The Aerostar was being flown on a passenger charter flight from Maitland, NSW to Olympic Dam, SA. Arriving in the circuit area, the pilot reported that he observed the windsock indicating a surface wind of 240 degrees at 10 knots. However, because the runway length of 1591 m was more than adequate, he elected to land on runway 06 with a tailwind component. He said he flew a normal approach using 45 degrees of flap, touching down about 150 m beyond the landing threshold. 

Shortly after touching down the aircraft was struck by a strong wind gust and diverged to the left. The pilot was unable to prevent the aircraft from departing the sealed runway surface and entering an area of soft earth. The right main landing gear collapsed, and the right propeller struck the ground, stopping the engine. After the aircraft had stopped, the pilot shut down the left engine and evacuated the two passengers. Nobody was injured.

A local pilot who arrived at the site shortly after the accident estimated the wind as 240 degrees at 25 to 35 knots, but with regular gusts from the north. He also reported that the wind conditions had been similar all morning. The pilot of the Aerostar subsequently reported that local pilots said that the terrain around the landing area can produce a wind funnelling effect along and across the runway, which may not have been readily apparent.

A significant factor in this accident was the decision of the pilot to land on a runway affected by a marked downwind component.

The date and timing of the occurrence are approximate only.

The aircraft was being used, on behalf of Airservices Australia, to conduct air tests on the static-line used with the PADS supply dropping system. A flat-strap static line, with a cone attached, was trailed behind the aircraft and its movement observed on a video camera. The tests indicated that the line could oscillate severely under certain conditions. On one of the test flights the line oscillated until it collided with the rear fuselage. The aircraft was not damaged, and its flying characteristics were not effected. The line was pulled in and a normal landing completed.

As a result of the tests, the flat-strap line was replaced with a round line and the cone modified. Subsequent tests indicated that this configuration was stable, and the oscillations were not repeated.

While taking off for parachute dropping operation, directional control was lost and the aircraft ground looped. The left main landing gear collapsed, substantially damaging the aircraft. All six occupants evacuated the aircraft without injury.

The pilot subsequently reported that, as take-off power was applied, the aircraft started to swing to the left. He applied right rudder to correct the swing, but the aircraft then swung rapidly to the right. He was unable to counteract the swing and closed the throttle to stop the take-off. However, the left wheel ran into a culvert and was torn off before the aircraft could be stopped.

The pilot reported that he was conducting a practice forced landing when the engine stopped. At 3,000 ft he had applied full carburettor heat, closed the throttle and commenced a glide descent. During the descent he applied power on four occasions to re-warm the engine. At about 600 ft AGL the engine stopped along with the geared wooden propeller. The pilot carried out trouble checks and declared a Mayday. He twice engaged the starter in an attempt to restart the engine but was unsuccessful. He subsequently reported that the starter seemed to be sluggish.

The aircraft touched down in a paddock on a downhill slope. It rolled about 30 metres, crashed through a farm fence, passed under a powerline, crossed a road, encountered a ditch and overturned.

At the time of the accident weather conditions were: CAVOK, outside temperature about 15 degrees Celsius, wind calm, visibility 10 km plus, nil precipitation.

This was the second accident to VH-IDB in the same area due to the engine stopping during practice forced landings. The previous accident occurred on 14 May 1997. Verbal reports of other instances have been received of Skyfox Gazelles experiencing engine stoppages after the throttle was closed and engine RPM reduced to idle.

After the first accident, the engine was inspected by a licensed aircraft maintenance engineer (LAME). Then, after the second accident with the same engine installed, it was inspected first by the LAME and then by the engine manufacturer's agent. No fault was found with the engine after each accident. Post accident tests on the starter, the battery and the engine have failed to detect a reason for the sluggish starter anomaly reported by the pilot.

Safety Action

As a result of complaints about the stalling of engines at idle, Airworthiness officers from the Civil Aviation Safety Authority (CASA), in conjunction with the aircraft manufacturer, conducted an investigation which included test flights in the Gazelle.

