The aircraft crashed into the back yard of a property in Munro Street. Initial information indicates the aircraft was about to land at St George after a flight from Toowoomba. The aircraft operator advises that the aircraft was privately hired for a return flight to St George.

The aircraft took off from runway 23 at Ballarat for a test flight. The take-off was witnessed by many operations and maintenance personnel who work on the airfield.

Witnesses advised that shortly after take-off they heard some loud bangs from the engine, which some described as backfiring, and then engine noise ceased. They estimated that the aircraft was 300 ft above the ground at this point. The aircraft then turned steeply to the left without much loss of height. After turning through approximately 180 degrees and levelling, the nose dropped and the aircraft spiralled steeply towards the ground. Immediately before impact, the engine surged to high power. Impact was

approximately 250 m beyond the end of the runway and 206 m to the left of the extended centreline.

The aircraft was destroyed by impact forces and fire. The impact was not survivable.

The terrain beyond the end of the runway was flat farming land with dispersed buildings and trees but provided suitable forced landing sites. Extensive fencing was the main hazard to a forced landing in this area.

The pilot of the aircraft was a licensed aircraft maintenance engineer (LAME) and his passenger was an apprenticed aircraft maintenance engineer. Both worked at the aerodrome. The pilot was also the owner of the aircraft, having restored it to flying condition approximately 5 years prior to the accident.

The pilot commenced flying in 1972. He flew regularly and in December 1990 was issued with a rating to fly at night under visual flight rules. In December 1992, he obtained his commercial pilot licence. At the time of the accident, he had flown approximately 890 hours, 240 of which were in the accident aircraft.

Persons who worked with the pilot at the airfield advised that in November 1995, 3 months before the accident, the aircraft was flown interstate by another pilot. During that flight, the engine was reported to have hesitated, run rough and had reduced power available. The aircraft was examined by a LAME but no fault was found. After the aircraft returned to Ballarat, the owner removed the engine-fuel nozzles and fuel distributor valve for testing. The test was satisfactory and the owner reinstalled the components.

The aircraft operated until 20 February 1996 when, as a result of further rough running and reports of the engine cutting out on take-off, the pilot removed the engine-driven fuel pump, the fuel control unit, and the throttle assembly for testing. The bushes on the mixture shaft of the fuel control unit were replaced, and some minor lint contamination was cleaned from the fuel control unit filter; otherwise, the units were found to be serviceable.

The pilot reinstalled the units and, on 29 February, the day of the accident, carried out engine test runs. After some initial setting up problems, the engine was reported to have operated satisfactorily, and the pilot was observed to taxi out to the runway and take off. The take-off was reported to be normal until the backfiring that preceded the accident.

It was determined that there was sufficient fuel on the aircraft for the flight as it had only been flown for about 1 hour since having the main tanks filled. The investigation was able to determine that at the time of the main impact, the fuel selector was selected to the right main fuel tank. The investigation could not determine the distribution of the fuel within the tanks, nor the exact quantity on board.

No abnormalities were found during the examination of the engine and components except that five pebbles were discovered in the cockpit- mounted fuel selector. Three were located in three of the five fuel passages in the strainer body. The other two were loose in the cavity between the main rotor and the strainer body and were able to sit over and partially block the fuel-feed passage.

The investigation is to be continued by the Victoria Police.

Synopsis

This report outlines the circumstances surrounding a fatal accident involving a Cessna 337 aircraft near Albany, in Western Australia on 13 March 1996. The aircraft crashed during a low-level inspection of a bay on the coastline to the east of Albany.

The accident occurred after the pilot lost control of the aircraft at low level. Loss of control was precipitated by a loss of power on both engines whilst the aircraft was being flown in a maximum-performance turn.

Loss of power on the rear engine was the result of fuel starvation, probably caused by un-porting of the fuel supply line during prolonged unbalanced flight. The reason for loss of power on the front engine could not be determined although it is possible that the pilot inadvertently selected the front engine to off whilst attempting to change the fuel selection on the rear engine from the main to the auxiliary fuel tank.

History of the flight

Witness evidence indicates the aircraft and pilot were hired so the passengers, Federal and State officers, could complete an aerial inspection of some unidentified drums located in a small bay 45 km east of Albany. An attempt to reach the drums on foot had failed because of dense undergrowth.

The pilot's post-flight report form shows the flight departed Albany Airport at 0901. Passengers' watches indicate the accident occurred between 0920 and 0925. There were no witnesses to the route flown or the accident.

The aircraft was reported missing at 1500, and the wreckage was located at 1615, on the edge of the bay containing the drums.

Pilot information

The pilot held a current commercial pilot licence with a Cessna 337 type rating. He was approved to conduct low-level operations. He was experienced in low-level operations, having worked as an agricultural and whale-spotting pilot for many years. He should also have been familiar with the coastline in the area of the crash, having operated in the area as a whale spotter.

The pilot held a Class 1 medical certificate. Post-mortem examination did not disclose any medical condition that may have been a factor in the accident.

Aircraft information

Cessna 337C VH-FAM was manufactured in the USA in 1968 and placed on the Australian register on 16 October 1968. The aircraft had completed 5,390 hours time in service. The rear engine, Serial Number IO-360-C-10344, had completed 393 hours of its 1,500-hour service life. The front engine, Serial Number IO-360-C-10014, had completed 1,726 hours of its 1,800-hour service life. The additional 300 hours available on the front engine were the result of earlier maintenance which extended the service life of that engine. An inspection of the aircraft's logbooks revealed that it had been maintained in accordance with the manufacturer's and regulatory requirements. A valid maintenance release was in force and no unserviceable items were recorded prior to the final flight.

The Cessna 337 is a twin-engine aircraft with the engines located fore and aft on the main fuselage. The rear engine is reported by the manufacturer to be the critical engine. Single-engine performance is less with only the front engine operating rather than the rear. In addition, the front engine is positioned below the total thrust line and the rear engine above it. Any loss of power in the rear engine will result in a nose-up pitching moment.

The estimated weight of the aircraft at the time of the accident was 1,927 kg. Maximum allowable weight was 1,996 kg. The pilot and one passenger were seated in the front row. The other two passengers were seated in each of the other two rows. The centre of gravity was within the defined envelope.

Meteorological information

The surface wind at 0900 at Albany was recorded as coming from 080 degrees at 16-18 kts. Meteorological advice indicates that the terrain around the crash site is conducive to the formation of small-scale eddies or rotors which cause turbulence. The pilot of the search helicopter reported that, 7 hours after the crash, conditions in the bay at low level were very turbulent. Wind conditions appeared to be significantly affected by the surrounding terrain. It was reported that the average wind conditions tended to push any aircraft operating in the bay towards the hills around the bay.

Communications

A review of the air traffic service recording tapes indicates that no radio transmissions from the pilot were recorded before or during the flight. There was no requirement for the pilot to contact an air traffic service agency. Wreckage and impact information

On-site inspection

It was apparent from aircraft and foliage damage that the aircraft impacted at a nose-down angle of approximately 60 degrees with the wings level. The left wingtip contacted the ground first, because of the slope of the terrain, followed by the lower, forward fuselage. Damage to the cockpit/cabin area was extensive. The fuselage section, aft of the wings, exhibited only moderate damage. The wreckage remained upright and mostly intact after the impact. There was no wreckage trail.

The direction of the impact indicated that the aircraft was flying towards the north, following the coastline in the bay, when it crashed.

The front engine was torn from the fuselage and suffered considerable impact damage to the crankcase, cylinders, and accessories. Engine oil was found on the ground under the engine. The rear engine suffered minor damage only, although the engine mounts fractured on impact.

The front propeller drive shaft fractured during the crash. Whilst both blades on the front propeller and one blade on the rear propeller were bent, neither propeller exhibited damage consistent with being under power at impact.

The landing gear was retracted.

The position of the flap motor worm drive indicated the flaps were extended to half travel.

All control surfaces were present, and all systems appeared to be working correctly prior to the accident. All damage was consistent with being caused by the impact.

Both wings were destroyed. The main fuel tanks, two in each wing, were split at the seams and exhibited severe distortion caused by the movement of fuel. The auxiliary fuel tanks, one in each wing, were intact although each had been breached. They also exhibited severe distortion caused by the movement of fuel. The distortion made it possible to determine that the main tanks were almost full and the auxiliary tanks half full, at impact. The right auxiliary was the only wing tank still containing fuel (3 L) at the time of the inspection. All other wing fuel had drained from the damaged areas. Both fuel sumps, one in each tail boom, were intact and undamaged. Only the left sump contained fuel.

The fuel lines running from the main and auxiliary tanks to their respective engine fuel filters were intact (right tanks to the rear engine and left tanks to the front engine). The fuel supply line from the filter to the rear engine was intact. The line on the front engine had been severed although fuel in the line was trapped by crimping adjacent to the fracture and the angle at which the wreckage was lying. No pre-existing blockages were found in the fuel system.

