The KAflex shaft had been fitted in accordance with the published requirements in the supplemental type certificate (STC) about 6 years, or 4,112.35 flying hours, prior to the occurrence. As far as could be determined, no certifications had been made that the shaft had been inspected in accordance with the STC inspection requirements during that period.

There was no flight manual supplement supplied as part of the STC, which would have alerted pilots to the specific inspection requirements and the significance of red dust production. As well as the daily inspection requirements for the shaft, the supplement did not include the warning not to disturb the bolts and to reject a shaft that showed evidence of turning of the fasteners.

The STC documents supplied with the shaft were not kept with the current helicopter logbook, although it was readily available. The retention of the STC documents in the archived logbook, instead of the current logbook, meant that the STC inspection requirements were overlooked and consequently not actioned by maintenance personnel. That resulted in the STC inspection requirements not being included as part of the routine or scheduled maintenance paperwork packages when they were assembled for release to service maintenance events.

The historical service record card for the shaft could not be located and had not been included in the helicopter records as required by the manufacturer in the STC accomplishment instructions. The minimal reference made to the card in the STC could easily have lead to the card being overlooked. When an example of the card was obtained, the appropriate location for the certification of maintenance activities was not readily apparent. The lack of a specific area on the card to certify completion of maintenance may have also contributed to the non-use of the card by maintenance personnel, including the 1500-hourly inspection certification in the airframe logbook. Also, inclusion of these inspection certifications in the airframe logbook could lead to the certification history for a driveshaft being lost if a driveshaft was subsequently moved from one helicopter to another. When overhaul became due, this service history would also have been unavailable to the manufacturer when the shaft and accompanying historical service card were returned to them.

The shaft failure had been initiated by fretting type movement at a flex frame bolted joint. This progressed until the joint failed, resulting in the gross overload failure of the remaining frames. That movement, in its early stages, should have been detectable by the presence of the red dust or loose bolted joints described in the STC inspection warnings. Had the maintenance and operating personnel been aware of the STC inspection schedules, they would have had a better understanding of the significance of the red dust and its ramifications.

Although the operator submitted that his personnel would have detected red dust had it been present around the flex frame joints, the wear pattern evident on the flex frame joint in Figure 1 was consistent with the flex frame fastener being loose for a period of time prior to the failure. The loose bolted joints were not detected. In this occurrence, the loss of bolted joint integrity may have progressed past the point where dust production may occur.

At 1702 central standard time on 14 June 2004, a Bell Helicopter Company 206B(II) Jetranger, registered VH-PHF, was being operated for a medical evacuation from Deep Well, NT, to Alice Springs. About 5 NM south-west of Alice Springs Airport, while cruising at 500 ft above ground level at 100 kts, the pilot felt a vibration and heard a loud bang accompanied by a reduction in main rotor torque. The pilot immediately placed the helicopter into an autorotative descent and broadcast a MAYDAY¹ to the Alice Springs Air Traffic Control aerodrome controller. The pilot landed the helicopter in a clearing and the five occupants were uninjured.

Initial inspection of the helicopter by the operator revealed that the KAflex² driveshaft between the engine free-wheeling unit and the main transmission had failed. The failed shaft and its components were removed from the helicopter and forwarded to the Australian Transport Safety Bureau (ATSB) for metallurgical examination. The examination found that the failures had occurred in the arms of the web elements, with all of the fractures typical of gross-overload failure from either a single or small number of cycles. Of significant interest was the separation of one of the flex frame unions (see Figure 1). This section showed evidence of wear and deformation consistent with looseness of the bolted joint fastener. Fatigue cracking had initiated from the bore or worn surfaces of the frame and propagated radially away from the hole, intersecting the side of the frame and freeing the connection.

Figure 1: Worn fastener bore and fracture in flex frame. Arrows indicate the point
of fatigue crack initiation and the direction of propagation.

Maintenance history

The helicopter's Maintenance Release was valid until 27 November 2004 or 12,351.80 hours in service, whichever occurred first. At the time of the failure, the helicopter had 12,168.25 hours in service. The operator had responsibility for the maintenance of the helicopter, which was conducted under a valid Certificate of Approval. The helicopter was operated in the Normal category, Day VFR.