It was found that the throttle mechanism had enough free play (also known as backlash) to reduce RPM below idle and stop the engine. This could occur if extra force was applied when pulling the throttle back to idle speed. To address this problem, the aircraft manufacturer issued Service Bulletin (SB) No. 20 on 28 October 1997. This SB provides the following information: "Idle speed may inadvertently be reduced below set minimum due to incorrect throttle stop adjustment." SB 20 requires operators to check the idle setting and adjust for backlash at the throttle stop in the cockpit. Compliance was mandatory, within two weeks or 10 hours time in service from receipt of the SB.

VH-WNZ was conducting a training flight, and the pilot had advised the Tower of his intention to conduct a simulated engine failure after take-off. The aircraft departed on an initial assigned radar heading of 060 degrees at 1710.40 EST.

VH-OGK was assigned an IBUNA 1 Standard Instrument Departure (SID), departed at 1712.30 EST, and had been instructed to remain on Tower frequency. While turning right as per the SID to heading 180 M, OGK passed behind WNZ. Separation between the aircraft reduced to 2.5 NM horizontally and less than 500 ft vertically. The required separation was 3 NM and 1,000 ft.

Investigation revealed that the aerodrome controller (ADC) did not adequately plan the departure of three aircraft, including WNZ and OGK. An arriving aircraft led to his decision to expedite the departure of OGK, thus compromising the separation between these aircraft. The ADC's decision to retain OGK on Tower frequency until instructed to transfer delayed the opportunity for the departures controller to issue instructions to OGK to ensure separation was maintained.

The pilot stated that while hovering the helicopter at about 50 ft over a dry riverbed, the engine note suddenly changed. He checked the instrument indications which confirmed that the engine RPM had increased. Almost immediately the indications returned to normal. He began to move the helicopter away from trees when the engine RPM again increased as drive to the main rotor system was lost.

During the attempted autorotational landing, the helicopter collided with a tree and landed heavily on the riverbank.

While vacating the helicopter, the pilot became aware of a "rumbling" noise and smoke coming from under the cabin. He attempted unsuccessfully to locate the source of the noise by isolating the electrical systems. He then switched off the battery.

Output from the helicopter engine is coupled through a V-belt drive system to the main transmission which drives the main rotor, and to the tail rotor drive system. The belt drive clutch control installation includes a linear actuator and electrical connections to a clutch control switch and warning light on the instrument panel. A cable and pulley interconnect the linear actuator to the clutch spring on the belt drive transmission. The clutch control switch, which is positioned on the lower left side of the instrument panel, has three positions, RELEASE, HOLD and ENGAGE (the normal operating position). With the switch in the ENGAGE position, the linear actuator retracts, applying tension through an idler pulley to the V-belts. The clutch warning light is on unless the clutch is fully engaged.

Examination of the helicopter rotor drive system did not identify any pre-existing defect which may have contributed to the accident.

The "rumbling" noise heard by the pilot was caused by the starter motor operating due to impact damage to the solenoid. There was no fire.

The linear actuator shaft was found to be extended by approximately 30 to 50% of its travel. It is normally fully retracted when the drive belts are correctly tensioned. The actuator, the clutch control switch, the clutch warning light and the associated wiring were tested and found to be serviceable. It was not possible to determine if the warning light was powered at the time of impact. The position of the linear actuator shaft may have changed following the accident when the pilot repositioned various switches and circuit breakers in an attempt to identify the "rumbling" noise.

The clutch control switch was fitted with a guard to prevent inadvertent operation. However, the retaining springs had weakened such that the guard was ineffective.

The circumstances of the accident are consistent with loss of rotor drive due to insufficient drive belt tension. However, the circumstances in which this occurred could not be determined.

The hot air balloon had taken off from Picton on an early morning joy flight to Camden. Surface wind conditions were calm, with a south-westerly wind of 24 kt at 1,000 ft. Surface winds were not expected to increase before late morning.

