Over a period of four days the aircraft operated twenty sectors with a high pitch squeal emanating from the L2 door. The squeal reportedly occurred at the top of climb and during cruise and stopped when power was reduced and descent commenced. On some sectors the squeal was not reported. On other sectors it was controlled by lowering the cabin differential pressure. At times the squeal was so loud that earmuffs were worn by the cabin crew, and passengers were seated away from the area. The investigation was unable to determine who issued the verbal instruction which resulted in the earmuffs being issued.

After many attempts at rectification the aircraft was removed from service and ferried to Melbourne. During the flight a licensed aircraft maintenance engineer was able to remove some trim and closely inspect the door. He found that above certain pressure differentials and airspeeds the lower door seal was able to flutter. The door was examined and found to be rigged to the upper limit allowed by the maintenance manual. The door assembly was rerigged to the lower limit. During a subsequent verification flight, the seal was found to be secure. The adjustment required was small, the door being lowered less than four mm.

The investigation disclosed that many factors had inhibited the early detection and rectification of the problem.

These included:

• an inadequate system of recording cabin defects,
• flight crews did not always record that the defect existed,
• flight attendants gave incomplete reports to flight crews, - a lack of timely advice to maintenance management, and
• the difficulty in visually identifying the seal flutter.

The investigation also found there had been inadequate communication links between:

• flight attendants and the company's cabin safety management,
• the flight attendants union and the company cabin safety management,
• cabin safety management and maintenance, and
• flying operations, cabin safety management and maintenance.

While it is undesirable to have a defect existing for so many sectors, this particular defect was difficult to detect in normal service and was not found until after the door trim was removed and the seal closely inspected in flight. The major safety concern was that on some sectors the cabin crew wore earmuffs which could have prevented them from immediately responding to some other problem, should one have occurred.

Safety actions taken

The operator is introducing a system of recording cabin defects that will stand alone from the technical report used by flight crews. This will require appropriate certification to transfer technical items into the maintenance log and will also enable cabin crews to have access to a running history of cabin reports.

The operator has also instituted an enhanced communications structure that will allow pilot reports to be rapidly conveyed to maintenance management, ensuring direct contact between operations, maintenance and cabin safety management. The operator's cabin safety management has advised cabin crews to make immediate telephone contact with them in the event of similar problems occurring.

The BAe146 was inbound to Kununurra from Curtin and the crew had been listening on both the area and the MBZ frequencies. They heard the pilot of the C207 make contact with Perth and discuss SARTIME arrangements and give his destination as Broome. This information indicated the C207 might be conflicting traffic for their arrival. They assumed the C207 was still on the ground at that time. As nothing further was heard from the pilot of the C207, the crew of the BAe146 made several attempts to contact him. When contact was established the C207 was 12 NM from Kununurra on the 242-degree radial at 5,500 ft. The BAe146 was 14 NM from Kununurra on the 250-degree radial at 3,800 ft, on descent.

The pilot of the C207 reported that his aircraft was fitted with one VHF set. After making his departure report on the MBZ frequency he changed to the area frequency and became engaged in a discussion with Perth flight service about his SARTIME. The discussion took longer than expected and during that time he could not hear any transmissions from the BAe146 on the MBZ frequency. When he changed back to the MBZ frequency he heard the transmission from the BAe146, and he answered immediately. He had tracked south of the Kununurra - Broome track to remain clear of cloud. He does not believe he came close to the BAe146.

It is likely that the discussion between the pilot of the C207 and Perth flight service, which the crew of the BAe146 assumed had taken place whilst the C207 was still on the ground, was the one that occurred after the C207 had departed. Consequently, the BAe146 crew did not hear the taxi and departure transmissions that preceded that discussion. The fitment of only one VHF set prevented the pilot of the C207 from adequately monitoring the MBZ frequency and, as a result, separation between the two aircraft could not be arranged in a timely fashion.

