At the time of the accident, the Cessna 210 was for sale. The pilot had conducted a pre-purchase inspection, including an engine run, the day before the accident. On the day of the accident, the pilot and two passengers boarded the aircraft for an evaluation flight to a nearby aerodrome. Following an uneventful engine run-up, the aircraft was observed to take off from runway 31R and flew for approximately 2.5 km at low level. The pilot subsequently carried out a successful ditching into a water-filled quarry. The pilot and both passengers successfully exited the aircraft but one passenger, who was unable to swim, drowned before reaching land.

The aircraft was manufactured in 1965 and had accumulated 8,013 airframe hours. All required maintenance had been performed and the aircraft had a current maintenance release, which listed no outstanding unserviceabilities.

During most of the previous year, the aircraft was in short-term storage at Mt Gambier SA. Approximately four months before the accident, a rough running engine was noted in the aircraft logbooks. The problem was later diagnosed as a cracked number 4 cylinder. The cylinder was subsequently replaced and the aircraft's 100 hourly inspection was carried out on 26 November 1999. On 24 December 1999, the aircraft was refuelled at Mt Gambier and flown to Moorabbin, Vic. by a ferry pilot who reported that the engine and aircraft performed normally. Except for one short flight five weeks before the accident, the aircraft remained on the ground until the accident flight. The engine was started and ground run four days before the accident flight. During the ground run, the pilot, who was not the accident pilot, noticed that the fuel selector valve was stiff but did not record the problem on the maintenance release.

Examination of the aircraft after it had been recovered from the quarry found that the flaps were at approximately the 10 degree position and the landing gear was retracted. Visible damage to the exterior of the airframe was limited to distortion of the main landing gear doors. The line to the manifold air pressure gauge was broken at the engine firewall. The throttle was in the idle position, the propeller control was in the full fine position and the mixture control was in the fully rich position. However, as the recovery of the aircraft from the quarry probably distorted the engine mounts, the positions of engine controls were not considered reliable. The magneto switch was in the BOTH position and both fuel pump switches were OFF. All engine instruments displayed readings that were consistent with a non-operating engine.

Engine and propeller

After recovery, the Teledyne Continental IO520 engine was dismantled for examination and all appropriate fuel and ignition system components were removed and tested. The propeller was examined and its governor bench tested. No fault that could have produced a significant loss of power was found with either the engine or the propeller. No blockages or loose baffles were found in the exhaust system.

Fuel system

The fuel selector valve handle was found in the left tank position but the selector valve, which is connected to the handle by two linkages, was found in the position of partly open to the left tank. The movement of the valve was stiff and the linkages were worn, allowing the movement of the valve to lag behind the handle, thereby giving a false indication of the valve's true position. The inside of the valve was corroded and the detent in the left tank position could not be felt when the valve was tested.

Fuel system testing

The position of the fuel selector valve had been index marked on the valve before its removal from the aircraft. The valve was rotated to the indexed position four times during testing and the flow rate measured. The measured flow rates at the indexed position varied between a minimum of 90 pounds per hour (PPH) and a maximum of 280 PPH. The engine manufacturer's specifications for the fuel system stated that the required fuel flow at full power was between 136 and 146 PPH. The engine manufacturer advised that with the fuel flow restricted to 90 PPH, this engine would develop approximately 75% of maximum power. The owner's manual stated that normal cruise power setting should be between 65% and 75% of maximum power.

When all the fuel system components, except the fuel selector valve, were connected to a test rig, they were confirmed to be capable of delivering to the engine a fuel flow that was slightly in excess of the engine manufacturer's specifications.

Fuel

Approximately 60 litres of fuel was drained from each wing tank after the aircraft was recovered from the quarry. When tested, that fuel was the correct type for the aircraft. A small amount of free water was present in the fuel. Ethylene Diamine (EDA) was the contaminant involved in the aviation gasoline contamination problem that occurred in southern Australia over the December 1999/January 2000 period. There was no EDA detected in the fuel samples taken from the aircraft, nor were any of the deposits, that are typically produced by EDA, found in any fuel system components.

Pilot experience

The pilot had held a Commercial Pilot Licence since 1962 and an Air Transport Pilot Licence since 1982. He had flown more than 13,000 hours on a range of aircraft including multi-engined turboprop aircraft. Since 1997, he had flown approximately 1,200 hours on Maule floatplanes and in the three months to the accident, almost all his flying had been on this type. He last flew an aircraft from the Cessna 200 series approximately three months before the accident, however, he had not flown the accident aircraft since 1966. He was correctly licensed and endorsed for the flight.

Witnesses

Six people located around the aerodrome witnessed the aircraft take-off. Their evidence was consistent on the following points:

• the aircraft engine ran normally during start, taxi and run-up,
• the engine note during the latter stages of the takeoff and the initial climb was unusual for the aircraft type,
• the aircraft became airborne at a point between 2/3 and 3/4 along the runway,
• the landing gear began retracting as soon as the aircraft left the ground, and
• the aircraft's angle of climb after takeoff was lower than normal.