The helicopter was maintained as a Class B type aircraft, with the airframe maintenance conducted in accordance with the manufacturer's prescribed maintenance procedures. In this instance, the Bell Helicopter Textron Company 206 Maintenance Manual BHT-206A/B-SERIES-MM-1 was identified in the Logbook Statement as the primary documentation. However, the driveshaft manufacturer's documentation was not identified as required supplemental documentation in that statement. The inspection schedule worksheets for maintenance were copied, for use by the maintenance personnel, directly from the airframe manufacturer's manuals. No supplemental maintenance inspection sheets were incorporated into any of the maintenance worksheet packages that specifically identified maintenance actions to be performed for the KAflex driveshaft.

The helicopter underwent a 300-hourly inspection for the issue of a maintenance release at Moorabbin, Victoria on 27 November 2003. No entries were recorded on the maintenance release with regard to any daily or periodic inspection requirements specific to the KAflex driveshaft.

Main driveshaft

The Jetranger was delivered with a proprietary designed main driveshaft. The driveshaft, which comprised a spherical coupling at either end of a torque shaft, was designed to transmit power from the engine freewheeling unit output adapter to the main transmission input quill. The engine's output shaft speed was around 6,000 RPM³ during helicopter operation. The application of power and flight and ground loads all contributed to drive line misalignment during operation. The spherical coupling design compensated for that misalignment under normal operating conditions. The main driveshaft was subject to ongoing monitoring through heat sensitive temperature indicators that detected overheating caused by inadequate lubrication, wear, or excessive misalignment of the drive train elements. Periodic maintenance of the driveshaft included disassembly, inspection and lubrication in accordance with the Bell Helicopter Textron Company 206 Maintenance Manual, to ensure the continuing airworthiness of the driveshaft.

On 19 January 1998, at 8,055.9 hours aircraft total time in service, the Bell manufactured engine-to-transmission driveshaft in the helicopter had been replaced with a KAflex unit. The new driveshaft was designed to replace the proprietary unit and had been marketed as providing reduced maintenance, longer time between overhauls, and greater reliability.

The Kamatics Corporation web page advised:

Helicopter flight manoeuvres generate high misalignment between the engine and the transmission, which must be accommodated by the connecting driveshaft. Such driveshafts, which rotate at speeds over 6000 RPM, often incorporate grease lubrication and seals. Designs of this type are susceptible to loss of lubrication, which results in overheating and possible failure, a major safety concern.

The KAflex driveshaft is a mechanical drive coupling which requires no lubrication or seals, and transmits power while accommodating high angular misalignment and length change through the use of flexible rectangular frames. These frames are bolted together at the corners in a truss-like arrangement, which are attached to shaft end fittings to allow for drop-in installation in the drive line. A fail-safe feature enables the coupling to continue to transmit power even in the unlikely event of a failure in a load carrying member.

KAflex driveshafts are custom designed for specific applications and selected because they offer superior, maintenance-free performance with extended 'on condition' service-life, resulting in unequalled reliability, increased readiness and cost effectiveness. They are supplied both as individual couplings and as complete driveshafts.

The KAflex driveshaft had been fitted in accordance with Kamatics Corporation supplemental type certificate (STC) SH 7767SW. The helicopter was then ground run and test flown with no defects found. While the KAflex driveshaft remained fitted to the helicopter, the requirements of airworthiness directive (AD)/Bell 206/79 Amdt 1⁴ were no longer applicable to the helicopter.

The STC documentation stipulated that, upon completion of the modifications and installation of the shaft, the historical service record was to be completed, applicable logbook entries made and the card to be kept with the aircraft logbooks. At the completion of its recommended time in service between overhauls, the shaft and the completed historical service card would be returned to the manufacturer for overhaul. While an entry in the helicopter's logbook was made for the installation of the KAflex shaft into the helicopter, no historical service record was found in the helicopter's logbooks pertaining to the shaft.

Maintenance and inspection

A copy of the STC was kept with the archived helicopter logbooks and worksheets, in a separate binder to the current helicopter logbook binder. Both binders were located at the operator's main office. The STC was available to the engineers maintaining the helicopter, but it was not identified by them as a document that they would need to refer to routinely in their maintenance activities.