The pilot in command reported that the flight proceeded as planned, and a normal approach was made for a landing on Camden aerodrome. The surface wind was 5-8 kt. Some 200 m from the landing point, at a height of about 50 ft, the rate of descent began to increase more than required. The pilot activated two burners for about 5 sec to arrest the descent. As expected, the balloon flew level for a short distance, then unexpectedly began to climb rapidly. The parachute valve was opened to counteract the climb, which stopped at about 200-250 ft. However, the balloon then began to descend rapidly. All burners were turned on in an attempt to arrest the descent, but to no avail. The balloon impacted the ground at an estimated rate of descent of 600-700 fpm.

The rapid deflation system was operated just before impact, but the balloon was dragged across the ground for about 30 m before becoming airborne again, reaching a height of about 20 ft before again contacting the ground and being dragged a further 40 m. As the balloon came to a halt the pilot reported that the deflated envelope flapped in the breeze for some time, suggesting an increase in the surface wind.

One lady sustained a fractured ankle during the landing. The other twelve occupants were uninjured.

FACTUAL INFORMATION

Background

The pilot/owner of VH-CSW had been approached some time before the accident with a request to allow passengers to abseil from his helicopter while it was in a hover. The pilot approached a Civil Aviation Safety Authority (CASA) representative who advised the pilot that he would need to submit a proposed amendment to the company's Operations Manual detailing how such flights would be conducted. The representative also advised that any proposed amendment would have to be approved by CASA before any such flights could be conducted and that the representative had already approached a CASA helicopter specialist to discuss safety aspects of the proposed flights.

Because of the CASA requirement for a detailed proposal for conduct of the proposed flights, the pilot decided to conduct some controllability tests prior to submitting his proposed Operations Manual amendment. Because the pilot declined to be interviewed or to supply the investigation with a statement, it is not known whether he conducted any calculations into the lateral centre of gravity (cg) implications of the proposed flights.

The pilot positioned the helicopter to a helipad near the accident site on the morning of the accident and carried out some local flights prior to the accident flight. Another helicopter pilot, who had flown for the company on a casual basis in the past, drove to the area with addition fuel for the day's planned flights.

A briefing was conducted by the pilot to explain how he proposed to conduct the experimental flight. Present at that briefing were the passenger, who had accompanied the pilot to the site in the helicopter that morning and who had an interest in the proposed abseiling flights, the casual helicopter pilot and another local person. A fourth person was also present at the scene but did not attend the briefing.

The proposed flight discussed at this briefing required that an abseiling rope be attached to the outside of the front cross-tube of the skid assembly of the helicopter and that one passenger be seated in the right seat of the helicopter. The rope would be attached in such a manner that it hung over the outside edge of the right skid tube. The pilot then intended to lift the helicopter to a hover about 10 m above the helipad and the casual pilot was briefed to walk under the hovering helicopter and to gradually apply his weight onto the abseiling rope. One of the persons briefed earlier had radio communication with the pilot through a hand-held radio transceiver.

History of the flight

The flight proceeded as planned initially and the pilot was able to remain in control of the helicopter with one person, the other pilot, who weighed about 70 kg, suspended by the abseiling rope. The person at the helipad stated that the pilot then contacted him by radio and asked that the other person standing there (who had not attended the briefing) also put his weight on the rope. That spectator quickly moved under the helicopter and applied his weight, reportedly about 80 kg, onto the rope.

The helicopter rolled rapidly to the right and began to go out of control. When it became evident to the persons on the rope that the pilot could not control the helicopter with both their weights applied, they both quickly let go and attempted to move away from under the helicopter. Although the casual pilot managed to get clear, the other person's ankle became entangled in the rope and he was lifted by the helicopter as it banked steeply right, (now out of control), and it proceeded to pivot 180 degrees to the right and move across the road towards a small service station, dragging the person below it by the leg.

One of the helicopter's main rotor blades struck the steel upright of a large illuminated sign; the helicopter then descended and struck the rear of a vehicle parked in front of the service station. The helicopter landed heavily beside this vehicle with its tail boom broken off but with its engine still running and main rotor still turning, causing major damage to the vehicle and multiple blade strikes to the building.

The pilot was winded in the accident, partly because the shoulder harness section of his inertia reel seatbelt had failed on impact. He managed, with some difficulty, to activate the emergency fuel cut-off to shut down the engine after unsuccessfully attempting to stop the engine by switching off the magnetos.