Of the five witnesses who heard the aircraft takeoff, four reported that the engine sounded as though it was running smoothly at less than full power. The fifth witness reported that the engine sounded uneven, as though it was "running on only four of its six cylinders".

Pilot's recollection

The pilot reported that on the day before the accident and the day of the accident, the aircraft's engine had started and run without difficulty. Both fuel tanks were dipped and found to be slightly less than one quarter full. For the accident flight, the fuel selector valve remained in the left tank position from the engine start until the aircraft ditched. The engine run-up, that did not include a full power check, had not shown any unserviceabilities with the engine. The takeoff was normal until the landing gear was selected up. The pilot recalled that at that point the aircraft experienced a gradual loss of performance and was unable gain any further altitude. He recalled that after takeoff the manifold air pressure gauge read 22 inches which is approximately 5 inches less than would be expected. The pilot reported that when he realised that the aircraft would not make it back to the aerodrome, he decided to land in the first clear area that he saw. When the water-filled quarry came into view, he decided that a ditching was the best option available to him.

Performance of accident aircraft

The evidence of the ground witnesses and the pilot was consistent on the following points.

• No sudden loss of performance occurred.
• The aircraft displayed lower performance than expected after takeoff; however, it was still able to climb to a height of between 50 ft and 100 ft.
• The aircraft was unable to maintain height during the later stages of the flight.

Data provided by the aircraft manufacturer indicated that a Cessna 210E aircraft with 10 degrees of flap extended, would become airborne after a takeoff run of 104 metres, under the same conditions as the accident aircraft experienced. Eyewitness statements indicated that the aircraft did not leave the ground until it was at least two thirds of the distance along the runway, a distance of 748 m. That represented a takeoff run 719% longer than expected.

The aircraft manufacturer advised that the operation of the hydraulic pump during the landing gear retraction cycle in a normally operating Cessna 210E aircraft consumed approximately 7 horsepower, which was 3.8 % of the engine output at the maximum power setting. That would result in a small reduction in climb performance during gear retraction. In addition, the drag of the landing gear doors, which open during the retraction cycle, would also reduce the aircraft's performance by a significant margin. If the engine was producing less than maximum power, the percentage loss of performance during the landing gear retraction cycle would be proportionally greater.

The distance required to stop the aircraft, if the takeoff had been rejected from the lift off point, would be approximately equal to the landing ground roll distance under the same conditions. From the lift off point, the aircraft had 750 m to run before the aerodrome perimeter fence. Half of this distance was sealed runway; the other half was level grass. The manufacturer's data showed that at maximum weight, 233 metres would have been required to stop on a grass runway.

Performance of pilot's most recent aircraft type

The Maule floatplane, the aircraft that the pilot had flown most often in the previous three months, at maximum weight had a take-off run of approximately 500 m from smooth water and longer from rough water. Compared to a fully serviceable Cessna 210E aircraft, the Maule floatplane normally required a longer take-off run and had a much lower rate of acceleration.

Related accidents

In 1984, a US registered Cessna 210 sustained total power loss in-flight and the pilot force landed the aircraft. A factor in the accident was the use of the auxiliary fuel pump in the high position (NTSB CHI84FA282). In 1976, an Australian registered Cessna 210 was damaged in a force landing after the engine lost all power as a result of incorrect use of fuel pump in the high position (BASI 197606853). The owner's manual stated that the electric fuel pump should not be switched to the high position during normal operation because richer mixture than normal will result.

In 1972, a Cessna 210 was damaged in a force landing after the engine lost all power. A factor contributing to that accident was the prolonged sideslipping of the aircraft with low level of fuel in the tanks. (BASI 197204960). The owner's manual advised that with fuel tanks one quarter full or less, prolonged uncoordinated flight can uncover the fuel tank outlets, causing fuel starvation and engine stoppage. It further advised that "prolonged" means more than one minute.

ATSB records contain reports of four accidents (BASI 197304110, 198101450, 198201391, 198802338) involving Cessna 210 aircraft in which vapour lock in the engine fuel system resulted in complete power loss.

History of the flight

The pilot hired a Beechcraft D55 Baron aircraft for travel associated with business commitments and submitted an instrument flight rules (IFR) flight plan from Bankstown to Warren and return. The flight to Warren proceeded normally. The pilot obtained an amended forecast before the return flight. The forecast indicated instrument meteorological conditions (IMC) for his arrival. Accordingly, he replanned via Bathurst in anticipation of an instrument arrival procedure to Bankstown.

The pilot departed Warren at 1628 ESuT and at 1705 reported 20 NM north west of Bathurst at 9,000 ft. That placed the aircraft inside controlled airspace without a clearance. The pilot was subsequently issued a clearance along the planned track. Approaching Bankstown the pilot encountered IMC and requested a Bankstown Radar Two arrival, anticipating a cloud break procedure from a radar vector. However, due to the missed approach flight path of that procedure conflicting with Sydney airspace requirements, the Sydney departures (west) controller, after determining that the pilot did not want the GPS approach, advised the pilot to expect a clearance for the Runway 11C Radar/Bankstown NDB/Sydney DME instrument approach. The pilot acknowledged the instruction. The approach controller subsequently observed the aircraft on radar to the right of the assigned approach track. He advised the pilot that he was right of track and cleared him to leave controlled airspace tracking to Bankstown along that procedure.