The pilot of the PA28 called Essendon tower completing a left orbit of the central business district and requested a clearance direct to Point Ormond.  His clearance was conditional on remaining clear of the DC3 which he believed he had in sight.  Shortly thereafter, the pilot of the DC3 reported a 200 metre near miss with the PA28.

The PA 28 pilot explained that at the time he was asked by the tower controller if he had the DC3 sighted, he saw an aircraft some distance off to the right, which soon passed abeam.  He presumed this was the DC3.  Believing that DC3 traffic had passed, he was shocked to see aircraft navigation lights directly ahead.  Immediately he altered heading slightly to the right but as the navigation lights continued to approach at about the same altitude, he banked steeply right and descended to 1,700 ft.

The PA28 pilot subsequently advised that judging distance to the head-on DC3 lights was difficult because of the background blackness of Port Phillip Bay.   For the DC3 pilot, the near miss was so sudden he did not achieve any evasive action.  Until the PA28 banked steeply, the DC3 pilot did not see it, probably because the PA28 navigation lights merged with the myriad of background lights of Melbourne suburbia.

In hindsight, the PA28 pilot believes each aircraft would have been more easily sighted if aircraft landing lights were on.

The Air Tractor 502 was conducting spraying operations late at night in fine conditions, with scattered cloud, no moon, and a barely discernible horizon.  After levelling at approximately 200 ft from a pull up and procedure turn, the pilot reported that his attention was drawn to another aircraft which was spraying a nearby property.  Whilst the pilot was watching the other aircraft the Air Tractor struck the ground at an estimated airspeed of 95 knots.

Wreckage distribution and pilot evidence indicated that, following the initial impact, the aircraft became airborne for a further 200 metres, with no engine or electrical power, before again striking the ground and sliding to a stop. The pilot, who had been wearing a full harness, escaped without injury.  The aircraft was destroyed.

The instructor reported that a normal approach was flown, with the student as the handling pilot. The aircraft was subsequently flared normally but was not held off, descending onto the runway in a level attitude and touching down on its three wheels simultaneously. The nose of the aircraft continued to lower onto the runway until the propeller struck the ground.

The operator subsequently reported that the lugs on the nose gear upper housing, which retain the hinge pin for the lower fork assembly, had failed. The nosewheel, complete with its fork assembly, had separated from the nose leg. There was no evidence of any pre-existing defects with the leg. The instructor reported that the landing did not appear to have contributed to the failure. The reason for the failure was not determined.

The pilot reported that it was the first time he had flown the type of glider, and it had been over 3 years since he had flown a glider with water ballast. Prior to the flight he had forgotten to drain the water, and he had departed with the ballast tank about half full.

As he made his approach to land, during an out-landing, the water shifted causing one wind to drop suddenly. The wing tip hit the top of a fence, yawing the aircraft to the right. The landing gear was torn off on touchdown and the left-wing tip then dug in causing the aircraft to ground loop to the left.

The pilot reported that there had been some nosewheel vibration, described as slight, during the take-off and that the aircraft had wanted to yaw away from the centreline. He had been able to regain the centreline with the use of moderate rudder input. Otherwise, the take-off was reported as normal.

During the landing roll, following the completion of a solo navigation exercise, the nose of the aircraft descended further than expected when the pilot attempted to lower the nosewheel onto the runway. He opened the throttle to complete a go-round; however, the propeller began striking the runway before this could be completed. The go-around was abandoned and the aircraft brought to a stop on the grass to the left of the flight strip.

Inspection found that the nosewheel and folk were missing and that the aircraft had slid to a stop on the nosewheel-folk attachment plate at the bottom of the nosewheel strut. The nosewheel tyre had made contact with the lower fuselage at some stage. The nosewheel was not located during an inspection of the airfield. The left main-gear leg was found to be bent slightly rearwards. Both propeller blades were scored and bent at the tips. No other damage to the landing gear was evident.

The nosewheel and folk were found on a beach north of Perth/Jandakot, one week after the occurrence. It was apparent the nosewheel had fallen from the aircraft during the navigation training flight.