Approximately 50 mL of fuel was removed from the fuel supply line, filter, and engine-driven fuel pump for the rear engine. This amount is representative of residual unusable fuel. The right fuel sump was empty although beach-marks indicated that it contained approximately 1.25 L of fuel at impact. Approximately 250 mL of fuel was recovered from the lines and filter for the front engine, the amount expected from a fully charged system. The left fuel sump contained 1.3 L of fuel and beach-marks indicated that it had been full at the time of impact. No water was evident in any of the fuel recovered. The fuel colour and smell indicated it was 100/130 avgas. The difference in the amount of observed fuel and fuel at the time of the crash was probably the result of evaporation. The on-site wreckage inspection took place 5 weeks after the crash (due to the unavailability of a suitable winching helicopter). Both fuel sumps were open to the atmosphere as a result of main tank damage and were lying at an angle which placed open lines above the fuel. The amount of evaporation was about the same for each sump.

Despite the damage to the main fuel tanks, the wreckage situation was such that it was possible for fuel to pool in the leading edge of the inboard left main tank following the accident. This fuel could have recharged the left fuel supply lines after the crash, leading to the fuel found during the post-crash investigation. A similar situation did not exist on the right side of the wreckage.

Although the cockpit/cabin area was destroyed, the engine control panel was recovered intact. The throttles, propellers and mixture levers were found set at a cruise power setting. There was evidence they had been locked in their pre-crash position by the impact. The ignition and battery switches were on, and the auxiliary fuel pump switches were off. The flap selector was set to just below the first detent. This corresponded with the position of the flap motor drive shaft.

The fuel selector panel indicated that the front engine was selected to the left main tank and the rear engine to the right auxiliary tank. The left fuel selector valve was also set to the left main tank: the cable was intact and exhibited little damage. The fact that the selected position corresponded with the valve position and that cable stretch was minimal indicates that the control position was probably the selected position. The right selector valve was set between the main tank and off positions. The cable was severely stretched and kinked. The centre wire had been pulled from a clamp on the actuating arm at the fuel selector panel. Stretching of the right fuel selector cable could move the selector valve from the auxiliary position, through the main tank and towards the off position. There was some restriction to movement of the right selector valve.

Propeller examination

The fractured drive shaft from the front propeller was examined in detail. No evidence was found in either the mode of failure (bending) or in the nitride coating on the shaft, to indicate the propeller was under any power at impact.

The rear propeller was still attached to the engine. There was no evidence of any rotational damage on the blades. There was some evidence that the rear of the spinner had rubbed against the engine cowl during the crash. The rub marks were minor, indicating that, whilst the propeller was rotating, it was not under power. The rear propeller was lying across one tail boom following the crash. There were no strike marks on the boom.

Engine examination

The engines were examined at an engine overhaul facility. The inspection indicated that both engines should have been capable of normal operation prior to the crash. No faults were found with the engine's electrical and fuel systems.

Aircraft fuel system

Fuel contents

The aircraft was refuelled to full main tanks 1 month prior to the final flight. At that time, the auxiliary tanks were estimated to be half full. The aircraft was kept locked in a hangar when not in use. The distortion of the fuel tanks during the crash confirmed that there was sufficient fuel on board at the time. There were no witnesses to the pilot's pre-flight inspection; therefore, it could not be confirmed that he had completed a fuel check, including a check for water. Fuel recovered at the crash site did not contain any water and there was no other evidence that fuel contamination was a factor.

Fuel caps

All fuel caps were fitted with vents and the vents were clear.

Fuel selector panel

The Cessna 337C fuel selector panel is mounted in the cockpit/cabin roof. There are separate selector knobs for the front and rear engines. The knobs are the same shape and size and are mounted longitudinally, with the knob for the rear engine at the rear.

To move the rear engine fuel selector from the main to the auxiliary tank position, the knob is turned one notch to the right. Moving the front engine fuel selector one notch to the right from the left main tank position, turns the fuel off. There is no bar to prevent accidental selection to the off position on either selector, as the knob must be moved through off to select crossfeed.

Selection of auxiliary on either selector depresses a button which changes the fuel gauge indication from main to auxiliary tank. The button is depressed as the lever moves and does not need to be depressed to move the lever. The normal method of fuel selection in the Cessna 337 is to operate the front engine from the left fuel system and the rear engine from the right fuel system.

The selector panel was dismantled and inspected. As was noted above, the operating cable for the rear engine/right fuel selector had been pulled from a clamp at the selector end of the cable. There were marks on the selector actuator arm which indicated that it had been pulled hard against the pinion gear shaft before the cable was pulled from the clamp. It was apparent that separation occurred during the crash process. Although the rear selector knob was set to the auxiliary tank position, it was free to move as a result of the separated cable. This, coupled with the witness marks on the selector actuator arm, indicates that the final knob position was not the selected position.

Fuel selector valves

The fuel selector valves were examined. Both valves appeared capable of normal operation. The restriction to movement reported in the right valve was determined to be excessive friction caused by a build-up of corrosion. It could not be determined if the corrosion was present prior to the crash.

The selector operation had been checked during the periodic servicing which took place 41 flying hours prior to the crash. Evidence was also available that the fuel selector controls and valves had operated correctly two flights prior to the final one. The selector had not been moved on the penultimate flight.

Fuel tanks

In the Cessna 337C the auxiliary fuel tanks, mounted inboard of the tail booms, feed directly to the engines via their respective selector valves.

The main fuel tanks, mounted outboard of the tail booms, feed the engines via a fuel sump located in each tail boom. Each sump has a capacity of 2.7 L and is gravity-fed from the main tanks. Fuel is drawn through an outlet, located in the top of the sump, by the engine-driven fuel pump. The outlet is displaced to the inboard side of the sump centreline. As a result, the unusable fuel, in a 45-degree banked turn, varies from 1 L to 1.25 L, depending on the direction of the turn. The unusable fuel in the right sump in a right turn is 1.25 L. The unusable sump fuel, in level flight, was measured at 500 mL. An auxiliary fuel pump is provided in the event of engine-driven pump failure.

Under normal conditions (balanced flight), unusable fuel is not a consideration as the fuel system is designed to keep the sumps full. In some circumstances (unbalanced flight), it is possible that gravity feed to the sump may cease. For example, in an unbalanced turn to the right, if the pilot introduces left rudder to help keep the nose up as bank angle is increased, gravity feed from the main fuel tanks to the right fuel sump will stop. The sump is above the main tanks and gravity and centrifugal force will tend to move the fuel downwards and away from the main tank outlet. Fuel will continue to feed to the left sump as the situation is reversed. Selection of the right auxiliary tank will overcome this problem as it bypasses the sump, and the outlet is on the lower part of the tank in a right turn.

At cruise fuel consumption rates (37.5 L/hour/engine), it will take just over 2 minutes for the rear engine to use the 1.45 L of useable fuel available in the right sump in an unbalanced turn. If the fuel selector is selected off or moved to a position were fuel supply is interrupted, tests indicate power loss will occur after approximately 9 seconds, as fuel supply is stopped downstream from the sump.

In the Cessna 337D and later models the auxiliary fuel tank is interconnected with the main tank and also feeds the sumps. This arrangement overcomes any possible fuel feed problems that might occur during an unbalanced turn.

Survival aspects

The crash was not considered survivable. Crashworthiness information indicates that aircraft occupants are unlikely to survive an impact at 60 degrees to the horizontal if the airspeed is more than 55 kts. As the stalling speed was probably in excess of 80 kts, impact speed would have exceeded 55 kts.

Although no evidence was found that the pilot had left a formal flight note with a responsible person, an associate of one of the passengers raised the alarm when the aircraft had not returned. Initial concerns were expressed to the manager of Albany Airport at 1330. A formal search was started by the Melbourne Search and Rescue Centre at 1500, once local attempts to establish the whereabouts of the aircraft had failed. The wreckage was located at 1615 by aircraft from the Western Australian Police Air Wing. The search aircraft initially flew over the wreckage without sighting it. The crew were directed back to the site by the signal from the emergency locator transmitter.

The aircraft was fitted with a fixed installation emergency locator transmitter that complied with TSO C91a. The transmitter was activated by the crash. However, its signal was not received by the search-and-rescue satellite until the day after the accident and therefore did not provide an early indication that a crash had occurred. The coaxial cable from the transmitter to the fixed aerial had been severed at the aerial. Movement of the coaxial cable during recovery of the bodies probably led to the signal being received by the satellite.

Organisational and management information

The operator held an appropriate air operators’ certificate. No organisational or management issues were identified as contributing directly to the accident. The drums were later found underneath bush on the edge of the bay. They were not visible from the air. A State Police helicopter was operating in the area of the crash site at the time but was not used in the search for the drums.

Terrain

The bay in which the accident occurred is open to the south-east and surrounded by hills up to 1,843 ft high. The average ground slope, which starts at the waterline, is 33 degrees. The bay is rectangular in shape and 300 m wide and 400 m deep. Calculation based on the manufacturer-supplied performance figures indicates a Cessna 337 cannot out-climb the terrain from inside the bay with only one engine operating.