The pilot contacted Bankstown tower and advised that he was flying the Bankstown Runway 11C Global Positioning System (GPS) approach. The tower controller, who had been expecting the aircraft to be on a Runway 11C Radar/Bankstown NDB/Sydney DME approach, queried the pilot as to which approach he was using. The pilot confirmed that he was flying the GPS approach.

Recorded radar data showed that the aircraft had closely tracked the Runway 11C GPS approach path to a point 17 NM from Sydney, heading about 120 degrees at 2,500 ft. The aircraft then turned left to a heading of about 065 degrees and descended to 600 ft. The descent rate during that period was between 2,200 and 3,400 ft per minute. The aircraft then turned right to about 240 degrees and climbed to 1,100 ft. The minimum altitude for that segment of the approach profile was 1,400 ft. The Sydney departures (west) controller observed the aircraft on radar and advised the Bankstown tower controller. A short time later, the tower controller heard the transmission 'India Lima Mike emergency emergency'. Subsequent transmissions from the tower controller to the pilot went unanswered.

A witness, located approximately 2.5 km east of the accident site, reported seeing an aircraft pass overhead on a westerly heading with its engines surging before stopping. Other witnesses saw an aircraft apparently attempting to land in a nearby field. They described its approach as steep and slow. The aircraft descended into a grass-covered gully and impacted the ground. The impact collapsed the extended landing gear and the aircraft, although otherwise intact, was substantially damaged. The pilot, who was the sole occupant, received severe head and facial injuries.

Witnesses described weather conditions at the time of the accident as overcast with light rain falling and visibility estimated to have been between 3 and 5 km.

The pilot later stated that he had not previously flown a Runway 11C Radar/Bankstown NDB/Sydney DME instrument approach. Although he had acknowledged the controller's instructions for the approach to Bankstown, his intention had been to descend to the lowest safe altitude (LSALT) on that track and, if not in visual contact with the ground, to climb and divert to Bathurst. He reported that when he was not visual at 600 ft he commenced a climbing right turn onto a reciprocal track with the intention of diverting.

The pilot reported that after initiating the climbing turn onto a westerly heading, the left engine failed. He carried out the initial actions for engine failure but did not check the fuel selection at that time. He reported that checking the fuel selection was an item of his memorised trouble checks that in a multi-engine aircraft are performed after the initial actions and prior to feathering and securing the failed engine. However, before commencing the trouble checks the right engine failed and the pilot discontinued any further checks. He broadcast an emergency radio transmission and concentrated on controlling the aircraft. When clear of cloud, he manoeuvred the aircraft to avoid some towers and positioned the aircraft for a landing ahead, clear of houses and power lines.

The pilot later stated that he had intended to change from the auxiliary fuel tanks to the main tanks before commencing the approach. However, anxiety at having to fly an unfamiliar approach in IMC had distracted him and he had forgotten to change tanks. The pilot had not referred to either the approach or landing checklists that each included a check of the fuel tank selection.

Pilot experience, qualifications and recency

The pilot held a Commercial Pilot Licence and a valid Class 2 Medical Certificate. His Command Instrument Rating was endorsed for ILS, LLZ, VOR and NDB approaches. A log book entry on 24 May 1996 certified him as competent to use the GPS for en route navigation only. His total instrument flight time was 129 hours. No instrument flight time was recorded in the 90 days prior to the accident. His instrument rating renewal on 13 April 1999 had included an NDB approach. He had subsequently recorded an NDB approach, in flight, on 13 August 1999 and had made two practice NDB approaches in a ground procedure trainer on 20 September 1999. Recent experience requirements in Civil Aviation Orders Part 40.2.1 specified that the holder of a command instrument rating must not carry out an NDB approach in IMC unless in the preceding 90 days the holder has flown that type of approach either in flight or in a synthetic flight trainer.

At the time of the occurrence the pilot was operating under the privileges of a Command Instrument Rating (Multi-engine). On 10 March 2000 the Civil Aviation Safety Authority promulgated Civil Aviation Order 40.2.3 "Private IFR Rating" that allowed private pilots who had received appropriate training to fly in IMC under conditions less strict than those required for an instrument rating. The Private IFR Rating specified a flight review at intervals of two years. Civil Aviation Advisory Publication 5.13-1(0) Private IFR Rating recommended that recent experience requirements of the command instrument rating be used for guidance as to instrument flight time and instrument approaches.

The pilot had recorded 45.7 hours on the Beechcraft Baron type. His initial Baron endorsement training was undertaken in the B58 model but all his recent time on type was in the D55 model.

Fuel system and management

The D55 Baron fuel system consisted of a separate main and auxiliary tank in each wing. A selector for each fuel system was located on the floor between the front seats. The selector had four positions marked OFF, AUX, MAIN and CROSSFEED. A placard on the fuel selector directed pilots to use the auxiliary tanks in level flight only. The Pilot's Operating Handbook advised pilots to preplan fuel and fuel tank management before the actual flight and to utilize the auxiliary tanks only in level cruise flight. The last item of the descent checklist was "Fuel Selector Valves - MAIN".

