An approved training pilot was conducting mustering endorsement training with another company pilot. At the time of the accident they were practising autorotational approaches with power termination prior to touchdown.

The training pilot had demonstrated a number of autorotational approaches from low and medium altitudes. He had also followed the trainee pilot through on the controls for a number of similar approaches. The trainee had exhibited problems with airspeed and rotor rpm control.

During the final practice, with the training pilot again following through on the controls, the trainee misjudged the approach. The training pilot decided to terminate the approach at 50 - 70 feet above ground level and with 30 - 35 knots forward speed. Although recovery action was taken the aircraft continued to descend, touching down heavily before the action had any significant effect on the rate of descent. The aircraft slid for approximately 30 feet before overturning.

It is possible that the training pilot was distracted by his instructional duties, and he misjudged the height required to recover from the autorotational descent.

While approaching 6000 feet on descent into Canberra the aircraft entered an area of severe turbulence. The turbulence had not been forecast and lasted for about one minute.

Prior to penetrating the turbulence, the flight attendants had prepared the aircraft for landing, but not all had taken their seats. Two of the non-seated flight attendants were thrown about the cabin but were not injured. The aircraft was climbed to 8000 feet for a smoother ride to allow all on board to be seated.

One seated flight attendant had the shoulder harness pull through the bulkhead fitting. It was suspected that the harness had been incorrectly routed through the fitting. A fleet check was initiated to inspect all seat harnesses for correct installation. No further examples of incorrect installation were found.

HISTORY OF THE FLIGHT

VH-EPJ was engaged in an instrument flight rules charter flight, with two passengers, from Telfer to Port Hedland. The flight had departed Port Hedland at 0650 that morning, arriving at Telfer at 0759. The aircraft remained at Telfer until 1526 after which, it departed for Port Hedland.

At a position 138 km south-east of Port Hedland, whilst cruising at 6,000 ft, the pilot observed that the left fuel quantity gauge was indicating empty. He selected the left fuel selector to 'crossfeed' and turned on the left fuel boost pump. After checking his fuel calculations, the pilot elected to continue the flight to Port Hedland. Within two minutes the left engine began surging. This was followed by surging in the right engine. The pilot completed the basic engine failure drills; however, due to his concern over control difficulties resulting from the yawing caused by the surges, he elected to shut down both engines, feather the propellers, and conduct an emergency landing. The aircraft landed without further incident on the Port Hedland to Marble Bar road.

After obtaining fuel for the aircraft the pilot flew it to Port Hedland.

PERSONNEL INFORMATION

The pilot of VH-EPJ, at the time of the occurrence, was also the Chief Pilot (CP) of the operating company. The CP was the only pilot to operate VH-EPJ after 29 June 1994. The pilot had completed his Beechcraft Baron endorsement in a BE55 which has a different fuel tank configuration to the BE58. During this endorsement the determination of fuel contents using the filling point tabs was discussed. Because the fuel tank arrangement in Baron aircraft does not lend itself to easy determination of the fuel contents, other than through the gauging system, it was recommended to the pilot that he always keep an accurate manual fuel log when operating the aircraft. Although the presence of wing mounted fuel gauges, in the BE58 was mentioned, the specifics of fuel management in the BE58 were not covered. This was left for the pilot to pursue. An endorsement in a BE55 qualifies a pilot to act as pilot-in-command of a BE58.

AIRCRAFT INFORMATION

VH-EPJ is fitted with rubber bladder type wing fuel tanks. An inspection of these tanks indicated that they had not collapsed at the time of the occurrence. The aircraft owner advised that the fuel gauging system had been calibrated twelve months prior to the occurrence and that the average normal fuel consumption for VH-EPJ was 113-115 L/hour.

Wing-mounted fuel quantity gauges provided the potential to indicate fuel quantities between 40 and 60 US gallons (151 L - 227 L) in each wing tank. The fuel tanks were fitted with tabs which had the potential to give a fuel indication of approximately 302 L in each tank.

The aircraft was not fitted with a placard or decal which indicated, to a pilot, what the tab indication meant. Nor was the information included in either the aircraft flight manual or company operations manual. The left and right exhaust gas temperature gauges (EGT), used to assist in accurately leaning the engine fuel mixture in flight, were unserviceable. When power was applied one gauge went immediately to a maximum reading and the other remained at zero. The EGT unserviceability was not entered in the aircraft's maintenance release document.