Stalling speeds

The Cessna 337 owner’s manual details the expected stalling speeds for various configurations. At 1,905 kg the stalling speeds are:

• At 0 degrees of bank: 65 kts with flap up, 60 kts with flap at one-third and 55 kts with full flap.
• At 30 degrees of bank: 69 kts with flap up, 64 kts with flap one-third and 59 kts with full flap.
• At 60 degrees of bank: 91 kts with flap up, 85 kts with flap one-third and 78 kts with full flap.

Thus, stalling speed reduces with the application of flap but increases significantly with increasing bank angles. The estimated stalling speed for the aircraft configuration was between 80 kts and 85 kts.

Post-accident inspection flights and other operational information

During the investigation, flights were conducted in a twin- and a single-engine aircraft over the bay area. The pilot of the twin-engine aircraft reported he could not remain inside the bay in a right turn with less than 60 degrees of bank selected. He was operating at a higher speed (120 kts) than the Cessna 337's assumed operating speed. The investigator in charge inspected the bay in a Cessna 182. His pilot flew a pattern which he thought was the best way to view the bay: this consisted of commencing an orbit over the bay at 900 ft (clear of the terrain and turbulence). When the drums were not sighted, he descended in a continuous 30–45-degree banked turn, at 75 kts with flap set, to 500 ft (the limit because of turbulence and safety).

Discussion with experienced pilots indicates that during low-level inspections it is not unusual for a pilot to introduce top rudder in a turn to allow bank angle to be increased and thereby improve the view below the aircraft without losing altitude in the process.

Flight tests and calculations indicate that to operate a maximum-weight Cessna 337 with half flap set, with both engines operating at cruise power, in nil-wind conditions and at low level in the bay area, requires a steep turn of more than 51 degrees of bank and an airspeed of no more than 90 kts. Once committed to the turn inside the bay, a pilot would have no option but to continue the turn to exit. Any adverse wind conditions would cause an increase in the bank angle required.

Experience indicates that operations at low level in confined situations place pilots in a high workload environment where they need to concentrate their attention outside the cockpit. In previous investigations it has been determined that this type of operation often leads to poor airspeed control. It can also lead to the unsighted operation of ancillary controls such as fuel selectors, particularly if the pilot is experienced on the aircraft type and does not need to look at the control to confirm where it is or how it operates.

Information from fuel starvation occurrences indicates that an engine will usually start surging rather than just stop when the fuel quantity available is low, particularly when operating in turbulent conditions. Usually, the first action by the pilot in the event of a sudden and unexpected power loss in cruise flight is to move the fuel selector to another tank that has fuel in it. If there is no response, this action is usually followed by selection of the auxiliary fuel pump. Fuel pump selection varies, depending on aircraft type.

Tests and research

BASI Report 87-116 (Australian Aviation Occurrences Involving Fuel Starvation & Exhaustion 1969-1986) concludes that pilot factors were involved in 89% of fuel exhaustion occurrences and in 45% of the fuel starvation occurrences reviewed. It goes on to report that 71% of the factors involved mismanagement of the fuel system. The report looked at all types of general aviation aircraft, both single and multi-engine.

To further refine the information obtained from Report 87-116, the air safety occurrence databases of the Bureau and the US National Transportation Safety Board were reviewed to determine the types of events that led to loss of power to both engines in twin-engine aircraft in general and to loss of power in one or both engines in the Cessna 337.

The following are explanations of some of the terms used.

Fuel exhaustion occurs when all the useable fuel in the aircraft has been consumed. Pilot miscalculation is often the main factor, although there are some occasions where problems with the fuel system may give the pilot false information or fuel is lost overboard.

Fuel starvation occurs when there is still adequate fuel on board the aircraft, but it is not being supplied to the engine(s) for some reason. Mismanagement of the fuel system is often the main factor. However, there are occasions where such problems as fuel contamination or fuel blockage may be factors.

Un-porting occurs when the fuel tank outlet is uncovered, and air enters the system. A low fuel state or unusual manoeuvres can lead to un-porting.

Mechanical failure/malfunction includes failure of an engine component or accessory, low oil pressure, low fuel pressure, fouled spark plugs and rough running.

"Undetermined" covers those factors which could not be or were not determined by the investigating authority.

The events listed as "other" include one-off occurrences such as icing conditions.

The review of the Bureau's database covered the years 1969-1996 and included power-loss occurrences (accidents and incidents) in all types of aircraft and the Cessna 337 in particular.

Fifty-nine occurrences involving loss of power in both engines in all types were identified. Of these, 31 were the result of fuel exhaustion, 20 of fuel starvation, six for other reasons and two were of undetermined origin. Sixty-five percent of the fuel starvation occurrences involved pilot factors.

There were 67 occurrences involving a loss of power on one engine in the Cessna 337. Of these, 52 were the result of mechanical failure/malfunction, 12 were due to fuel starvation and three were for undetermined reasons. There were seven occurrences where there was a loss of power on both engines in a Cessna 337. Four were the result of fuel exhaustion and three involved fuel starvation. Two of the three involved pilot factors. The factors in the third were undetermined.

The review of the National Transportation Safety Board's database covered the years 1985-1995 and Cessna 337 accidents involving a loss of power on one or both engines.

There were 12 accidents involving loss of power on one engine and 23 accidents involving loss of power on both engines. Four single-engine accidents resulted from mechanical failure/malfunction, four were for undetermined reasons, two were from fuel starvation and two occurred for other reasons. The records for the multiple-engine failure accidents indicate 11 resulted from fuel exhaustion, eight from fuel starvation, three were for undetermined reasons and one was the result of un-porting of the fuel supply lines. A breakdown of factors in the National Transportation Safety Board's recorded accidents could not be determined from the information available.

No occurrences were identified, in either database, where a loss of power in both engines resulted from mechanical failure/malfunction.

The one report of un-porting in a Cessna 337 involved a pilot entering a wings-level, steep descent whilst there was minimum fuel in the tanks. Both engines stopped during the descent as a result of un-porting of the fuel lines.

The manufacturer reported that a military version of the Cessna 337C was used extensively as a forward air control aircraft in Vietnam without any similar fuel feed problems being reported. Forward air control often results in extreme flight attitudes.

Anecdotal evidence indicates that the rear engine of early model Cessna 337s occasionally stopped without warning during varying phases of flight. Some of these stoppages led to accidents when the loss of power was not identified early enough by the pilot. A common reason for these stoppages was not formally identified.

Overview

It is evident from information provided on the purpose of the flight, the location of the crash, the impact direction, and the damage, that the pilot was conducting a low-level inspection of the bay area in an attempt to find the drums. To complete this task, he had to fly the aircraft in a steep, right turn at a slow speed. During the turn both engines lost power. The loss of power led to a loss of control and the pilot was unable to recover the situation prior to impact.

Engine power loss

The lack of rotational damage to either propeller indicates that both engines had lost power prior to impact.

The lack of fuel in the system supplying the rear engine suggests fuel starvation contributed to its loss of power. The fact that the sump was depleted indicates that the fuel supply to it was interrupted. As no blockages were found, the most probable reason is that a prolonged, unbalanced, right turn stopped fuel feed to the right fuel sump from the right main tank. In less than 3 minutes, all useable fuel in the sump had been consumed and the engine stopped.

One flight conducted during the investigation indicated that the accident aircraft probably commenced an orbit over the bay area at a higher altitude than that used for the final circuit. When unable to see the drums, the pilot probably descended until he entered the bay on the final orbit at low level. As a result, the turning-time required to deplete the useable fuel in the sump could have been exceeded.

No direct evidence was available to establish why the front engine was not producing power. The investigation found that fuel was available and that all the engine systems were probably serviceable. The engine controls were all selected to the operating position at impact.

The research indicates there are no recorded occurrences where a loss of power to both engines resulted from mechanical failure/malfunction of the engines. The most common reasons for loss of power to both engines in a twin-engine aircraft (and in particular the Cessna 337) are either fuel exhaustion or fuel starvation. There was adequate fuel on board the aircraft; therefore, fuel exhaustion is not a consideration. As a result, fuel starvation is considered to be the most likely reason for the loss of power to the front engine. Pilot factors were identified in 45% of fuel starvation occurrences in all types of aircraft and in 65% of those involving twin-engine aircraft. In the absence of any evidence indicating a problem with the aircraft systems, pilot factors are considered the most probable contributors to loss of power to the front engine.

As it is common practice to select a different fuel tank following sudden power loss, it is possible that the pilot inadvertently selected the front engine off whilst he was attempting to restart the rear engine by changing the fuel tank selection. The design of the fuel selector switches and the pilot's concentration outside the cockpit may have contributed to his action. Fuel supply to the engine was probably re-established when the pilot realised his mistake and reversed the selection. Fuel then flowed backed into the lines, but the engine had insufficient time to restart.

Loss of control

The pilot had extended the flaps to reduce the stalling speed and increase his safety margin. Evidence indicates the aircraft was probably flying at 85-90 kts. This speed gave the pilot a small margin above the stall and allowed the aircraft to remain inside the bay. Under the circumstances, any interruption to engine power would have resulted in a sudden reduction in flying speed. This may have been sufficient to cause the aircraft to stall. If the rear engine lost power first, the nose-up pitching moment would have exacerbated the situation. The tendency for the prevailing wind to push the aircraft towards the hills could have resulted in an unconscious action by the pilot to increase bank and tighten the turn, thereby further reducing the safety margin. Turbulence may also have been a factor.