Spectators arrived to assist the pilot and passenger from the severely damaged cabin and to extinguish a fuel fire that had started below the ruptured left fuel tank.

The two occupants of the helicopter and the person caught by the leg by the abseiling rope all received minor injuries. The helicopter was destroyed and severe damage was caused to the service station buildings and the vehicle parked in front of the office. Although there were persons in the buildings when the helicopter crashed, none of them were injured.

The aircraft

The helicopter had been manufactured in 1970 and its data plate identified it as a Bell 47G-3B-2. It had been converted some time later to a Bell 47G-2A-1, although registration records show the type as a Bell 47G-3B-1.

Examination of the aircraft at the accident site and a search of maintenance records proved that the aircraft was suitably registered and that the maintenance release was current. No defects that would have affected the flight were apparent. The failed pilot's shoulder harness inertia reel was recovered for failure analysis.

The pilot was suitably licensed and endorsed on the aircraft type.

ANALYSIS

The Flight Manual for the helicopter type did not cover abseiling operations and there was no procedure for the attachment of ropes or lifting devices to the skid tubes in the manner used on the accident flight. There was no centre of gravity (cg) chart provided in the manual for loads attached to the skids. In normal operations, any loads that are to be suspended below the helicopter are attached to the purposely installed cargo hook mounted under the helicopter close to the cg. CSW was fitted with such a device which has both electrical and mechanical methods for instantly releasing the load when required, especially in an emergency.

The forward skid cross tube was about 780 mm forward of the optimal longitudinal centre of gravity and any load applied on a rope attached to it would cause the helicopter to pitch nose-down as well as causing it to roll to the right.

The Flight Manual for the Bell 47G series of helicopter has a section on loading of cargo litters (if installed) which states that the maximum load allowable on a litter is 102 kg. Litters are mounted fore-and-aft on the skid cross-tubes between the cabin side and the outside of the skid. An asymmetric loading chart is provided with the litter kit and is used for calculation of maximum allowable difference between loads on the left and right litters. As an example of the sensitivity of this type of helicopter to asymmetric lateral loads, if there is a passenger of 75 kg in the right seat, it is only possible to carry about 55 kg more on the right litter than on the left to remain inside allowable lateral cg limits. As the outside of the skid tube is several centimetres outboard of the litter centreline, any load applied at the skid tube would have a greater adverse effect on the lateral cg than the same load on a litter.

On the accident flight, the pilot was simulating an asymmetric load of about 150 kg (no load on the left skid tube, two persons hanging from the right skid), attached to the helicopter in such a manner that the lateral cg limits imposed by the manufacturer were exceeded. This lateral cg exceedance combined with a pitch-down of the nose as the load was applied, resulting in a loss of control.

The carriage of a passenger of about 80 kg in the right seat would have resulted in the helicopter being close to a laterally balanced condition before any load was applied to the rope. If it had been decided to conduct a proving flight after calculation of the lateral cg implications, the conditions would have been better simulated with no passengers in the cabin or with one occupying the centre seat. It would be expected that the person from the right seat would have been the one abseiling from the helicopter. A seating configuration which left the right seat empty would have reduced the exceedance of lateral cg experienced on the accident flight.

The two persons applying their weight to the rope attached under the helicopter released their hold when they realised the helicopter was not maintaining a steady hover. As one of the persons stepped backwards, his ankle became entangled in the rope. He was lifted by the helicopter then dragged up a bank onto a road. This dragging load, which could not be released from the aircraft in an emergency, probably prevented the pilot from regaining control before the helicopter struck the sign.

Although abseiling from helicopters is frequently carried out in military operations, the aircraft used are normally larger, cabin-class helicopters and it is possible for persons abseiling from the helicopter to do so from each side simultaneously, thus reducing the risk of an exceedance of lateral cg limits.

Investigation into the failure of the pilot's shoulder harness inertia reel revealed that the unit fitted to the helicopter was not approved for installation in Bell 47G aircraft type.

SIGNIFICANT FACTORS

The pilot was attempting to explore the effect on helicopter control of a lateral load applied to the outside of the right skid tube.

