had not met any of these criteria. His most recent night flying had been conducted in late July 1997.

Flying conditions at Osborne

Some high cloud was present in the mine area. Visibility was good, but the night was dark with no moon and no visual horizon. The wind was blowing from the north-east at right angles to the runway and about 10 to 15 kts in strength.

A parking area and lit windsock were located on the southern side of the airstrip near its south eastern end. The runway lighting system contained a series of lights which pilots could use as a glide slope indicator when landing towards the south-east. Since the pilot had never previously landed at the airstrip during darkness it is not known whether he was aware of this feature.

The pilot was flying the aircraft at an altitude of about 60 ft in a steep left orbit near the Elizabeth River. When the pilot reversed the direction of turn, two of four aileron segments, and what appeared to be a small piece of fabric, fell from the left wing. The pilot later reported that the control column locked after he heard a loud bang. He retarded the throttle as the aircraft dived out of control into the river. Impact with the water destroyed the left wing but the cabin and right wing remained largely intact. The pilot was able to extricate himself from the submerged cabin, but his efforts to free his passenger were in vain.

Several weeks after the accident, a helicopter search found two aileron segments still joined by the interconnecting torque tube. They were recovered from mangroves 80 m from the river. A third aileron segment from the left wing was not found.

The investigation found evidence of deterioration of the timber wing ribs/aileron hangers from which the ailerons were suspended. The outer layer of plywood displayed "tide" marks where the trailing edge of the wing ribs had been immersed in moisture. Some ribs were stained by black mould and had deteriorated due to timber rot. The aircraft had been standing under a shade cloth roof in hot, humid, tropical conditions for an extended period. Moisture had accumulated in the wing and had gone unnoticed.

About a year before the accident, two outboard ribs/aileron hangers had been broken when the left aileron was pushed against a hangar door. A subsequent repair had been carried out in accordance with factory repair instructions. Small sections of fabric had been removed from the upper wing surface to facilitate the repair and had been replaced afterwards. Fabric covering the adjacent third rib/aileron hanger had not been removed for inspection. The part was not recovered from the river.

Examination by a specialist from the CSIRO Forestry and Forest Products Division found that the rib/aileron hangers suspending the inboard aileron segment of the left wing were profoundly affected by two types of wood rot. The wood rot caused a significant reduction in strength of the timber laminates. In addition, over-tightened bolts holding the aileron hinge compressed the timber laminates.

Witness evidence indicated that there were no birds in the immediate area when the aircraft descended out of control. The wreckage was examined for any evidence of bird-strike damage. None was found. Two of three propeller blades were broken by impact with water and rocks. The pilot's assessment that the control column locked, was confirmed by discovery of a bent torque tube holding the inboard segment of left aileron which had streamlined in the airflow.

The operator's procedures did not ensure that the implications of the aircraft type change were properly communicated to those involved in the preparation and acceptance of the pallet. Consequently, the pallet was accepted for shipment without recognition that it was oversize for the aircraft.

When advised by the leading hand that the pallet could not be loaded, the load controller apparently focussed only on the available area within the cargo compartment. He had not recognised that the overhang prevented the pallet from being manoeuvred through the cargo door. The high level of ambient noise and assumptions made by both the controller and the leading hand may explain why neither understood the point the other was attempting to make.

The processing of the LIR was inadequate in that the defence intended by the leading hand's written confirmation of the aircraft loading was circumvented. The load controller, having assumed that the loading was in accordance with the original documentation, dispatched the aircraft without ensuring confirmation of the final loading status.