Impact sequence

The steep nose-down attitude indicates the aircraft was probably in an aerodynamically stalled condition for some time prior to impact. Considering the pilot's experience, the stalled condition probably resulted from a loss of control at inspection height. In a more controlled situation, the pilot would have attempted to manoeuvre the aircraft to a crash landing, and any last-minute stall would have been less severe. The relatively intact nature of the wreckage and the lack of severe damage to the aft fuselage indicates the loss of control occurred at low altitude.

Summary

As the right sump fuel contents approached the unusable level, it is likely the rear engine began to surge rather than just lose all power immediately. Re-establishment of fuel supply from the auxiliary tank would have corrected the situation and prevented complete power loss. The pilot probably attempted to change the tank selection. The lack of fuel in the rear fuel supply lines indicates that this did not occur. Although there is no substantive evidence to explain the loss of power to the front engine, it is possible the pilot inadvertently selected it off instead of selecting the rear engine to the auxiliary tank.

Failure of the rear engine alone may have been sufficient to cause the loss of control, particularly if the pilot was distracted from flying the aircraft by the engine/fuel situation. Failure of both engines at a critical point in a maximum-performance turn in a confined area will almost certainly lead to loss of control.

The low operating altitude probably prevented recovery from the loss of control situation before impact.

1. The task requirements and the terrain conditions meant the pilot had to fly the aircraft in a continuous maximum-performance right turn at low level. Whilst they approached the limits, these conditions were still within the aircraft's and the pilot's capabilities. As the aircraft was operating at or near the limits, there was little margin for error. The margin available was insufficient to prevent loss of control when the situation changed unexpectedly.
2. To improve visibility, the pilot probably introduced left rudder and increased the angle of bank, thereby creating an out-of-balance condition.
3. A prolonged, unbalanced turn probably led to fuel starvation and loss of power to one engine.
4. A sudden, unexpected loss of power during a maximum-performance turn resulted in loss of aircraft control.
5. The loss of power and control occurred at low altitude and there was insufficient height to effect recovery.

FACTUAL INFORMATION

The pilot had positioned the helicopter at "Fernleigh" homestead on the previous Friday and had not flown the helicopter again until the day of the accident. Flying commenced at 0645, with the helicopter being flown to "Bunburra" property to treat serrated tussocks. Operations ceased there at about 0945, and the helicopter was flown back to "Fernleigh". Further flights were carried out until 1200, when flying stopped for a lunch break. Each treatment flight took approximately 10 minutes, with the helicopter returning to reload with chemical.

After the pilot and loader/driver had finished lunch, they refuelled the helicopter. Two further flights were completed before the helicopter took off again at 1305. This was to have been the last flight in the treatment area before moving to a new location. The pilot had advised the loader/driver that during his return from the treatment area, he would inspect for regrowth another area he had previously treated. Before take-off, the property owner reminded the pilot of the presence of power cables in the area that he was going to inspect. The map location of the cables was not reviewed by the pilot. When the helicopter had not returned by about 1315, the loader/driver and property owner became concerned. Their initial search failed to find the helicopter, but shortly after, the owner of "Belmedie", an adjoining property, advised them that it had crashed.

The helicopter had struck power cables at a height of 26 m whilst tracking in an easterly direction at a calculated airspeed of 45 kts. The power cables were located in the area that the pilot had planned to inspect for regrowth. There were no witnesses to the actual flight path of the helicopter preceding the collision. A witness at "Belmedie" had heard and glimpsed the helicopter near the homestead. Weather conditions at the time of the accident were fine.

The accident site was located on the western side of an open area of rising ground. On both sides of the flight path lay heavily wooded hills. Obstructions to the flight path were two sets of diverging (approximately 46 degrees) 22,000-volt power cables, strung from a pole on the top of a hill to the left of the flight path. The first set, with a span of 224 m, went to the "Belmedie" homestead. The second set, with a span of 428 m, went to a pole on rising ground across and to the right of the flight path.

An engineering examination of the helicopter did not find any anomalies or defects that may have contributed to the accident. Damage to the helicopter structure was consistent with the main and tail rotors having come into contact with power cables. The main rotor blades had initially contacted the power cables, followed by the tail rotor. The cables then passed between the tail-rotor gearbox and the pitch change links around the tail-rotor drive shaft, before breaking. The dynamics of the cable strike resulted in the main rotor slicing off the tail boom forward of the tail-rotor gearbox. Other damage sustained by the helicopter was consistent with severe forces generated during the subsequent ground impact sequence. No emergency locator transmitter (ELT) was fitted.

ANALYSIS

The approach of the helicopter close to the homestead, en route to the inspection site, was consistent with a practice followed by pilots to locate and avoid power cables running to building sites. It would appear that, once having located the power cables to the homestead, the pilot continued towards the area to be inspected. The poles which supported the 428-m cable, which the helicopter struck, were located in heavily wooded areas. The cables could not have been easily seen, due to poor background contrast.

Initial contact with the power cables was by the helicopter main rotor blades, followed by the tail rotor, before the cables broke. The dynamics of the cable impact resulted in the main rotor slicing off the tail boom. As a result, the pilot experienced loss of control as the helicopter was now without directional control, coupled with a significant forward shift of the centre of gravity.

It is likely the pilot either forgot or did not see the second set of cables before colliding with them.

SIGNIFICANT FACTORS

1. The pilot did not adequately establish the location of the power cables prior to the flight.
2. The nature of the terrain in the vicinity of the power cables inhibited the capacity of the pilot to see them.
3. Flight control of the helicopter was lost when the tail boom and tail rotor gear box were severed from the helicopter.

Reports indicated that the pilot had intended to conduct a test flight in his gyrocopter to check the propeller installation. A short time later he was observed flying in an easterly direction at an estimated height of 400 - 500 ft above ground level, when two loud bangs were heard.

A search located the wreckage of the gyrocopter which was mainly intact except that its wooden propeller had separated from the engine crankshaft flange. Pieces of the shattered propeller, a section of rotor blade skin, and the propeller mounting hub/ pre-rotator drum were found along the flight path, within 230 m of the gyrocopters impact point.

Examination of the engine revealed the crankshaft flange had failed at the attachment bolt holes, allowing the propeller and mounting hub to separate and pass up through the rotor system, striking a rotor blade and tearing a large section of skin away from its spar. The gyrocopter then became uncontrollable and descended rapidly to the ground due to loss of rotor speed and aerodynamic lift.

The propeller attachment bolts, although bent, were still attached to the hub/pre-rotator drum. There were no indications that the bolts had been loose, or the propeller had broken prior to the flange failure.

The crankshaft flange was manufactured from a Subaru cast iron flywheel, machined to a thickness of about 6.5 mm, with an associated loss of material integrity. The flange probably failed due to accumulative stresses imposed on it during operations from fluctuating propeller loads, engine RPM changes, and abrupt aircraft manoeuvres.

The crankshaft flange was manufactured from an unsuitable material. No other faults or malfunctions were found with the gyrocopter, or its controls, which may have contributed to the accident.

FACTUAL INFORMATION

History of the flight

A witness heard the aircraft pass King Island aerodrome at 0455 EST at the same time as he noticed the pilot-activated 10/28 runway lights illuminate. The pilot reported to Melbourne Control that he would be completing a runway 10, non-directional beacon (NDB) approach. A short time later he broadcast that the aircraft was at the minimum descent altitude, which is 640 ft above mean sea level (AMSL) for a runway 10 NDB approach. He also broadcast that there was a complete cloud cover. The aircraft did not enter a missed approach procedure but was heard to fly towards the south-east from overhead the NDB, which is located 1.3 km south-south-west of the centre of runway 10/28. A second witness, located near the NDB site, reported observing the aircraft's lights to the south-east. At 0507 a farmer heard the aircraft pass low over his house shortly before it crashed into trees, 3.5 km south-east of the aerodrome. The first responders arrived at the accident site at about 0530. The pilot had not survived.

Damage to aircraft

Parts of the aircraft were torn off by tree and ground impact. However, the fuselage remained substantially intact until it was destroyed by fire after it had come to rest.

Pilot information

The pilot was correctly qualified and endorsed to perform the flight. He held an air transport pilot (aeroplane) licence, a night visual flight rules rating and a current multi-engine command instrument rating.

The pilot's total flying experience included 185 hours at night, of which 36 hours were as pilot in command, 6 hours as dual and the rest as co-pilot in twin turbine-engine regular public transport aircraft. He was approved to fly as pilot in command of charter operations to King Island in Piper PA 31 Chieftain aircraft on 9 May 1995 following company check flights with the chief pilot. The pilot had completed 104 flights to King Island, mostly in daylight, but not all in Chieftain aircraft. He had flown from Moorabbin to King Island at night as pilot in command of Chieftain aircraft on three occasions since 9 January 1996.