No evidence (fuel staining) was found which would indicate that fuel had vented from the aircraft in flight. The aircraft performed normally during the ferry flight to Port Hedland following the occurrence.

COMPANY DOCUMENTATION

The company operations manual contains the following requirements:

• Section A:3:6:

'Fuel Usage Records. The aircraft flight time and fuel log sheet includes a section which enables a continuous check of the fuel usage and fuel remaining at the completion of each trip. The pilot in command shall complete this section at the completion of each trip. Fuel used by each aircraft shall be calculated on a monthly basis by the Chief Pilot and fuel consumption determined'.

• Section A:4:6:

'The Pilot in Command is to ensure that sufficient fuel is carried to proceed from the departure aerodrome to the destination aerodrome or alternate aerodrome if required...All company aircraft shall carry a fixed reserve of 45 minutes at the holding rate and a variable reserve of 15% of the flight fuel at the cruise consumption rate. The flight fuel is to be calculated at the consumption rate for each aircraft shown in section B:1:1'.

• Section B:1:1:

'Flight Planning. Use greater of A) 180 knots at 110 litres per hour from departure to destination plus alternate if required, plus 15%, plus 45% (79 litres) or B) single engine use 125 knots 60 litres per hour from critical point to destination plus 15%, plus 10 minutes (18 litres). Taxi Fuel: 10 litres' [plus 45% should read 45 minutes - typographical error].

The operations manual did not contain any information on the determination of fuel contents other than by refuelling the aircraft to full tanks and/or using a manual fuel logging system. Nor did it contain information which would indicate to the pilot that other conditions, such as extended climb time or low altitude operations would affect the fuel burn.

PILOT OBSERVATIONS

The pilot indicated that he was not aware of any method of determining the fuel contents in a BE58, other than by reference to the main fuel gauges, except for refuelling the aircraft to full tanks and maintaining an accurate fuel log from then onwards. The aircraft had been refuelled, for the occurrence flight, at Port Hedland and the pilot had completed fuel drain checks for water at both Port Hedland and Telfer. During the flight from Telfer, he had checked the fuel indications each 10 minutes and he did not notice anything abnormal until immediately before the surging started. At that point the pilot noticed that the left fuel gauge was indicating empty and the right gauge 1/4. The right gauge had dropped rapidly to about 1/8 when the right engine also started surging.

The pilot initially thought there was a problem with the contents of the left tank only and that he had sufficient fuel in the right tank to continue to Port Hedland. He was aware that he would arrive at Port Hedland (a further 27 minutes flying) with little or no fuel remaining. His initial decision was to continue with the flight. When the right engine started surging, he knew the problem was one of exhaustion rather than starvation and elected to complete a precautionary landing.

As the EGT gauges were unserviceable the pilot relied on the fuel flow gauges to lean the fuel mixture to the cruise setting.

The pilot also indicated that he was concerned that the aircraft could be damaged if it was left parked on the road for any length of time. He did not make an attempt to contact a licensed aircraft engineer to inspect the aircraft before proceeding to Port Hedland as he knew it was unlikely that one would be available at short notice.

FUEL CALCULATIONS

The pilot had determined that 430 L of fuel was onboard the aircraft at Port Hedland, prior to departure, by reference to the company fuel remaining record, for which, as the only pilot of the aircraft since the 100 hourly inspection on 2 June 1994, he had been solely responsible. The record indicated that the fuel tanks had been last filled to capacity on 17 July 1994, and that the aircraft had subsequently flown some 38 hours.

The pilot had maintained the fuel record from that time by:

• including fuel added to the aircraft (using known amounts from drums or refuelling contractors) and
• by deducting fuel used (calculated using 110 L/hour).

The pilot used 110 L/hour for flight times recorded by the flight time meter, plus an allowance for fuel usage during taxi of between zero and 15 L, depending on the actual taxi time. He had determined the 110 L/hour usage rate as a result of a flight to and from Perth in July 1994 which had permitted the use of full fuel tanks. The operations manual also indicated that 110 L/hour should be used for flight planning. Subsequent monthly checks of fuel consumption rates, as required by the company operations manual, had not been completed.