The nosewheel folk was made of carbon-fibre material and information from a composite expert indicates some the carbon fibres had been damaged when an excessive side-load had been applied to the nosewheel structure. Despite the damage, the folk did not fail completely at that time. It is likely this initial damage occurred prior to the flight on which the nosewheel fell off.

It is possible the damage was not evident to the pilot during his preflight inspection prior to the navigation training flight. Delamination of composite structures, following excessive loading, usually shows up as a white discolouration of the naturally black material. The black, composite, folk had been painted white to match the aircrafts colour scheme. This probably disguised any signs of delamination.

Additional loads, during the take-off and possible movement in flight, caused the damaged folk structure to fail completely and the nosewheel was then able to fall free, colliding with the lower fuselage on the way.

The type and extent of damage to the left main gear and the lack of any other damage to the nose structure indicates that the side loads were applied either during a landing with the aircraft yawed to the right or when the aircraft was turned to the right, on the ground, at too high a speed. It is unlikely the damage was the result of a heavy landing or collision with an object. The initial damage probably occurred during night flying training on the evening prior to this occurrence although, nothing unusual was reported. Neither of the possible reasons may have appeared worth reporting to the maintenance section at the time, particularly if there was no apparent loss of function.

At Youanmi, fuel is supplied in 200 L drums. It is the pilot's responsibility to move the drums and pump the fuel into the aircraft. The pilot reported that he did not add the extra fuel as the gauges indicated there was 1,270 lb of fuel remaining and this was more than the minimum requirement. Although there was a discrepancy between the expected fuel burn on the flight up and the apparent actual burn, he indicated he did not consider it was significant. He assumed the difference was the result of a minor gauging problem. The pilot had a 15-minute wait before the passengers arrived, and he was able to depart for Perth.

The return flight was uneventful until top of descent. As the pilot commenced descent the right engine lost all power. During his trouble checks the pilot noted that both fuel gauges read 240 lb. He was unable to restart the right engine and the flight was continued on one engine. The pilot reported the engine shutdown to air traffic services and indicated that he suspected fuel starvation. The aircraft was cleared direct to Perth. Although the pilot believed fuel might be the problem, he decided not to divert into Pearce air force base but instead continued to Perth. He did however, request to track via overhead Pearce. The pilot later observed that had the second engine failed, he could have glided safely to either Pearce or Perth.

Following the aircraft's arrival in Perth, and after the passengers had disembarked, the fuel gauge indications were reported as 40 lb on the left and fluctuating between 100 and 180 lb on the right. The aircraft was defueled and 15 L (26 lb) removed from the left tank and 3 L (5 lb) from the right. When all fuel was drained from the aircraft the left gauge read 0 lb and the right gauge 125 lb.

The pilot's fuel planning figure indicated the aircraft would use 1,574 lb of fuel for the return flight to Youanmi, without including any of the variable and fixed reserves. The fuel drain indicated the aircraft actually used 1,619 lb. The pilot's flight plan indicates that the time intervals for some of the legs flown were longer than planned. In addition, there was the 6% difference in fuel burn between the two flight planning programs considered. It is not possible to determine, without additional research, the accuracy of the program used by the pilot. These differences probably account for the additional fuel used.

An engineering investigation determined that both fuel gauges had been replaced 3 days prior to the occurrence and a fuel calibration completed. This calibration had been completed using a system common to general aviation (the gauges were reset to zero and known amounts of fuel added to check the accuracy of the indications) but not recommended by the manufacturer. The gauging system was reported as serviceable after this calibration. Following the occurrence, the right gauge was reset to zero and another calibration was completed using the general method. A subsequent calibration using the manufacturer approved method (special test equipment) found that the right gauge was under-reading by 25 lb.

The reason for the false reading reported on the right gauge during and following the occurrence was not identified.

One of the purposes of fuel flight planning is to provide pilots with a means of cross checking the accuracy of their fuel gauges. It would be unlikely that any aircraft would burn 43% less fuel than expected on any one flight (379 instead of 679 lb). Consequently, it should have been obvious to the pilot that the gauge readings were unreliable when he checked the quantities at Youanmi, particularly in the light of his general and type experience. Why the pilot did not identify the discrepancy as significant could not be determined from the evidence available.