The lateral centre of gravity limitations of the helicopter were exceeded, resulting in loss of control.

The pilot was unable to regain control before impact with a sign when a person became entangled in the rope attached to the helicopter.

The pilot was tasked with flying passengers from Port Augusta to Cook, via Ceduna. After arriving at Cook the aircraft was flown without passengers to Nullabor for an overnight stop. The following day, he was to return to Cook to collect a group of eight passengers and fly them to Port Augusta. The aircraft had been refuelled to full capacity at Ceduna on the flight out to Cook. During this refuelling the pilot did not note the amount of fuel used on the flight from Port Augusta. The pilot calculated that the fuel remaining at Cook for the return flight would be 439 litres. He also calculated that the fuel required to return to Port Augusta from Cook would be 452 litres. This fuel requirement was calculated using a fuel flow of 140 litres per hour and a true airspeed of 170 knots with a 15-knot headwind. The fuel requirement included a 100-litre fixed reserve. The pilot was aware that there was a 13-litre shortfall between the fuel on board and the fuel required. He did not eliminate this discrepancy by adding fuel at Nullarbor where it was available. The weight of the aircraft on departure from Cook was calculated by the pilot as being 35 kg below maximum take-off weight. However, this calculation was based on 423 litres of fuel, rather than 439 litres.

The aircraft departed Cook at 0007 UTC and the pilot passed an estimate with his departure report of abeam Ceduna at 0116. The aircraft reported abeam Ceduna at 0116 and gave an estimate to flight service for Port Augusta of 0238. This estimate gave a time interval which was 9 minutes longer than planned by the pilot. The pilot had submitted a flight plan to ATS that indicated a groundspeed of at least 165 knots would be achieved during the flight. The pilot had planned the flight expecting the true airspeed to be higher, and therefore the groundspeed to be higher, on the return flight. He also expected to obtain favourable winds at various altitudes and therefore did not plan to use the reserve fuel of 100 litres. He also assumed that he would obtain a better fuel flow than the 140 litres per hour that he had used during the planning for the flight. The flight progressed uneventfully until approximately 90 nm from Port Augusta when the pilot noticed that the fuel remaining on board had decreased to a lower level than expected. He elected to divert to Wudinna as he was now in some doubt as to whether the aircraft would make Port Augusta with the fuel remaining on board.

The decision to divert to Wudinna was based on the fact that the company held fuel stocks at this location. He advised flight service that he was diverting due to a higher-than-normal fuel burn. Flight service declared an alert phase at this time due to the pilot reporting 20 minutes endurance with a time interval to Wudinna of 10 minutes. En-route to Wudinna the pilot noticed that the airfield was situated beyond a heavily timbered area and he elected to carry out a precautionary search-and-landing in a cleared paddock whilst there was fuel remaining, rather than continue over the timbered area where he might suffer fuel exhaustion. The pilot advised flight service that he was carrying out a precautionary landing and they declared a distress phase due to the unsure fuel status of the aircraft. The pilot completed the precautionary search-and-landing without damage to the aircraft or injury to the passengers. The position of the precautionary landing was 19 nm north-east of Wudinna. The company was advised, and the aircraft was recovered to Wudinna later that day.

The estimated fuel remaining on board was less than 30 litres. In his report to the Bureau as part of the investigation, the pilot indicated that he operated the aircraft during both the outbound flight to Cook and the return flight to Port Augusta using fuel flow and exhaust gas temperature (EGT) settings from another company aircraft. These settings had resulted in an actual fuel flow of approximately 165 litres per hour.

Significant Factors

1. The pilot used a fuel flow setting during flight planning that was not representative of that which the aircraft would achieve during flight.

2. The pilot departed from an aerodrome where fuel was available knowing that the fuel on board was less than that required for the flight, with reserves.

The pilot was conducting circuits to gain familiarization with the aircraft. Late in the landing roll he encountered a willy-willy, which caused the aircraft to swing. Using right brake, the pilot was unable to control the swing, and the right wheel assembly broke off. The right wing fell to the ground and the pilot exited the aircraft when it stopped.

The aircraft was not fitted with an electronic locator beacon.