The pilot indicated that he mistrusted the fuel quantity indications provided by both the main fuel indicating system and the wing mounted fuel gauges as their readings did not correspond with the contents indicated by his fuel log. Although he believed the gauging system to be inaccurate, the pilot did not enter it as an unserviceable item in the aircraft's maintenance release. As a result of the pilot's mistrust of the aircraft fuel quantity gauges, the company fuel remaining record was the sole source of fuel quantity information used by the pilot.

During the investigation, the pilot reported that following the incident he had detected an error in the fuel remaining record. An entry on 4th October 1994 indicated a fuel added amount of 200 L when the actual amount had only been 100 L. The pilot indicated that he had ordered, and believed he had received, 100 L per side. However, the fuel docket issued by the refuelling agent indicated a total of 100 L.

CAA SURVEILLANCE

The pilot was interviewed by the CAA on 16 June 1994 and granted CP approval on 20 June 1994. The CAA's Aviation Safety Surveillance Program calls for one operational surveillance visit per year on an operator such as the one involved in this occurrence. No operational surveillance had been carried out since the appointment of the new CP. A CP is appointed only after they have satisfied the authorised CAA officer, at interview, that they have the qualifications, knowledge, and experience to perform the duties set out in CAO 82.0. The authorised CAA officer advised that the CP, in this case, had satisfied the CAO requirements.

ANALYSIS

Cause of the Surging

Possible explanations for the surging were:

1. A non-fuel-related problem in the engines

- This possibility was considered very unlikely as both engines were similarly affected at the same time.

2. Fuel quality problems

- This possibility was considered very unlikely as the previous refuel had been conducted at Port Hedland using high standard refuelling equipment. Further, the pilot had completed fuel drain checks at both Port Hedland and Telfer with negative results.

3. Fuel starvation due to a mechanical defect in the aircraft (such as collapse of the fuel tank bladder) or the fuel system, resulting in interruption of the delivery of fuel to the engine, despite adequate fuel remaining on board the aircraft.

- This possibility was considered unlikely in view of the subsequent successful flight from the incident site to Port Hedland and an inspection of the fuel system did not disclose a collapsed bladder or any other problem.

4. Fuel exhaustion.

- This is considered the most likely cause and was the focus of the investigation. The following possible reasons were considered.

(a) There were faulty fuel drain valves, or the fuel drain valves were not fully closed.

- There was no evidence of fuel leakage around the fuel drain valves to support this possibility.

(b) There was a theft of fuel.

- There was no evidence to support this possibility.

(c) The pilot had miscalculated the fuel contents.

Miscalculation of fuel contents

An analysis of the flight suggests that the likely fuel on departure from Port Hedland, assuming a usage rate of 113 L/hour (based on information provided by the aircraft owner), 10 L taxi allowance and fuel exhaustion, was 226 L.

total of 130 L and Telfer to the incident site was 40 minutes @ 113 L/hour for a sub-total of 76 L) and adding the taxi fuel (two periods of taxiing @ 10 L each for a sub-total of 20 L).

As the calculated amount of fuel was 226 L, a discrepancy of 204 L existed with the 430 L calculated on the pilot’s fuel remaining record. Whilst 100 L was accounted for earlier in the report, a discrepancy of 104 L remained. This amount may be attributable to an actual fuel flow rate in excess of 110 L/hour. Over the 38 flight hours since the last positive check of fuel on board, a rate of 113 L/hour would account for the discrepancy.

A difference of 3 L/hour is not considered significant, on any single flight if the pilot is aware of the actual fuel state on departure, as it would be adequately covered by the variable (15%) and fixed reserve (45 minutes) that must be carried as part of the operations manual requirements. However, such a small difference can be significant over a longer period of time as was the case leading up to this occurrence. Had the pilot, in his role as CP, carried out the monthly calculation of fuel usage, as required by the operations manual, he probably would have been aware of the higher-than-expected fuel usage rate and made allowances for it.