The pilot reported that the aircraft was cruising at 7000 feet, when he heard a loud bang accompanied by severe vibration. He immediately transmitted a "Mayday". During a subsequent orbit, clear of cloud, the pilot saw a paddock with a likely landing strip. The throttle was closed and a glide approach commenced.

During the descent, the pilot noticed that the strip was obstructed with irrigation pipes. He then decided that the only viable alternative for a forced landing was on a nearby road. Vehicles were observed (headlights were on) travelling in the same direction and groups were widely spaced. During the landing roll (which was subsequently found to be downwind) the left wing struck a road speed sign at about 60 kts IAS.  The aircraft then slewed left off the road into a fence.

The pilot activated an Emergency Locator Transmitter which was fixed in the empennage. The transmissions were heard by the pilot of an overflying aircraft and reported to Brisbane Air Traffic Control.

The pilot later reported that the loss of power was due to a sudden failure of the No 3 cylinder barrel, occurring at the barrel to cylinder head shrink line.

FACTUAL INFORMATION

The aircraft was being operated as a scheduled passenger service from Singapore to Sydney, with the co-pilot as the handling pilot. Whilst cruising at flight level (FL) 370 over the centre of the Australian continent, turbulence was encountered. A diversion of about 30 NM to the right of track was carried out to avoid local storms, before returning to the original track. The seat belt signs had been on for some time and the cabin crew needed to resume the meal service which had been in progress. The pilot in command (PIC) requested a clearance to climb to FL390, which was approved. Conditions were smooth and clear at the higher level so the seat belt signs were turned off and the meal service resumed. The aircraft was being controlled by the number two autopilot.

Later, prior to reaching the top of descent, the co-pilot commenced the pre-descent review and briefing, which included a review of the aircraft fuel status. When the fuel system synoptic display was selected on the electronic centralised aircraft monitor (ECAM), a lateral fuel imbalance was observed, with the left inner fuel tank about 800 kg heavier than the right. The PIC elected to correct the situation and the fuel imbalance checklist was reviewed. The checklist required that the four fuel crossfeed valves be opened by depressing the four fuel crossfeed valve switches on the centre overhead panel, then turning off the fuel pumps in the lighter tank. The valve positions are monitored by observing the fuel system synoptic on the ECAM screen on the centre instrument panel.

The engine driven hydraulic pump push button switches are also located on the centre overhead panel, immediately above the fuel crossfeed valve switches. These switches are similar to the fuel crossfeed valve switches and are activated by the same push-button switching action. The engine driven hydraulic pump push button switches are not guarded to prevent inadvertent activation. All valve positions are monitored on the respective fuel or hydraulic system synoptic displays on the ECAM, on the centre instrument panel. Standard operating procedures require that, when a switch is manually activated, the respective valve position is monitored on the ECAM before any further switching actions are performed. This procedure is to confirm that the correct selection has been made.

The PIC reported that he raised his hand and placed his right index finger on a fuel crossfeed valve switch. Before depressing the switch, the co-pilot remarked that the fuel management was under automatic control, and in the forward transfer configuration at the time. The PIC removed his finger from the switch to refer back to the fuel system synoptic on the ECAM but then decided to continue with the fuel imbalance procedure. He then depressed what he thought was the number one fuel crossfeed valve switch, then the remaining switches in succession, leaving his finger on the last switch. He observed the white switch lights illuminate as each switch was activated but neither of the crew monitored the valve positions on the fuel system synoptic on the ECAM. After depressing the last switch, the PIC observed the green hydraulic system low pressure warning on the ECAM, heard one warning chime, together with the master caution light and alarm, and also heard the auto-pilot disconnect warning. He looked up and saw that he had inadvertently pushed all four engine driven hydraulic pump push button switches off and immediately switched them all back on.