The pilot flew for the King Island operator on a part-time basis. His normal full-time employment involved flying Metroliner aircraft (a twin-engine turboprop aircraft with a maximum weight exceeding 5,700 kg) as co-pilot for a domestic airline. As co-pilot, he had accrued 1,837 hours. The pilot was on annual leave from his full-time employer at the time of the accident.

To be promoted to pilot in command of a Metroliner the pilot had to first accrue 500 hours as pilot in command of multi-engine aircraft operating under instrument flight rules. This was to comply with Civil Aviation Order 82.3. To meet this requirement, the pilot's full-time employer gave him permission to work part-time for the smaller regular public transport / charter operator. At the time of the accident, he still needed to accrue a further 247 hours.

It was reported that the pilot was well rested prior to the accident flight. The tape recordings of the pilot's radio calls, made shortly before the accident, do not indicate he was suffering any significant stress.

The pilot passed his last aviation medical examination on 23 February 1995. There were no restrictions on his medical certificate. He was not known to be suffering from any ailment.

No evidence was found to indicate that the pilot had ever experienced the combined conditions of flying an NDB approach to the minimum descent altitude, with the cloud cover at or near that altitude, and at night in very dark conditions with no ground lighting apart from runway lights. No evidence was found that he had flown circling approaches in simulators under simulated dark night conditions. However, during his experience as a co-pilot on turboprop aircraft, he had flown at night to aerodromes which had a low level of ambient light.

Meteorological information

The pilot received an AVFAX of relevant weather forecasts prior to the flight. He discussed the AVFAX contents with another company pilot who was also planning an early morning flight to King Island in a Chieftain. The AVFAX included the aerodrome forecast for Moorabbin, the relevant area forecast, and the aerodrome forecast for King Island.

At hourly intervals, the automatic weather station at King Island aerodrome measured and recorded wind velocity, air temperature, dew point temperature, QNH, and rainfall. These readings were transmitted electronically to the Melbourne control operator within one minute of their recording. Also, a trained weather observer estimated and recorded the cloud amounts and heights at three-hourly intervals.

At 0500 the King Island aviation special weather report was wind 320 degrees at 5 kts, temperature 15 degrees, dew point 15 and QNH 1003. No evidence was found that the pilot received the 0500 weather report. The 0300 and 0600 observations included a report of a complete cloud cover. No cloud base was reported at 0300. At 0600 the base was reported as 1,000 ft above ground level.

A post-accident Bureau of Meteorology assessment indicated that the meteorological situation at King Island at 0507 on 8 February 1996 would have included complete cloud cover with a base at 1,000 ft above ground level, or possibly lower, with the possibility of fog or mist.

A ground witness at King Island aerodrome reported that he did not see the aircraft or its lights when he heard it fly near the aerodrome at the beginning of the NDB approach. A policeman who drove to the accident site from Currie, about 20 minutes after the accident, advised that conditions were very misty and very dark.

A second pilot from the same company flew a Chieftain to King Island, arriving shortly after the accident. At 0525, he commenced the runway 10 NDB missed approach from 1.6 km west-north-west of the runway 10 threshold. He conducted the missed approach because the runway lights were intermittently obscured by cloud below 640 ft AMSL and there was fog in the area. The aircraft returned to Moorabbin.

Aids to navigation

The only ground-based approach aid for aircraft arriving or departing from King Island aerodrome is the NDB. Several other NDB approaches were conducted at King Island on the day of the accident. No problems were reported concerning the serviceability of the King Island NDB.

Communications

Communications between Melbourne Control and VH-KIJ en route to King Island were excellent. At 0445 the pilot reported that he was changing to the King Island mandatory broadcast zone frequency. At 0451 he reported to

Melbourne Control that he was going to conduct an NDB approach. At 0502 he advised that he would call Melbourne Control again by 0505. At 0506, the controller asked the pilot to confirm that operations were normal. When there was no reply, he attempted to contact the aircraft several times until 0510. After that he requested the pilot of the second Chieftain, estimating overhead King Island at 0510, to try to ascertain if KIJ had landed safely. No radio transmissions were recorded from the pilot of KIJ after 0502.

The last comment by the pilot of KIJ, made on the mandatory broadcast zone frequency and recorded on the King

Island aerodrome aircraft movement recording tape, was for another party to stand by. The operator's agent at King Island aerodrome, reported that he had attempted to contact the pilot by radio but was unsuccessful. The pilot's comment was probably a response to this transmission. It was not possible to determine the time that the comment was made. At no stage did the pilot advise of an in-flight problem.

Aerodrome information

King Island aerodrome had three runways. Only runway 10/28 was equipped with runway lights. The lights were pilot-activated and standby power was available. The runway 10/28 lights were activated when KIJ first passed the aerodrome and remained on until after the accident. Runway 10/28 was 1585 m long and 30 m wide. No visual approach slope indicator system was installed. Three unlit obstacles were clearly depicted on the Civil Aviation Safety Authority's aerodrome chart but only the NDB was marked on the Jeppesen charts used by the pilot. These were the NDB mast at 238 ft AMSL, a second mast at 173 ft AMSL, and third mast at 152 ft AMSL.

A mandatory broadcast zone with a radius of 15 NM was centred on King Island aerodrome.

During the investigation pilots reported that the area around King Island aerodrome had low ambient lighting at night. In hazy or misty conditions or when there was extensive cloud cover there was often no visible horizon and no lighting other than the runway lights to provide a visual reference point.

Terrain/obstacles

King Island aerodrome elevation was 132 ft AMSL at its reference point. The highest obstacle within the prescribed 2.66 NM circling area was the NDB mast at 238 ft AMSL. From the air by day, the terrain within the prescribed circling area looked quite flat. However, the tops of the trees involved in the initial impact were estimated to be 227 ft AMSL.

Runway 10 NDB instrument approach

The minimum sector altitude for a radius of 25 NM was 1,800 ft AMSL. The outbound track for category B aircraft (which includes the Chieftain) was 325 degrees for 2.5 minutes. Aircraft were required to be established on the inbound track of 130 degrees not below 1,300 ft AMSL. Minimum descent altitude was 640 ft AMSL. Circling minimum was 740 ft AMSL. A missed approach required a climb, from overhead the NDB, on 130 degrees to 1,800 ft AMSL.

The pilot used Jeppesen charts which showed the minimum descent altitude for the runway 10 NDB approach as 540 ft AMSL with actual aerodrome QNH set on the altimeter sub-scale, and 640 ft AMSL with forecast QNH set.

Emergency locator transmitter

The aircraft was not fitted, nor was it required to be fitted, with an emergency locator transmitter.

Wreckage, flight path and impact information

Within the limitations created by the post-impact fire, no evidence was found that aircraft components or systems were factors in the accident. The landing gear was down, and the flaps were estimated to have been extended to 15 degrees at impact. A ground witness reported that the engines sounded as if they were normal and producing power immediately prior to impact. Post-accident inspection indicated that the engines should have been capable of normal operation and that they were producing power at impact.

Witness information and post-accident flight tests indicated that the aircraft probably tracked 135 degrees from the NDB for 3 km before commencing a 30-45 degree banked left turn towards the threshold of runway 28. The aircraft was left wing low and heading 360 degrees when it collided with trees during the turn. The aircraft had descended from 640 ft AMSL to 227 ft AMSL prior to impact.

Altimeters

The aircraft was equipped with two barometric altimeters which were severely burnt during the accident. It was not equipped, nor was it required to be equipped, with a radar altimeter. One altimeter retained a sub-scale setting of 1007 hectopascals. The other altimeter was too damaged for the altimeter setting to be ascertained. Forecast QNH for departure Moorabbin was 1007 hectopascals. On the AVFAX received by the pilot prior to departure, the forecast local QNH for his arrival at King Island was 1005. At 0415, Melbourne Control advised the pilot that area QNH was 1007. In contrast, the QNH recorded by the King Island automatic weather station, seven minutes before the accident, was 1003. This reading was passed to the Melbourne Control operator within one minute of its recording. Had the pilot requested an updated QNH from Melbourne Control shortly after 0500 he would have been given 1003.

The sub-scale setting on the altimeter being referenced by the pilot could not be determined. The company chief pilot reported that both altimeters in KIJ had been accurate to within plus or minus 20 ft during flights prior to the accident. The allowable instrument flight rules tolerance was plus or minus 60 ft with an accurate QNH set. A sub-scale error of +4 hectopascals (1007 instead of 1003) could result in the aircraft flying 120 ft lower than expected.

Medical information

The post-mortem and toxicology tests performed on the pilot revealed no medical problem which may have contributed to the accident.

Fire

No evidence was found of in-flight fire. There was a post-impact fire which destroyed most of the aircraft.

Survival aspects

The accident might have been survivable except for the post-impact fire.

Possible misinterpretation of visual circling criteria

No evidence was found to indicate that the pilot would have deliberately descended the aircraft below the circling minima prior to becoming visual.

After discussions with several very experienced instrument-rated pilots, approved testing officers, chief flying instructors, chief pilots and flying operations inspectors, it became apparent that many instrument-rated pilots continued to misinterpret the departure and approach procedures set out in Airservices Australia's Instrument Approach and Landing Charts, page 2, paragraph 1.5, and in particular note 1.