SIGNIFICANT FACTORS

1. The pilot was not sufficiently aware of all the methods for determining the actual fuel contents.
2. The use of 110 L/hour for both fuel planning and actual consumption was insufficient to cover all types of operations in the Beechcraft Model 58.
3. As CP, the pilot did not calculate the actual fuel consumption on a monthly basis therefore he deprived himself of more accurate fuel consumption figures on which to base his fuel log.
4. The pilot believed, probably incorrectly, that the fuel indicating systems were inaccurate and as a result he did not make sufficient use of them in his fuel calculations.
5. When convinced that the fuel indicating system was too inaccurate to use, the pilot did not place the system unserviceable. This deprived him of another method of determining the fuel contents.
6. Both exhaust gas temperature gauges were unserviceable, and the pilot did not attempt to have them repaired. This deprived him of accurate fuel mixture control which, in turn, may have led to a higher-than-expected fuel consumption.
7. The pilot made an error in one entry in his fuel log which created the impression that the aircraft contained 100 L more fuel than it actually did.
8. The operator's documentation either did not provide sufficient fuel management guidance to the pilot, or the guidance was incorrect.

While cruising at 2500 feet in visual conditions, the propeller suddenly detached from the engine. Because his radio had not been working well, the pilot quickly made an emergency call from his mobile telephone to the police emergency number. He then concentrated on making a forced landing onto a country road. During the landing roll, the aircraft hit a dirt bank which resulted in substantial damage. An inspection of the aircraft after the accident revealed a broken crankshaft which was the reason the propeller had detached from the engine.

The pilot had been cleared to land on runway 36 which had a light crosswind blowing from the left. A normal approach with the left wing being kept slightly low to counter the wind was made with the pilot intending to touch down on the runway threshold.

An approach speed of about 75 kts was maintained to touchdown, with the aircraft contacting the runway heavily causing all three landing gear legs to collapse. The aircraft then slid to a stop on the runway.

FACTUAL INFORMATION

Sequence of events

The balloon, with the pilot and six passengers on board, was launched following a delay of about 30 minutes due to the late arrival of a passenger, and the need to change to an alternative launch site. It was the last of three balloons in a loose formation travelling to the north-east under the influence of a weak south-westerly breeze. The initial cruising height was about 1,000 ft above ground level (AGL). Soon after passing to the west of the central business district, the balloon descended to a height estimated by witnesses to have been between 100 and 200 ft. During this phase of the flight the passengers, by raising their voices, were able to talk with people on the ground.

The pilot then climbed the balloon to about 500 ft before descending to an estimated 50 to 100 ft AGL. The flight continued at low level for approximately 3 km before it became apparent that the balloon was on a collision course with a house. The pilot and passengers reported that the balloon would not rise despite extensive use of the burners. The balloon basket struck the wall of the house and was dragged over the roof.

Two people who observed the balloon, one before impact and the other after impact, reported that they had seen an open panel on the side of the balloon.

The balloon continued and collided with the roof of a second house one street to the north-east, approximately 150 m from the first collision. This impact penetrated the roof structure of the house. The basket remained on the roof for about a minute, during which the pilot continued to use the burners in an attempt to obtain lift. However, the skirt of the balloon caught fire. The basket then broke free from the roof, demolished a section of brick fence, and the balloon envelope became entangled in a roadside powerline.

Having used the burners for several minutes, the pilot freed the balloon from the powerline and climbed to about 500 ft. The skirt fire self-extinguished. About 5-10 minutes later, the pilot selected a small park as a landing area. She had not changed fuel tanks, consequently, only one burner was available for the final approach to land. The balloon landed short, striking the bank of a small creek. The basket was dragged on its side until the envelope draped over a set of powerlines. All occupants climbed from the basket when advised by the pilot and ground crew that it was safe to do so.

Balloon examination

No pre-existing fault was found with the balloon envelope or the burner and fuel system. However, fire damage to the balloon skirt and to the lower envelope was evident. The basket was abraded and some of the cane structure had fractured.