When the autopilot disconnected automatically, due to the reduction in hydraulic pressure, the aircraft began to pitch up and the stall warning sounded twice. The PIC advised the co-pilot that he was taking control of the aircraft and gave a nose-down command to his side stick control, but did not utilise his side stick priority push button. At the same time the co-pilot also gave a momentary nose-down command to his side stick control before realising that the PIC had control. The aircraft pitched down, then up, before level flight was regained and the auto-pilot was re-engaged. During the event the aircraft climbed about 350 ft before returning to FL390.

The loss of hydraulic pressure to the hydraulically powered flight controls and the prevailing aerodynamic forces acting on the airframe, resulted in the aircraft pitching nose up. This configuration could not be reversed until hydraulic power was restored. However, during this period both crew had instinctively moved their respective side stick controls to a nose-down command. As the side stick inputs are summed, this action resulted in a rapid pitch reversal as hydraulic pressure was restored, and required a further nose-up correction before level flight was regained.

The resulting negative and positive vertical accelerations which occurred caused unrestrained passengers and cabin crew in the rear of the aircraft to be thrown about and injured.

ANALYSIS

The investigation identified a safety deficiency in that there were several factors which increased the potential for the inadvertent activation of the engine driven hydraulic pump push button switches. The switches were not guarded; they were similar in appearance to the fuel crossfeed valve switches; and were activated by the same push button switching action. They were also located immediately above the fuel crossfeed valve switches on the centre overhead panel.

The PIC had initially placed his finger on a fuel crossfeed valve switch. However, before he activated it, his attention was drawn to the ECAM fuel system synoptic on the centre instrument panel, when co-pilot commented that the fuel system was in automatic forward transfer. Whilst his attention was drawn to the ECAM his finger moved slightly higher and was then positioned over an engine driven hydraulic pump push button switch. Deciding that he would continue with the fuel imbalance procedure, the PIC looked up to the centre overhead panel to what he thought was the fuel panel and depressed the switch to the off position. He then continued to depress the remaining three engine driven hydraulic pump push button switches to the off position. Neither of the crew were alerted to the incorrect selection at that time as the ECAM fuel system synoptic was still displayed.

As the aircraft was relatively new, there would have been minimal internal hydraulic system leakage. The aircraft was flying in still air so there was little demand on auto-pilot input to the various flying control hydraulic actuators. The hydraulic system accumulators continued to provide some residual pressure whilst the main hydraulic pressure was bleeding off, so it took some time before the hydraulic pressure depleted sufficiently to trigger the hydraulic system low pressure warning, though the hydraulic pump low pressure warnings had occurred. This time allowed the PIC time to switch off all the hydraulic pumps before a warning was observed and corrective action taken.

SIGNIFICANT FACTORS

1. The four engine driven hydraulic pump push button switches were not guarded to prevent inadvertent activation.

2. Inadvertent selection of the four engine driven hydraulic pump switches to the off position resulted in the loss of hydraulic system pressure to the hydraulically powered flight controls. The autopilot disengaged automatically due to the reduced hydraulic pressure.

3. The crew did not monitor the fuel crossfeed valve positions on the ECAM fuel system synoptic, following each switching action, in accordance with standard operating procedure.

SAFETY ACTION

As a result of the investigation into this accident, the Bureau of Air Safety Investigation issued the following safety advisory notice SAN960163:

"Airbus Industrie, Civil Aviation Authority (Singapore), Bureau Enquetes-Accidents (France), Civil Aviation Authority (UK), Transportation Safety Board (Canada), Federal Aviation Administration (USA) and the Civil Aviation Safety Authority (Australia) should note the circumstances surrounding this occurrence and the safety deficiency identified by BASI's investigations."

SAFETY ACTION BY MANUFACTURER

Airbus Industrie, as a result of this advisory notice, and subsequent investigations by their own staff, undertook the following safety actions:

1. Issued a notification to all operators of the aircraft type of the details surrounding the occurrence.

2. As an additional precautionary measure, Airbus Industrie has decided to develop a modification introducing a guard on each engine driven hydraulic pump push button switch.