Single fuel quantity indicators for both the left and right fuel systems were mounted on the pilot's lower panel. A toggle switch on the electrical sub-panel enabled selection of the quantity indication for either the main or auxiliary tanks.

The later model Beechcraft B58 Baron has a single tank in each wing, simplifying fuel selection and fuel quantity indication.

Wreckage examination

Examination of the wreckage did not reveal any pre existing defect that may have contributed to the accident sequence. There was no fuel in the auxiliary tanks. Approximately 170 litres of Avgas was recovered from the main tanks. The fuel selectors and the fuel quantity gauge switch were selected to the auxiliary tank positions. The wing flaps were not extended.

The aircraft was certified for IFR flight and was equipped with a GPS receiver that met the requirements for conducting GPS non-precision approaches. Documentation found in the wreckage included a set of current approach charts but the investigation was unable to determine which approach chart the pilot had used for the approach.

Shoulder harness attachment

The upper attachment of the pilot's shoulder harness failed during the accident sequence. The harness had no inertia reel and required manual adjustment. The pilot reported that he had been unable to adjust the shoulder harness as firmly as he desired. Examination of the upper attachment found that the installation was not in accordance with the approved modification and the bolt had pulled through the window pillar (see photos Fig. 1 and 2 below). The attachment for the right seat shoulder harness conformed to the approved modification.

Figure 1: Photograph of the failed shoulder harness attachment showing (above) where the bolt had pulled through the pillar.

Figure 2: Photograph of the bolt and harness end fitting.

Attachments for the shoulder harnesses had been installed in January 1973. This was carried out to comply with Airworthiness Directive AD /GENERAL/28 "Safety Belt and Harness Installations" that was issued in May 1972. The AD required an additional, single, shoulder strap to be fitted to the front cockpit seats of all Australian registered aircraft. The aircraft's logbook indicated compliance with the AD in accordance with an approved design drawing.

Figure 3: Cross-section through pilot's window pillar as appeared on the approved modification drawing.

Figure 4: Cross-section through pilot's window depicting the installation as found on VH-ILM.

The investigation was unable to determine how and when this attachment was altered. Maintenance records showed that since modification the aircraft had been resprayed and the pilot's side window panel had been replaced. Either procedure may have required removal and reinstallation of the attachment. The attachment in the aircraft, as depicted in the diagram at Fig. 4, was significantly weaker than the approved method of attachment shown in the diagram at Fig. 3, thereby reducing the protection provided by the design. The fact that it was an incorrect installation may not have been apparent to maintenance personnel. The approved installation drawings were held by the organisation that originally performed the modification and were not readily available for reference by subsequent maintenance personnel.

Maintenance personnel normally referred to the illustrated parts catalogue for identifying correct components and assemblies. That catalogue, produced by the aircraft manufacturer, did not incorporate modifications by other than the aircraft manufacturer. As such, the modification of the shoulder harness installation did not appear in the catalogue. Approved maintenance data, as required under Civil Aviation Regulation CAR 2A, provided guidance as to the information and documentation required to assist licenced aircraft maintenance engineers (LAMEs) to carry out aircraft maintenance, including modifications.

Civil Aviation Regulation 50A entitled "Aircraft Log Book" stated in part;

"…the holder of the Certificate of Registration for an Australian aircraft must:

(a) keep a log book for the aircraft, and
(b) make the log book available, and other documents referred to in the log book, available to CASA and to persons engaged in maintenance on the aircraft…."

Investigation into a fatal floatplane accident at Calabash Bay, NSW on 26 July 1998 (Occurrence 199802830) also found incorrectly fitted seat belt attachments. The report noted that failure of these attachments might have contributed to the severity of injury to the occupants. In both the Calabash Bay accident and this occurrence, aircraft had been found to have been operated with incorrectly attached restraint systems.

Two Fairchild Industries Inc S227 aircraft, VH-EEP and VH-UUQ, operating under instrument flight rules (IFR) were inbound to Mackay at approximately 0408 eastern standard time. EEP was from Rockhampton, maintaining FL140 and was being followed by UUQ on the Brisbane track at FL160. Near the descent point, approximately 55 NM south of Mackay, UUQ was above and abeam EEP when the Swampy sector controller issued instructions for the crew of EEP to leave control area on descent. The lower level of controlled airspace was 4,500 ft. About 1 minute later the crew of UUQ requested descent and advised the controller that they had EEP in sight. The controller instructed the crew to descend to FL140 and then to FL130. As EEP was descending through FL130, as indicated by the aircraft's Mode C altitude readout on the controllers' radar display, the lateral distance between the two aircraft reduced to 4.5 NM while there was less than the required vertical separation standard of 1,000 ft between them. The required radar separation standard was 5 NM. There was an infringement of separation standards.