Pilot experience and currency

During the previous 90 days, the pilot had flown 2.5 hours in command on the balloon type, and had completed a currency check on 18 September 1994. However, she was not familiar with the area over which the accident flight was conducted. Wind Effect

The Bureau of Meteorology analysed the meteorological conditions spanning the flight time of the balloon. The analysis indicated a 5 to 10 kt wind to 500 ft from a south-south-westerly direction, with a more westerly component at the lower levels. The air in the lowest layers was stable and the wind strength was too weak to cause mechanical turbulence. It was possible that uniform descending air was present in the lee of hills around the Wavell Heights area, but the effect would have been light. This analysis was based on an open terrain assessment, consequently the conditions may have been enhanced by local funnelling between buildings.

The houses which were struck by the balloon were some 500 m downwind of a low ridgeline. The pilot of another balloon operating in the area at a greater height said that he had noticed a flattening of the balloon envelope, indicating an increase in wind velocity or possibly a down draft. Flattening of the envelope would have the effect of expelling hot air, resulting in a loss of lift. The pilot of the accident balloon reported that the balloon had been pushed into the side of the house by a very severe downdraft. Continuous power from both burners had been used by the pilot in an attempt to regain control, but sufficient lift was not achieved.

ANALYSIS

The weather conditions as assessed by the Bureau of Meteorology and those reported by the pilot of another balloon, suggest that the pilot of VH-AQF should not have experienced difficulty due to wind effect alone. However, the operation of a balloon at a very low height requires that the pilot must be extremely aware, particularly of the topography and the local conditions. The pilot's ability to constantly anticipate the influence of these factors on the performance of the balloon is critical to its safe operation.

The pilot reported that a severe downdraft had affected the balloon. However, it is likely that the balloon had been in a subtle descent for some time before the pilot realised. Consequently, power was applied too late to avoid collision with the house. The pilot did not then consider the option of deflating the balloon and terminating the flight but applied continuous power from both burners. The dragging basket and surface wind partially flattened the envelope. However, the pilot persisted, despite further collisions, until the balloon lifted clear. The pilot subsequently landed the balloon heavily when only one burner was available to control the descent.

The reported open balloon panel was most likely a rotational vent. The open vent should not have significantly affected the performance of the balloon.

CONCLUSIONS

Findings

1. The pilot was qualified to conduct the flight.
2. The balloon launch was delayed, resulting in exposure of the flight to winds which were stronger than anticipated.
3. The pilot was not familiar with the area over which the flight was conducted.
4. The balloon entered a descent, probably caused by a weak downdraft or windshear.

Significant Factors

1. While being operated at a very low height the balloon entered a descent.
2. The pilot applied power too late to prevent collision with a house.

SAFETY ACTION

Action by the CAA

On 19th October 1994, the District Flight Operations Manager of the Archerfield District Office of the Civil Aviation Authority issued a variation to the operator's permit to operate hot-air balloons. Paragraph 5 (g) (iii) of the operator's permit was amended to read:

(5) The pilot in command of the balloon shall conduct the flight(s) in accordance with the following procedures:

(g) not fly a balloon in free flight;

(iii) within a radius of 300 metres of any city, town or populous area below 1000 feet a.g.l.;

Prior to the variation, sub-paragraph (iii) began with the words 'except during launch and landing'.

With effect from 31 October 1994, the above Variation to Permit was rescinded and replaced with an Operations Manual Directive which was issued to all Commercial Balloon Operators in the Archerfield District Office's area of responsibility. The directive was issued to assist in ensuring safety by improving the reliability of commercial balloon operations in the vicinity of built up and populous areas. Other requirements were also imposed which are designed to assist in the safe conduct of these flights by limiting operations in areas which are likely to cause an increase in risk. The directive specifically mentioned that Operations Manuals were to be amended to address conditions for flight below 1,000ft above obstacles over built up areas; it is not acceptable to conduct prolonged flight at low level during launch or landing procedures.

Interim Recommendation

The Bureau issued interim recommendation IR950093 on 19 May 1995 which stated:

The Bureau of Air Safety Investigation recommends that the Civil Aviation Authority review the rules pertaining to hot-air balloons, particularly in relation to flight below 1,000 ft over populous areas.  In conducting this review, the acceptability or otherwise of such operations to the community, should be considered. CAA Response

The CAA response to IR950093 is stated in part as follows:

"I refer to Interim Recommendation IR950093, Occurrence 9402878 involving VH-AQF on 9/10/94 and concerning the review of rules pertaining to the flight of hot-air balloons over populous areas.