The following extract from Aeronautical Information Publications - Visual circling is relevant to the accident:

"When visual reference has been established within the circling area at or above the minimum descent altitude, further descent below the minimum descent altitude may occur provided that: a. the aircraft is maintained within the circling area; b. visual reference can be maintained; c. the approach threshold or approach lights or other markings identifiable with the approach end of the runway to be used are visible during the subsequent visual flight; and d. obstacle clearance of at least 300 ft (category B) is maintained along the flight path until the aircraft is aligned with the runway, strip or landing direction to be used.

"Note 1: For the purpose of this paragraph visual reference means clear of cloud, in sight of ground or water along the flight path, and with a flight visibility not less than the minimum specified for circling."

The aircraft crashed within the prescribed circling area. Whether the pilot maintained at least the 2.4-km minimum visibility specified for visual circling is unknown.

This investigation determined that many pilots misinterpret the words "in sight of ground or water" to mean no cloud is anticipated between the aircraft and the ground or water along the flight path, whether by day or night, despite not being able to see the ground, water or obstacles. Some pilots believe it acceptable to have 4/8 of cloud below the aircraft at night during a 300-ft obstacle clearance circling approach in very dark conditions. Other interpretations were also evident.

Research has indicated that many pilots in their training have been taught a wrong interpretation of the meaning of "in sight of ground or water along the flight path."

The authoritative interpretation from the Civil Aviation Safety Authority is that "in sight of ground or water along the flight path" means that pilots must be able to physically see the ground, water, and obstacles along the flight path before descending below the minimum descent altitude to apply a 300 ft obstacle clearance buffer.

Previous action on misinterpretations of visual circling criteria

In the Bureau's report 9301743 of the accident involving Piper PA-31-350 Chieftain VH-NDU at Young NSW on 11 June 1993, BASI interim recommendation IR930231 recommended that the Civil Aviation Authority review:

"(a) the adequacy of instructions to flight crew for maintaining a safe height above terrain at night, and

"(b) the phraseology used in aeronautical information publications, departure and approach procedures, instrument and approach and landing charts, paragraph 1.5 with a view to making it less susceptible to misinterpretation".

As a result of this recommendation the Civil Aviation Authority modified the words in the Aeronautical Information Publications to read as follows: "visual reference means clear of cloud, in sight of ground or water along the flight path".

From the same report BASI interim recommendation IR9300234 recommended that the Civil Aviation Authority review the obstacle terrain guidance information provided for flight crew in other than high-capacity regular public transport operations. This review was intended to ensure that flight crew have an adequate knowledge of terrain associated with the route flown, including obstacle terrain information for non-precision and circling approaches. The following is part of the Authority's response:

"The requirement to avoid obstacles by 300 feet is to be complied with using visual reference only, i.e. the pilot must be able to ensure all obstacles lit or unlit are avoided visually. At night this may not be possible. Thus the pilot may only be able to descend when he is aligned with the landing runway and able to use the documented obstacle limitation surface, and, the Civil Aviation Authority will review the practices of other authorities in respect to the provision of terrain information on instrument approach charts with a view to determining whether the current practices need to be changed".

BASI investigation report 9302851 of an accident involving Piper PA-31-350 Chieftain, VH-WGI, in Tasmania, was produced after report 9301743. Report 9302851 identified that the visual circling criteria continued to be misinterpreted despite the fact that the Civil Aviation Safety Authority had amended information on visual circling contained in Aeronautical Information Publications.

Night visual approaches

During a visual approach, a pilot relies on a combination of visual cues and instrument indications to judge the flight path, rate of descent and closure rate. However, clearance from the ground is maintained by visual reference only. During an instrument approach, ground clearance is maintained by reference to a set instrument flight path and an established minimum altitude until the ground is in sight.

When conducting night visual approaches where there are overcast conditions, low levels of ambient light and no visual cues on the ground prior to the runway lights, pilots are less able to adequately judge rates of descent and closure rates appropriately. In many previous accident investigations and research studies using simulators, these "black hole" conditions have been associated with pilots flying low approaches and impacting the ground before the runway threshold. However, the same conditions can also produce high approaches. The black hole conditions in this accident were also exacerbated by the lack of a defined horizon or other visual cue information on the ground beyond the runway lights.

Previous Bureau investigations of night take-off and landing accidents have determined that pilots often have difficulty accepting that operating conditions do not meet the requirements for visual flight, particularly when they can see the runway or helipad lights and there appears to be adequate visibility. However, in black hole conditions a pilot must revert to instrument approach procedures to ensure an adequate level of safety. At many aerodromes this is not an option as there is no precise approach aid. Consequently, pilots often persist with a visual approach despite the conditions.

Automatic weather stations

An automatic weather station was installed at King Island. This provided information electronically to the air traffic service operators in Melbourne. This information was available to the pilot on request.

Automatic weather information broadcast facilities have been established at a number of airports throughout Australia. More automatic weather information broadcast facility installations are planned. The automatic weather information equipment, which is linked with the automatic weather station, normally transmits weather information on the airport navigation aid frequency. Information from Airservices Australia suggests that the locations of automatic weather information broadcast facilities are determined by the aviation industry and the Bureau of Meteorology.

King Island is currently serviced by four airlines and a number of charter operators. At present Airservices Australia has no plans for an automatic weather information broadcast facility to be installed at King Island.

Statistics of similar accidents

During the investigation, a search of Bureau of Air Safety Investigation and National Transportation Safety Board (USA) records for similar occurrences was completed.

Recent Australian accidents identified were:

• BASI investigation report 8802354, Piper PA-31-350 Chieftain, VH-HOX, Coffs Harbour NSW, 7 April 1988; fatal accident at night within the circling area in marginal weather.
• BASI investigation report 9301743, Piper PA-31-50, VH-NDU, Young NSW, 11 June 1993; fatal accident at night within the circling area in marginal weather.
• BASI investigation report 9302851, Piper PA-31-350 Chieftain, VH-WGI, Tasmania, 17 September 1993; fatal accident at night within the circling area in marginal weather.

A computer search of US accident records since 1991 disclosed 17 accidents in which aircraft flew into the ground within the circling area in dark-night conditions. Most of these accidents involved instrument flight rules flights and marginal weather.

ANALYSIS

Accident location

The accident site and aircraft configuration were consistent with the aircraft being on a left base turn for runway 28. Wind conditions were suitable for a landing on runway 10 or 28. Had the pilot intended to land on runway 10, there was no reason for the aircraft to fly so far south-east of the NDB before turning towards the north.

Obstacle clearance

After 104 flights to King Island, the pilot was probably confident that he could avoid the obstacles within the circling area. However, it was possible that the pilot was unaware that the terrain/obstacles to the south-east were about 100 ft higher than the aerodrome reference elevation.

Final flight path

The evidence provided about the misinterpretation of the visual circling criteria may be relevant to the pilot's actions. Information from the witness who reported seeing the aircraft's lights indicates the aircraft had probably descended below the cloud base prior to or during the final turn. Despite this, the reported weather conditions of a dark and misty night with no defined horizon and no ambient lighting made it unlikely that the pilot could maintain visual contact with the ground or obstacles as required by the circling criteria. He would, however, have been able to see the runway lights. If the pilot was not aware of the correct meaning of the criteria, he probably assumed that having only the runway lights in sight was acceptable. Consequently, he continued a visual approach in conditions that were not suitable for visual flight.

Accident investigation and anecdotal evidence indicate that pilots are reluctant to accept that conditions do not always meet visual requirements when they can see the runway or helipad lights during an approach, especially in marginal weather conditions on a dark night. This lack of acceptance often leads to misjudgement of the aircraft's performance during the approach which, in turn, can lead to unexpected ground impact.

Because the pilot did not request updated QNH information, it is probable that he did not have the King Island QNH of 1003 hectopascals set. This would have resulted in the aircraft being closer to the obstacles than the pilot realised.

As the pilot had few visual cues during the turn, he would have relied on the altimeter and vertical speed indicator to help judge his descent rate. At the same time, he would have needed to check the aircraft's position in relation to the runway by reference to the runway lights. This regular transition between visual and instrument flight, in what were instrument flying conditions, probably distracted the pilot to the point where he was unaware of the proximity of the trees until it was too late to prevent the impact.

The evidence indicates that it is unlikely that the pilot had encountered a similar combination of conditions in either his flying or training experience.

The combination of a misunderstanding of the circling criteria, lack of adequate visual conditions, lack of recognition that adequate visual conditions did not exist, the black-hole effect, incorrect QNH, distraction and lack of experience probably caused the pilot to misjudge the descent rate during the approach to runway 28.

The Bureau believes that AIP/DAPS IAL-2, 1.5 should be amended to clarify the instruction and differentiate between day and night circling approaches due to the apparent misunderstandings of the circling requirement. The difficulties of maintaining visual reference with the ground or water at night should also be considered. Clear visual sighting of ground obstacles is almost impossible at night. Therefore, a different limitation may be appropriate.