The controller advised the crew of EEP that "traffic is UUQ". That was acknowledged by the crew who also advised that they had UUQ in sight. The air traffic system short term conflict alert activated and the controller queried both crews with respect to their ability to maintain their own separation on descent. Both crews acknowledged and advised that they could maintain their own separation. At this stage UUQ was ahead and above EEP. The controller instructed the crew of UUQ to leave control area on descent. Shortly after both crews reported transferring from the Swampy sector frequency to the Mackay mandatory broadcast zone frequency. As EEP was passing 7,000 ft it entered instrument meteorological conditions and the crew lost sight of UUQ. The crew of EEP contacted the crew of UUQ to establish the relative positions of the aircraft and found that the UUQ crew had descended their aircraft so that it was below EEP. The crew of EEP reduced power and manoeuvred their aircraft in an endeavour to increase the lateral spacing between them and UUQ. The aircraft subsequently landed at Mackay.

Both the area forecast and the Mackay terminal area forecast indicated the possibility of instrument meteorological conditions below 10,000 ft during the period when the aircraft were expected to be in the area.

The controller managing the Keppel sector, adjacent to the southern boundary of the Swampy sector, controlled the aircraft before the crews transferred to the Swampy sector. The Keppel controller noted the similar groundspeed readouts from the aircraft and queried both crews with respect to their respective indicated airspeeds. The crews both advised their indicated airspeeds as 205 kts. The controller advised the crew of UUQ that as they were about 2 NM behind EEP and that the next sector would probably make them second in the arrival sequence and, "if you would like you can start to reduce speed back to about 20 kt groundspeed reduction would probably fit you nicely behind". The crew acknowledged the transmission and advised that they had that aircraft in sight and were gaining on it.

The Keppel controller did not instruct the crew of UUQ to reduce speed and was not responsible for arranging the arrival sequence into Mackay. If the controller had issued such an instruction the crew of UUQ would have been required to read back and comply with the speed requirement. The Keppel controller informed the crew that he would advise the next sector that they had EEP in sight and subsequently told the Swampy controller. The Keppel controller did not advise the Swampy controller that he had pre-warned the crew of UUQ to possibly expect a speed requirement for sequencing. The investigation did not establish why the crew of UUQ did not reduce speed.

The crew of EEP was operating on the Keppel sector frequency and heard the advice passed by the controller. The crew later reported that they did not hear the response from the crew of UUQ but expected that aircraft to be following them on arrival into Mackay.

Due to the early hour, the Swampy sector was not busy and there was little other traffic in the area. The controller at the Swampy sector was endorsed and rostered for duty on the adjacent Daintree sector, but was not endorsed for the Swampy sector. The rostered and endorsed controller for the Swampy sector had left the position for a break. In that situation it was normal practice for the controller managing the adjacent position to monitor the radio frequencies and communication links while the position was vacant. If a radio or coordination call occurred, an appropriately endorsed controller would be recalled to operate the position. Immediately prior to and during the occurrence, the non-endorsed Swampy controller did not recall the other controller. Although the controller was not endorsed on the Swampy sector, the standards, procedures and techniques used were common to both Swampy and Daintree sectors. The controller should have been capable of maintaining separation using either radar, vertical, lateral, longitudinal or visual standards or a combination of these standards. The controller later reported that he believed that visual separation was being applied and consequently did not ensure that radar or vertical separation standards were maintained while the aircraft descended.

The Manual of Air Traffic Services (MATS 4-5-1) detailed how responsibility for separation may be assigned to a pilot using visual separation. For arriving aircraft above FL125 a controller was to instruct the pilot of one of the aircraft involved to follow and track behind the other aircraft, provided the pilot has reported sighting the aircraft and at least one of the aircraft is on descent. This was particularly so when the following aircraft was faster. In this case, a controller should have confirmed that the pilot was capable of following the slower aircraft. The Swampy controller did not instruct either crew to follow the other aircraft nor did he confirm whether the pilot of UUQ could follow EEP. Prior to a controller issuing any control instruction requiring a pilot to keep an aircraft in sight, the controller should consider a number of aspects that may limit a pilot's ability to comply. One of the aspects related to restrictions on atmospheric visibility that may not have been apparent to the pilot.

The Aeronautical Information Publication (AIP) contained a number of references about the application of visual separation. The references were different to what was in MATS. Also, the reference that related to controlled airspace provided little guidance to assist pilots in the application of visual separation. AIP ENR13-3 paragraph 3.2.1.d stated, "under certain conditions, the pilot of one aircraft may be given the responsibility for separation with other aircraft. In this circumstance, the pilot is also responsible for the provision of wake turbulence separation".

Controllers were responsible for assessing their fitness for an operational shift and if there were any doubts they were expected to notify a supervisor. The controller managing the Swampy sector had recently experienced some difficulties with obtaining satisfactory rest during his time off at home and had also been involved in a traffic incident the evening prior to the occurrence shift. The controller later reported that at the time he believed that he was capable of undertaking the shift despite his recent experiences.

Human performance varies during the day, tending to correspond with the body's circadian rhythm. Generally, the standard of human performance of some tasks decreases during the early morning hours. The reduction in performance is separate to that observed due to sleep deprivation. Additionally, an individual's ability to recognise the on-set of fatigue or a reduction in performance diminishes with fatigue and low points in the circadian rhythm.