The Authority has examined the circumstances surrounding this occurrence and does not entirely agree with the

Bureau's summary of the deficiency. Balloons have no special privilege permitting them to fly at altitudes below 1000 ft over built up areas. They are only permitted to operate below 1000 feet in the same circumstances as aeroplanes, i.e. when approaching a suitable place for a landing or when taking off from a suitable place, in accordance with Civil Aviation Regulations 92(1)(d) and 157(4)(e)

Whilst passengers may have been told that it was normal to cruise at low level and converse with people on the ground, this is not the case during flight over populous areas. The intention is that balloons only fly low to the minimum extent necessary to manoeuvre to a suitable landing area.

It is clear that the operator involved in this accident may have become accustomed to carrying out prolonged cruising flight at low level over populous areas. This practice may have led to the balloon encountering a local windshear effect at a considerable distance from the intended landing site. A lack of currency and situational awareness on the pilot's part are also considered to be factors.

As a result of this occurrence the Authority has issued directions to a number of balloon operators in respect of the conduct of flights over cities, requiring them to define standards of operation, training and currency for their pilots and standards for the launch and landing areas used. The material is currently being reviewed with a view to defining more precisely what is safe, expeditious, and permissible, compared to that which is not permitted or appropriate.

Community perceptions, as well as considerations of the differences between the dynamics of balloon and aeroplane or helicopter flight, will be taken into account in carrying out this review".

This response was categorised by the Bureau as CLOSED-ACCEPTED.

An emotionally disturbed young woman was allowed to board the aircraft as the sole passenger for the 25-minute flight. At top of climb, the passenger attacked the pilot and made an attempt to exit through the pilot's side window. In the process, she disturbed the aircraft and engine controls. The aircraft descended out of control from 4,500 ft and after pushing the passenger out of the cockpit, the pilot regained control by 3,000 ft. After being threatened by the pilot, the passenger remained subdued for some time. However, approaching Bundaberg, she again became agitated and harassed the pilot. The aircraft landed safely.

Resort staff allowed the passenger to board the aircraft unescorted despite the availability of suitable staff. The passenger had been "playing up" all weekend had caused problems at the resort.

The aircraft was being vectored for an intercept of the runway 35 localiser at Canberra airport. While flying the 160-degree radial, 17 nautical miles from Canberra, on a heading of 310 degrees at 6,000 feet, a false capture of the localiser occurred. The crew identified the false capture, reselected the heading mode and held that mode until positively established on the localiser. The approach and landing was completed without further incident.

The investigation found that the problem was not new and had been known for a number of years, although very few formal reports had been raised. The domestic operators had issued notices to their Boeing 737 crews advising details of the problem. The operators required that any further occurrences were to be reported to air traffic control, followed up by a formal air safety incident report. The Civil Aviation Safety Authority issued a NOTAM containing similar advice and requirements.

In March 1995 the Bureau's data base was modified to specifically record false localiser captures. In the 18-month period to August 1996 the data base captured 17 events that had affected all the major types of high-capacity regular public transport aircraft on the Australian registry. While the majority were reported on approaches to Sydney airport, false captures were also reported at Canberra, Wellington and Christchurch airports.

False capture results in many different indications depending on the type of aircraft, the equipment fitted, and the mode selected on the autopilot or flight director. With certain modes selected an aircraft can deviate from the desired track and follow the false signal.

Research has shown that this phenomena has been recognised overseas and some work has been carried out in an effort to identify the causal factors.

One group believes that the false capture occurs due to significant scalloping, or signal reversals, in the region of the localiser where the signal is normally saturated. Testing of transmitters at Australian airports by the Civil Aviation Safety Authority has shown that the radiated signals meet the International Civil Aviation Organisations requirements for signal propagation.

Another organisation suggested that the capture criterion for some flight management systems and the algorithm for some digital receivers make invalid assumptions about the localiser signal.

Due to the lack of substantive data, the Bureau's Safety Programs group has initiated a research project aimed at identifying the extent of the problem, quantifying its effect and making recommendations as necessary. The results of the project will be published in due course.