The Bureau is also concerned that the obstacle clearance heights of 300 ft for category A and B aircraft, and 400 ft for category C and D aircraft, leave very little margin for height deviation. During the high workload of circling approaches, as the aircraft configuration changes with flap and landing gear extension, an inadvertent height loss of 200-300 ft could occur. Pilots are not required to demonstrate a visual circling approach at minimum obstacle clearance height as part of the initial instrument rating test and renewals.

SIGNIFICANT FACTORS

1. The pilot continued a visual approach in conditions which prevented him from maintaining adequate visual clearance from the ground or obstacles and which made visual judgement of the approach difficult.
2. The pilot probably did not recognise that the conditions were not suitable for a visual approach.

SAFETY ACTION

As a result of the investigation, the Bureau of Air safety Investigation issued interim recommendation IR960027 to the Civil Aviation safety authority on 28 August 1996.

"IR960027

"The Bureau of Air Safety Investigation recommends that the Civil Aviation Safety Authority:

"(i) amend AIP/DAPS IAL-2, 1.5 to clarify the intent of the instruction and differentiate between visual circling approaches conducted during the day and at night;

"(ii) critically review the obstacle clearance height to assess whether these minimum heights are appropriate from an operational viewpoint; and

"(iii) require pilots to demonstrate a visual circling approach at the minimum obstacle clearance height during the test for their instrument rating and at subsequent renewals."

The Civil Aviation Safety Authority responded on 12 November 1996 as follows:

"I refer to your interim recommendation IR960027 concerning the accident involving PA31-350, VH-KIJ at King Island on 8 February 1996.

"(i) AIP amendment List 17, effective 5 December 1996, includes a revision of DAPS IAL 2, para 1.5. The revision clarifies requirements for visual circling which are: applicable by night or day, and applicable only during daylight.

"An AIP SUP addressing the AIP changes has been produced with an effective date of 7 November 1996.

"(ii) 300 feet minimum obstacle clearance is the internationally accepted requirement for visual circling for

Category A and B performance aircraft as published in ICAO PANS OPS Doc 8168. The Civil Aviation Safety Authority does not have any evidence to indicate that a trained and recent pilot who is attempting to comply with the obstacle clearance requirements specified for visual circling is placing his or her aircraft in an unsafe situation. The revised text of AIP DAP IAL 2 para 1.5 will further promote safety in this regard by clarifying the requirements for visual circling at night and advising pilots to maintain the maximum practical obstacle clearance during visual circling.

"(iii) Appendix 1 of CAO 40.2.1 specifies the flight test requirements for the initial issue and renewal of instrument ratings. Para 2.1 (f) of the Appendix specifies that the applicant shall demonstrate proficiency conducting a circling approach. However, the present flight test form for the conduct of an instrument rating test does not list a circling approach as a test item. This shortcoming in the flight test form is being addressed in the course of a review of all flight test forms commenced several months ago. It is anticipated that a new instrument rating flight test form will be promulgated and distributed to industry early in 1997."

The Civil Aviation Safety Authority responded again on 13 January 1997 as follows:

"I refer to BASI draft Air Safety Occurrence Report 9600399 concerning the accident involving Piper PA31-350, VH-KIJ, near Moorabbin Victoria on 8 February 1996. The following comments are forwarded for your consideration.

"The Authority agrees with the recommendations contained in the subject draft ASOR. Also, in regards to IR960027:

"(1) AIP DAPS IAL 2 paragraph 1.5 has been amended, with effect 5 December 1996, to clarify the intent of the instruction. The revised text provides requirements for visual circling at night and by day.

"(2) The obstacle clearance heights required for visual circling are those included in the procedures of ICAO Doc 8168 OPS/611 Volume 1 which have been adopted by Australia. These procedures are accepted and are used worldwide. The Civil Aviation Safety Authority considers that the minimum obstacle clearance heights specified in the PANS OPS procedures provide an appropriate safety margin for visual circling operations."

Classification of response: OPEN (The Bureau considers that the response does not meet some or all of the criteria for acceptability for a recommendation that the Bureau considers to be significant for safety. The Bureau will initiate further correspondence.)

The Bureau issued interim recommendation IR960054 to Airservices Australia on 27 August 1996 as follows:

"IR960054

"The Bureau of Air Safety Investigation recommends that Airservices Australia review the criteria used for the installation of AWIBs, taking into account the types of operations at the airport, the frequency of RPT operations, the geographic location and prevailing meteorological conditions."

Airservices Australia responded on 15 November 1996 as follows:

"Re: Occurrence 9600399 generating Interim Recommendation: IR960054

"The Bureau of Meteorology has been installing Automatic Weather Stations (AWS) at aerodromes around Australia for a number of years. AWS are now a vital component of the weather observation network and make a significant contribution to aviation weather products, in addition to other services provided by the Bureau.

"AWS transmit to both the Bureau and to Airservices Australia the basic elements of wind direction and speed, pressure (QNH), air temperature, dew point, relative humidity and ten-minute rainfall. The AWS observations are distributed to Air Traffic Service units and are also stored in the AIS/MET database. In essence, the AWS observations form part of the preflight and inflight information service and, as such, are available on request.

"At the 1994 consultative meeting, industry endorsed a Bureau proposal to make AWS information available via telephone; this being facilitated by means of a Bureau developed device (known as Aerodrome Weather Information Broadcast [AWIB]) connected to each AWS. Industry also endorsed a proposal to make the AWIB information available on navigation aids which were collocated with Bureau AWS.

"With respect to installation and priorities, industry endorsed the proposition that any installation to support telephone (Bureau) or navigation aid (Airservices) access would form part of the normal equipment (AWS and navigation aid) maintenance programmes of the respective organisations. This endorsement was based on the understanding of AWIB connection issues at that time. It is on this basis that the Bureau is progressively implementing a national AWIB installation programme, providing industry with access to the broadcast information via telephone. The Bureau is currently planning to introduce some 20 AWIB per year. However, this installation programme is rather flexible as it depends largely on Bureau Regional Office priorities as to when and where it is carried out. Funding of the Bureau's installations was an integral part of the proposal endorsed by industry.

"Airservices Australia's involvement with such facilities to date has been limited to:

"(a) permitting AWIB to be connected to the Mount Gambier VOR for the initial proof-of-concept trial;

"(b) the use of AWIB (connected to a local navigation aid) as a replacement for obsolete ATIS facilities at four other non-towered locations; and

"(c) introducing AWIB at Canberra as an out-of-hours ATIS supplement which eliminates the need for costly ATC support of out-of-hours RAAF flights, while at the same time directly assists industry to meet Canberra's new Noise Abatement Procedures (ATIS ZULU retained on NDB, AWIB on VOR).

"The only criteria applied in the selection of these sites was that of cost benefit to Airservices. Each of these installations was progressed as an individual requirement and not as part of any programme.

"Whereas the Bureau's costs were not high, were readily defined and were therefore endorsed as a part of the proposal, Airservices is still in the process of identifying the cost of AWIB/navigation aid connection at other locations. It is intended that this costing information form part of a proposal to industry seeking funding for an installation programme. As you might appreciate, the cost of these connections varies considerably from site to site, and as they may involve several kilometres of new cabling work can be quite significant. Airservices' technical staff estimate the average cost of new cabling works at over $23,000 per kilometre.

"The other component of this Airservices proposal will be a prioritised installation schedule. The priorities are expected to be derived from work currently being undertaken by a group (chaired by the Bureau) which is determining a programme for AWS upgrades/enhancements. The remainder of this group is made up of representatives from Airservices, CASA, industry, and the aerodrome owners. The criteria being used by the group specifically includes consideration of the rate of aerodrome utilisation by RPT and other IFR aircraft, critical weather locations and the availability of alternates.

"In the meantime, Airservices will continue to make AWS-derived observations available as part of the preflight and inflight information services."

Classification of Response: CLOSED - ACCEPTED

The aircraft departed Wee Waa at about 1755 EST with a load of 1450 litres of Endosulphan and Delphin spray mixture, to spray a cotton property 5 km east of Walgett. At about 1900 the aircraft was noticed by a witness in the adjoining property making a spray run from the north towards south. At the end of the run the aircraft pulled up and commenced a turn initially to the right then reversed the turn to the left. The left turn continued and the aircraft flew into the ground. The aircraft appeared to be operating normally up until the accident.

This accident was not subject to on-site investigation.

FACTUAL INFORMATION

The aircraft had been hired in Victoria by the pilot for an extended private business trip to south-east Queensland.

The aircraft disappeared on a NOSAR (no search and rescue watch) flight from Caloundra to Kooralbyn via Jacobs Well. The aircraft had not been reported missing and there was no evidence that the pilot had made any arrangements for a private SARWATCH. The pilot was not rated for flight in instrument flying conditions.

On 8 January 1996, a body, later identified as that of a passenger in the aircraft, was washed up on a beach near Caloundra and some aircraft wreckage was found on a beach at the south-eastern end of Bribie Island. The pilot's body was found in the same area the following day. Later, a member of the public handed to police additional pieces of wreckage which he had found in the area on 5 January.