History of Flight

The day before the accident, the Lockheed L-382G Hercules was being used to conduct a United Nations charter flight from Darwin in the Northern Territory to Dili in East Timor. During the approach for landing, when the landing gear was selected down, the main gear indication showed that the left main gear had not fully lowered. The crew checked the electrical and hydraulic systems but no fault was found. They also reported that after a fly-past of the control tower, the air traffic controller advised that the gear appeared down and locked. The crew then cycled the gear up then down and it lowered normally with the indication showing the gear down and locked. The landing at Dili was made without further incident. The flight engineer reported that he inspected the landing gear after landing at Dili and found no faults in the landing gear system. He said that he suspected that a micro switch might have been the cause of the indication. The aircraft operator did not have any maintenance personnel stationed at Dili and the crew did not report the problem to the operator's maintenance organisation.

The aircraft returned to Darwin the next day. At about 1000 (CST), while the aircraft was on approach to Darwin airport, the crew lowered the landing gear. The nose and right main gear indicators showed that the respective gear was down and locked but the left main gear position indicator showed unsafe. Still suspecting an indication problem, the crew raised and lowered the landing gear several times, but the left main gear indicator continued to show an unsafe condition. The crew conducted a fly-past of the control tower and the controller confirmed that the left main gear was not down. Having confirmed that the nose and right main landing gear operated correctly and that the left main landing gear would not move, the pilot in command allocated flying duties to the copilot. The pilot in command and flight engineer conducted the checklist actions and attempted to lower the gear using the emergency procedures. The attempt to lower the gear hydraulically by using the landing gear override selector valve was unsuccessful. An attempt to lower the gear using the manual drive failed because the emergency engaging handle could not be moved. The flight engineer unsuccessfuly attempted to manually move the shift lever on the forward gearbox of the left landing gear from "power" to "manual" and the loadmaster then attempted to lower the gear by disconnecting the universal joints on the vertical torque shafts of the left landing gear. However, the castellated nuts on the bolts of both wheel vertical torque shaft universal joints could not be undone without using a spanner. Even using a spanner, only two of the four nuts had been undone after about 30 minutes.

At about 1020, the crew of a C5 Galaxy military cargo aircraft also inbound to Darwin advised air traffic control that their aircraft was experiencing a hydraulic problem. Twenty minutes later, the crew of the L-382 informed air traffic control that the aircraft would be making a gear-up landing. The pilot in command requested that the airport's Rescue and Fire Fighting Service (RFFS) lay a foam path along the last two thirds of Runway 36. The air traffic controller informed the crew that a landing on Runway 29 was preferred, and that the L-382 was number two in the "emergency landing sequence". Air traffic control intended that the Galaxy land first followed by the L-382.

The air traffic controller later informed the L-382 crew that the airport RFFS advised that laying foam was not standard procedure. The controller also advised that if foam was laid, no foam would be available to attend the aircraft after it landed. The pilot in command was concerned about the potential for fire caused by sparks during the landing and he requested a clearance from air traffic control to perform the landing on the grass alongside the runway.

By the time two of the nuts on each of the universal joints had been undone, the fuel state of the L-382 was approaching 1,500 lbs or about 20 minutes endurance. The pilot in command decided that due to the low fuel state, there was insufficient time to undo the remaining nuts before a landing was required and he advised the controller of the aircraft's low fuel state. Concerned at the chances of the aircraft slewing off the runway after touchdown, the pilot in command also decided that the nose and right main landing gear would be raised for the landing. The L-382 was cleared to track for final approach Runway 29. The Galaxy diverted to the Royal Australian Air Force Base at Tindal.

The pilot in command subsequently decided that it would be more prudent to land on the runway because of possible obstructions on the grass area. After briefing the passengers, the crew conducted their own emergency briefing, including actions after touchdown, shutdown actions, and evacuation routes. The pilot in command assumed control of the aircraft during final approach and conducted the gear up landing.

At 1104, the L-382 landed on Runway 29. Touchdown was made at approximately 90 kts and the aircraft slid about 300 metres before stopping adjacent to Taxiway D. The aircraft remained straight on the runway and none of the crew or passengers were injured. They evacuated the aircraft soon after it came to rest. Although a flash fire erupted at the rear lower fuselage area while the aircraft slid along the runway, the fire did not spread. The RFFS applied foam to the area around the aircraft after it came to a halt.

Damage to the Aircraft

The aircraft sustained extensive lower fuselage structural damage due to the scraping along the runway. All rib lower end-caps aft of the nose-wheel bay were damaged and there was evidence of a flash fire in the rear section of the lower fuselage. Some damage to the electrical wiring located in the lower fuselage area had also been sustained.

Weight and Balance

The aircraft remained within the published centre of gravity and weight envelopes throughout the flight.

Fuel

The aircraft taxied at Dili with about 17,600 lbs of fuel, for the flight to Darwin. The departure fuel load was estimated to be 800 lbs more than the amount needed for the flight and required reserves. The aircraft arrived at Darwin with over 8,000 lbs of fuel and landed with about 1,500 lbs of fuel remaining.