The aircraft took-off normally but climbing through about 400 ft it suffered a sudden, complete engine failure. The pilot carried out emergency procedure checks, which included changing from the left fuel tank to the right, but when this failed to immediately restore power he changed back to the left tank and pumped the throttle vigorously without success. The pilot then carried out a forced landing into the sea adjacent to the shore. All occupants evacuated the aircraft uninjured.

The aircraft had been refuelled the previous day to its maximum capacity of 330 litres and then flew for 2.3 hours. The pilot then dipped the fuel tanks which showed 170 litres of fuel remaining. The next morning, he did not recheck the fuel quantity using the dip stick but relied on the fuel gauge readings which at times are inaccurate. These indicated 40 litres in the left tank and 80 litres in the right. The aircraft then flew for another .6 of an hour using the left tank. The engine failed during the next take-off.

An inspection of the aircraft revealed that the right tank contained fuel, but the left tank was empty. The fuel consumption rate was advised as being 70 litres/hour.

The pilot had not selected the tank which indicated the greater quantity of fuel for take-off, and when the engine failed he changed from the left tank to the right, but did not wait long enough for air to expel from the engine fuel feed line to allow resumption of fuel flow from the right tank to the engine. Pumping the throttle would not have had any effect on regaining power as the engine is fuel injected.

The aircraft was being used for a pipeline survey, involving a number of landings and take-offs, prior to the accident. As the aircraft approached Red Hill, a survey check point, the pilot noticed that the fuel gauge indicated that the equivalent of 170 l of fuel remained. He elected to proceed to Paraburdoo, 116 NM away, to refuel. The pilot calculated that the trip would take 77 minutes, and he would arrive at Paraburdoo with approximately 40 l of fuel remaining (the minimum operational requirement).

About 20 minutes after leaving Red Hill the pilot noted that the fuel gauge still indicated 170 l. It remained at this reading for a further 10 minutes before beginning to oscillate and fall rapidly. The pilot reassessed the readings and determined that he still had sufficient fuel for the flight.

Approximately 5 km north-west of Paraburdoo, as the aircraft was descending through 3500 ft, the turbine temperature and engine RPM began to fluctuate. The pilot assumed an engine malfunction and entered an autorotational descent rather than risk a more significant problem during a normal approach.

When 500 ft above the ground the pilot observed that the area he had selected for landing was unsuitable. He adjusted his approach to land at a more suitable site however, it was further away. During the extended glide, to the new site, the pilot allowed the rotor RPM to decrease (the low RPM warning sounded during the approach) and in the ensuring heavy landing a main rotor blade struck and damaged the tail boom. The engine continued to run at idle during the approach and was shut down by the pilot after landing.

A post-accident inspection found that the fuel gauge indicated the fuel tank contained the equivalent of 19 l of fuel.

Draining the tank retrieved around 15 l of a fuel/water mixture (confirmed by laboratory testing) from the aircraft. Additional fuel/water was found when the fuel contents sender units were removed for testing. Tests on the fuel contents sender units indicated that it was possible for the gauge to stick between readings of 114 and 190 l.

Calculations indicate that the helicopter left Port Sampson with an endurance of 220 minutes and that it had been operating for about 206 minutes at the time of the engine malfunction. The low fuel pressure lights did not illuminate during the accident sequence indicating that the fuel inlet remained covered at all times. Background information indicates that the turbine engine fitted to the Bell 206 helicopter is tolerant of water held in suspension in the fuel however, ingestion of a fuel/water mixture where the percentage of water has increased as a result of its settling out of the fuel into the bottom of the tank, may cause fluctuations similar to those reported by the pilot.

The most likely reason for the engine malfunction was that the low fuel state raised the percentage of water in the fuel/water mixture to a point where it began to effect engine performance.

Evidence indicates the pilot did not carry out water disclosing check at Port Sampson prior to departure. Had he done so he may have determined that water was present and taken action to correct the situation before it led to an emergency landing.

The low RPM state prevented the pilot from arresting the descent sufficiently to prevent a heavy landing.

The pilot did not complete a flight plan or a fuel log for or during the flight. Consequently, he had to rely solely on the fuel gauge indications which, in this case, may have been incorrect.