Recorded data from the Brisbane Terminal Area radar (TAR) revealed that the aircraft tracked along the Bribie Island coast to a point three kilometres north-north-west of Woorim near the ocean beach. It disappeared from radar at the completion of a 180-degree left turn at 1456 EST. The aircraft was outside controlled airspace at the time and its transponder was not operating. The Bureau of Meteorology weather radar information showed a large rain squall at the southern end of Bribie Island between 1440 and 1520.  Correlation with the TAR-recorded information showed that the aircraft entered the area of the rain squall. Another pilot, who was conducting a scenic flight near the Glass House Mountains, confirmed that the southern end of Bribie Island was obscured by a heavy rain squall at the time of the disappearance.

The aircraft was fitted with an ELT which complied with TSO C91. No reports were received that indicated that the ELT operated during the accident sequence.

The aircraft has not been recovered.

ANALYSIS

Radar plots showed that the aircraft entered a heavy rain squall. The 180-degree turn may have been an attempt by the pilot to regain visual flight, but during the turn, control of the aircraft was probably lost.

SIGNIFICANT FACTORS

1. The pilot did not hold an instrument rating.
2. The aircraft entered a rain squall.

At a result of the investigation into this occurrence, the Bureau of Air Safety Investigation forwarded the following interim recommendation to the Civil Aviation Safety Authority on 4 November 1996:

IR960127

The Bureau of Air Safety Investigation recommends that the Civil Aviation Safety Authority:

1. review the Aviation Safety Surveillance Program to ensure that new commercial operators are adequately monitored and inspected until a demonstrated history of safe operation is known;
2. align the scheduled surveillance period of the Aviation Safety Surveillance Program to that of the validity period of the air operators certificate;
3. reconsider the flight review requirements for Chief Pilots with the view of bringing them into line with the current situation for Chief Flying Instructors, as an additional method of surveillance;
4. review the adequacy of the approval and assessment requirements for Chief Pilots who do not have a demonstrated history in flight operations with a commercial operator;
5. review the current situation regarding Aeroplane Flight Reviews, to allow for appropriate notification to the Civil Aviation Safety Authority and recording of the results.'

FACTUAL INFORMATION

Sequence of events

The occupants of the Piper Lance aircraft had planned an overnight stop at a farmhouse near the proposed landing area. The pilot in command spoke to his contact (another pilot) at Boddington by telephone at least three times on matters related to the landing area. Landing area details were discussed in depth and the contact indicated that the pilot in command should land towards the east in the paddock he recommended. It was also recommended that he complete a right circuit at 1,800 ft above mean sea level (1,000 ft above ground level) to remain clear of the surrounding terrain. Immediately prior to departure the pilot in command was told that the wind at the landing area was a north-westerly at 15 kts. During one conversation the pilot in command indicated that he would land at Narrogin if the landing area at Boddington was unacceptable.

The aircraft circled Boddington for 15 minutes before it made an apparent approach to land. During this time, it made a low pass alongside the proposed landing area.

The pilot in command did not follow instructions. He completed a left circuit at 500 ft above ground level and approached from the north-west at 45 degrees to the proposed landing direction. The aircraft descended to within 3 m of the ground, with landing gear and full flap extended. Power was then applied and a go-around commenced. The aircraft turned slightly left, passed between trees and continued to climb over rising ground. It was observed to veer towards the right during the climb.

Evidence indicates that the landing gear and flap remained extended during the climb.  The aircraft collided with a tree 900 m from the go-around point. The collision occurred 25 m above the ground and 3 m below the top of the tree. The aircraft was extensively damaged by the collision. It then crashed in a dam 50 m beyond the tree and to the right of the original flight path.

Damage to the aircraft

Damage to the aircraft indicated that its flying characteristics were adversely affected by the tree collision. Tree impact had severely damaged both wings and destroyed the control system located in the aircraft's lower fuselage. The pilot in command could not have exercised any control after that point.

Weight and balance

The aircraft's weight at the time of the accident was estimated at 1450 kg. Maximum weight was 1633 kg.

Personnel information

Despite the pilot in command's considerable flying experience, information provided by his associates indicates that he had always flown into and out of landing areas that were marked as flight strips. As a result, the pilot in command had not previously been confronted with the complexities of assessing whether a farm paddock met terrain clearance and aircraft performance requirements.

Post-mortem examinations did not disclose any medical condition that might have contributed to the accident.

Meteorological information

The weather conditions were recorded as fine, temperature 28 degrees Celsius with a wind from the north-west at approximately 8 kts.

Proposed landing area information

The paddock chosen for the landing was aligned east-west across a valley and included a grassed area 1,200 m long and 100 m wide. There were no other suitable paddocks in the immediate area. The grassed area was clear of obstructions but was not marked out as a landing strip. The proposed landing area sloped up from a river, towards the east and was contained within the grassed area. The slope was two degrees (3.4%) for the first 900 m, increasing to four degrees (6.8%) at the eastern end. There were ridges, 35-45 m higher than the surrounding terrain, at each end of the landing area. A displaced threshold, to assist the pilot in command during his approach over the high ground and trees, was marked by a car parked 300 m in from the western edge of the paddock. The car was also parked facing into wind, as arranged with the pilot in command, to indicate the wind direction. It was expected by the ground party that the pilot in command would make an approach for a landing towards the east (uphill). This direction meant the aircraft would land downwind.

A post-accident inspection of the paddock and an assessment using the approved landing weight chart indicates the proposed landing area did not meet the specifications contained in Civil Aviation Advisory Publication No. 92-1(1), Guidelines for Aeroplane Landing Areas. The average longitudinal slope was 5%. This exceeded the maximum of 2%. The landing distance available, after object clearance requirements were considered, was 530 m. This was 190 m less that the minimum distance calculated from the landing weight chart for the prevailing conditions. The pilot in command did not attempt to use the proposed landing area although it had been recommended to him by another pilot.

The approach direction used by the pilot in command was along the valley rather than across it. This allowed the pilot in command to fly a shallower approach than would have been the case had he used the direction recommended. It also provided a better climb-out route in the event of a go-around. However, fences and a dry watercourse reduced the actual landing distance available in this direction to 300 m, much less than the distance required.

Wreckage information

The wreckage was examined at the accident site and after removal to storage. The inspection of the airframe did not disclose any defects that may have contributed to the accident sequence.

Witnesses reported that the engine sounded normal during the climb-out after the go-around. Inspection of the engine indicated it was capable of normal operation and was developing power at the time of impact with the water.

An anomaly was found with the rear wing spar attachment points that might have affected the structural integrity of the aircraft. Post-accident inspection of the wreckage disclosed that the rear wing spar attachment points had been modified some time prior to the accident. The attachments had been cut to turn bolt holes into slots. The aircraft manufacturer has not approved this type of modification. There were indications that one attachment had been loose prior to the accident. No records were found relating to the modification. Discussion with one of the aircraft's owners indicated that the modification had not affected the aircraft's performance. It compared favourably with other PA32s also flown by the owner of the accident aircraft.

Aircraft performance

Performance calculations based on flight manual and manufacturer's information, indicate that the aircraft should have been capable of climbing clear of the terrain and trees following the go-around with the landing gear and flaps extended.

ANALYSIS

Proposed landing area

The recommendation by the ground contact and the acceptance by the pilot in command that the proposed landing site was suitable, indicates that neither had an adequate understanding of the parameters for an acceptable aeroplane landing area as set out in GAAP 92-1(1).

The pilot in command did not attempt to land in the paddock recommended by his ground contact at Boddington. This was probably because a later assessment from the air indicated that it was unsuitable.

The investigation could not determine why the pilot in command was not aware of the landing distance limitation on his actual approach path. It is possible that the obstructions did not become apparent until late in the approach as the fences and the dried watercourse would have blended into the surrounding dry grass. The attempted go-around probably indicates that the pilot in command realised the area was unsuitable during the latter part of the approach.

Go-around and climb performance

During the go-around the pilot in command did not retract either the flaps or the landing gear. The investigation could not determine the reason for this.

Performance calculations indicate that the aircraft should have been capable of climbing clear of the terrain and trees following the go-around. Why it did not could not be determined. Nor could it be determined why the pilot in command did not turn the aircraft further right, away from the trees and towards lower ground. The Piper Lance has an extended forward fuselage, restricting forward and downward visibility during a climb. It is possible that the pilot in command was not aware of the aircraft's proximity to the tree and that the collision was completely unexpected.

Wing spar attachment modification

It was not possible to determine when or why the holes in the wing rear spar attachments had been modified. It could not be determined what effect the modification and/or the loose wing spar attachment might have had on the structural integrity of the aircraft, particularly during the collision.

SIGNIFICANT FACTORS

1. A landing was planned and attempted in an area where there were no suitable landing sites.
2. The aircraft's climb performance was less than it should have been during the climb-out from the go-around.

SAFETY ACTION

As a result of the investigation into this occurrence, the Bureau of Air Safety Investigation advised the Civil Aviation Safety Authority of the details of the wing spar attachment modification.