Personnel Information

The pilot in command had accumulated 9611 hours of flying experience and had about 4428 hours on the Hercules series of aircraft including the military C-130 and civilian L-382. The copilot had about 2,300 hours flying experience of which nearly 600 had been gained on the L-382. The flight engineer had 1,224 hours operating experience of which about 484 hours were in the L-382. The flight engineer reported that he had not conducted, nor had he witnessed, a practice emergency lowering of the landing gear. There was no regulatory requirement for him to conduct or witness a practice emergency lowering of the landing gear.

Operator Information

The operator conducted freight and passenger flights in several countries including remote areas of Africa, South America and Antarctica.

The operator reported that because many of its operations were in remote areas, crews were expected to assess aircraft serviceability according to the company operations manuals, Aircraft Flight Manuals (AFMs) and Minimum Equipment Lists. There was no published guidance available to the crew regarding the maintenance requirements following landing gear incidents such as the one they encountered during the approach to Dili.

Meteorological Information

The weather conditions were clear with a 10 kt breeze from the north-west.

The Main Landing Gear

The main landing gear of the L-382 comprises a main assembly on each side of the aircraft. Each assembly has two shock struts each equipped with a brake assembly, wheel and tyre assembly, torque strut, and ball screw retracting mechanism. The brake, wheel, and tyre assemblies are mounted on the bottom of each shock strut. The two shock struts are mounted in tandem and connected to each other by a torque strut.

Each main landing gear retracting mechanism has horizontal and vertical torque shafts, three gearboxes, a hydraulic motor, ball screws, and strut vertical guide tracks. An emergency manual drive is provided for extension and retraction of the main landing gear following a hydraulic system failure.

The left main landing gear gearboxes are mounted on the left wheel well structure over the forward and aft shock struts, and connected to each other by a horizontal torque shaft. A hydraulic motor and a manual gearbox are mounted forward of the drive assembly. A vertical torque shaft extends down from each of the two gearboxes to a universal joint mounted on top of each of the ball screws. Each ball screw is anchored through a ball bearing pillow block at the upper end, and through a trunnion in the shelf bracket at the lower end. A ball nut on the ball screw is connected to the strut lower flange. The vertical ball screw assembly is installed on each of the two main landing gear struts. Each screw is centred between the strut guide tracks, and mounted to the inboard wheel well wall through its upper pillow block and trunnion assembly at the lower end of the ball screw shaft. Consequently, rotation of the screw by the vertical torque shaft causes the ball nut attached to the landing gear strut lower mounting flange to travel up or down, directly raising, or lowering the landing gear.

Main Landing Gear Operation

When the landing gear selector in the cockpit is positioned either up or down, an electrical signal commands the selector valve to direct hydraulic pressure to the hydraulic motor mounted on the front of each forward gearbox. Each of the forward gearboxes rotate the attached vertical torque shaft and screw assembly while also driving the horizontal torque shaft to drive the associated rear gearbox. The rear gearbox turns the rear vertical torque shaft and attached ball screw assembly.

Lowering of the landing gear can also be achieved by several alternative methods, depending on the type of failure encountered. Following a failure of the selector valve, over-ride buttons on the landing gear selector valve can be used to manually direct hydraulic power to the hydraulic motors of the main landing gear. The main landing gear can also be lowered using the manual shift on the forward gearbox. Pulling the emergency engaging handle moves the shift lever on the forward gearbox from power to manual thereby engaging the manual gearbox and releasing a spring-loaded brake. The main gear should then free-fall. If the gear fails to free-fall, a hand-crank is available to wind the gear down into position. A malfunction that locks any component of the system could prevent the main gear from moving. Consequently, the AFM advises that the universal joints mounted on top of each of the ball screws be disconnected. The gear should then free-fall under its own weight. A wrench is provided to wind down the landing gear should it not free-fall once the universal joints were disconnected. The universal joints are connected to the torque shaft by bolts and secured with castellated nuts. The nuts and their associated bolts should be tightened during installation to a torque of 25 to 30 inch pounds.

The aircraft manufacturer reported that it was possible, though undesirable, to land the L-382 model Hercules with one landing gear leg down on one side and both gear legs down on the other.

Engineering Investigation

A field structural repair was carried out in Darwin before the aircraft was flown to Singapore for final rectification of the damage. The landing gear was secured in the down position for the flight. The engineers conducting the repairs visually inspected the landing gear and found no abnormalities. The aircraft was then jacked clear of the ground, and functionally tested the landing gear. They found that the left main gear failed to extend normally. When the engineers attempted to hand crank the gear down, they noted that there was a high resistance within the system and the left main gear would not lower. When the universal joints were disconnected on the left main landing gear assembly, the forward strut lowered freely but the rear strut remained up. On closer inspection and disassembly, the left rear main gear ball screw assembly was found to have excessive backlash and the grease on the ball screw was found contaminated with accumulated debris. The engineers also reported finding several defects within the disassembled ball screw assembly including; excessively worn ball inserts and numerous chipped and distorted bearing balls in the ball nut assembly. Three circular scores with deep gouges were found on the internal surface of the ball nut assembly return sleeve and the scores coincided with the positions of the bearing balls. The ball screw was also bowed. The engineering organisation concluded that the damage was consistent with the bearing balls not riding normally or freely along the sleeve, with the greatest resistance probably occurring when the bearing balls rode across the gouges. The examination found no faults in the left landing gear hydraulic motor or associated gearboxes. The operator and aircraft manufacturer reported that there had been no previous failures of the type of ballscrew fitted to the accident aircraft.

An inspection of the universal joint castellated nuts and associated bolts found that none of the nuts could be fully unwound without the use of a spanner. A subsequent materials analysis of one of the castellated nut and bolt units revealed that the thread of both the nut and bolt had been deformed by the imposition of a load or loads along the axis of the bolt. Excessive tensile loads being applied during the tightening of the bolts or a load caused, for example, by abnormal operation of the landing gear could cause the damage. The reason why the threads of the nuts and bolts deformed could not be determined.

The aircraft manufacturer reported that no records could be found specifically stating that difficulty was experienced in removing the nuts from the bolts in the flanged connection.

The manufacturer had, however, introduced a torque shaft with a quick release feature that replaced the nuts and bolts. The manufacturer reported that the feature was introduced as a product improvement to make it easier and safer to disconnect the torque shaft from the ball screw.

Maintenance

A review of the aircraft maintenance documentation revealed no precursory event or events that may have indicated to the operator or the flight crew that there was an impending problem with the left main gear. No landing gear system faults were recorded during the previous block check. The manufacturer advised that the inspection requirements of the L-382 included a daily, a "B" check (the earlier of every 6 months or 600 flying hours), and a "C" check (the earlier of 2 years or every 2,400 flying hours). The daily and "B" checks required the checking of ball screws for general condition, cleanliness and lubrication. The "C" Check required a more detailed examination. The aircraft operator reported that the approved company maintenance schedule for the L-382 aircraft utilised block checks that occurred every 600 hours flying time. The block checks incorporated the "B" and one quarter of the "C" checks. Consequently, the aircraft would complete the manufacturer maintenance requirements every four block checks. The aircraft underwent a block check 647 hours before the incident at Dili. The aircraft was operating on an approved ten percent extension to the servicing schedule and was due to be flown to Singapore after the return flight to Darwin to undergo the block check.

The damage found within the ball screw assembly could be identified only when the unit was dismantled. There was no maintenance requirement to disassemble the unit for an in-service inspection. The ball screw assembly was installed as a new component and the aircraft operator approved documentation indicated that life of the component was 6 years or 6,300 flying hours. The ball screw assembly involved in the accident had been in service for 5,506 hours and 65 months and was due for replacement during the next block check. The ball screws were cleaned every 50 flying hours and were due for cleaning within six flying hours after the flight to Dili.

The aircraft manufacturer reported that the failure of the main landing gear to lower properly during the approach to Dili should have been subject to a maintenance investigation using the trouble-shooting procedures detailed in the aircraft maintenance manual. According to the manufacturer, the main landing gear system should have undergone an extensive maintenance inspection including extension and retraction testing before the aircraft was released for continued operation.

Operating Procedures

The AFM, in part, stated, "If the main and nose landing gears fail to extend after normal actuation of the landing gear lever, attempt to identify the malfunction before making further attempts to lower the gear". The operator's standard operating procedures provided the procedure for emergency lowering the landing gear but did not provide any further elaboration on the AFM requirements.

The aircraft operator reported that landing gear indication problems had occurred on other occasions; mainly due to spurious electrical signals that cleared with the reselection of the landing gear position.

The crew had been provided with a mobile telephone but the mobile service in Dili at the time was reported as being unreliable.

Passenger Evacuation

One of the loadmasters in the crew gave a pre-flight briefing to the passengers before the departure from Dili. Once the decision was made to make an emergency landing at Darwin, the senior loadmaster briefed the passengers on emergency procedures for the landing. The loadmasters reported that procedures were conducted in accordance with the operator's loadmaster training manual and drill cards carried by each loadmaster.

Before landing, the passengers were briefed that the front (crew) door and left emergency exit number two would be used. They were told to move towards the nose of the aircraft after exiting and to remain clear of the aircraft propellers. The passenger next to the exit was briefed about operating the emergency door and only to exit after the propellers had stopped. The passengers complied with the evacuation briefings and no injuries were reported.

Rescue Fire-fighting Services

The Airport Services Manual provided a discussion and guidance on the technique of foaming runways for an aircraft emergency including gear-up landings. It noted that, "Flouroprotein foam, film forming flouroprotein foam and aqueous film forming foam are not considered suitable for runway foaming operations due to their short drainage time". The Civil Aviation Safety Authority (CASA) reported that all the Airport Fire Services in Australia audited by CASA use Aqueous Film Forming Foam. Consequently, Airservices Australia and the Civil Aviation Safety Authority had an agreed policy that foam paths would not be laid at Australian airports. This decision was due, in part, to the poor persistence qualities of the foam agent (the foam path would only exist for a short period) and because aircraft frequently missed the foam path during landing. Both factors reduced the effectiveness of laying foam paths. The investigation was also advised that an attempt to lay a foam path for the aircraft probably would have exhausted the total stocks of foam agent held by Darwin RFFS, leaving none to use on the aircraft after it had landed.