Although the STC and the manufacturer's website stated that the driveshaft was 'maintenance free' Sections 3 and 4 of the STC detailed the inspection and maintenance regime that the manufacturer expected to be performed while the driveshaft was in service. Section 3 of the STC detailed pre-flight, 100-hourly and 1,500-hourly inspections that were to be performed throughout the 6,000 hour service life⁵ of the shaft. The inspection advice described examination of the flex frames for the production of 'red dust' showing up as a red metallic residue. Section 3 also contained requirements for a 6,000 hour inspection (which it referred to as maintenance requiring return to the manufacturer) and conditional inspections after specific events, such as an overtorque, an overspeed, a sudden stoppage, a hard landing or a lightning strike. Section 3 also contained the following bold type warning with regard to flex frame attachment hardware:

WARNING
DO NOT disturb or tighten flex frame nuts or bolts. Evidence of turning
fasteners by wrench or other means is cause for rejection.

Section 4 of the STC listed maintenance requirements for the driveshaft. That information stated that there was no periodic maintenance requirement for the KAflex driveshaft.

The operator's managing director stated that at no time did any of his personnel detect the production of red dust residue on the shaft. He also advised that had there been red dust production around the flex frame joints, his personnel would have detected it and prevented the failure. The manufacturer stated that red dust is usually produced in the initial stage of loss of integrity of the bolted joints, but noted that this was not always the case and operators should be vigilant with regard to inspection for loose bolted joints.

From the time of installation of the shaft on 19 January 1998 to the time of the occurrence on 14 June 2004, there were no entries detailing the conduct of periodic or 1,500-hourly inspection certification requirements of the KAflex driveshaft in the helicopter's logbooks, or in the worksheets for maintenance for the issue of a maintenance release. This represented the 4,112.35 hours in service for the KAflex driveshaft. However, there was no stipulation in the STC Section 3 instructions to require certification for the completion of the inspections detailed in that section.

There was also no amendment insert in the helicopter's flight manual for the daily inspection as described in the STC, Section 3 - DAILY INSPECTION BEFORE FIRST FLIGHT OF THE DAY. There was also no stipulation in the STC that the flight manual should be amended in order to make that information readily available to the pilot.

When interviewed, the pilot in command was asked to describe the execution of a daily inspection of the helicopter. While a detailed explanation was given to the interviewer, at no time was the driveshaft manufacturer's warning caveat mentioned or alluded to by the pilot. As this is a bold type warning in the manufacturer's documentation, it should have been a recall item readily identified during this discourse.

Historical service record

There was minimal reference made to the historical service record in the STC. A copy of the card was obtained from the local Australian distributor for KAflex. There was no provision on the card for certification of the 100-hourly and 1,500-hourly inspections. The historical service record card had not been incorporated into the maintenance records for the helicopter as required by the manufacturer in the accomplishment instructions of the STC.

¹ International radio broadcast for urgent assistance.
² A proprietary name for a driveshaft manufactured by a Unites States company, Kamatics Corporation.
³ Output shaft speed at 100% main rotor RPM.
⁴ AD/Bell 206/79 Amdt 1 detailed the inspection and installation of a Visual Aid Overheat Indicator on a Bell manufactured main input drive shaft assembly. It was later cancelled as those requirements were incorporated into the 100-hourly maintenance servicing requirements.
⁵ At the time of shaft installation, the STC stipulated a 4,000 hour service life. Service Instruction 2348 Revision "E" dated September 1999 extended this service life to 6,000 hours between overhauls.

FACTUAL INFORMATION

On 15 June 2004, a Hiller Aviation UH-12E helicopter, registered VH-HMT, crashed during an agricultural operation, after the pilot reported a loss of tail rotor authority. The pilot was not injured in the accident. The horizontal stabilizer was found about 150 m from the wreckage site, indicating that it had separated from the helicopter in-flight. The operator reported that damage to the stabiliser and the tail rotor blades was consistent with the separated stabiliser coming into contact with the tail rotor blades, as it was still connected to the helicopter by the taillight wiring. The operator reported that an engineering examination found no evidence of tail rotor system failure.

The horizontal stabiliser spar tube failed in the area where the tube passes through a collar in the doubler attached to the inner stabiliser rib (refer to figure 1).

Figure 1: Horizontal stabiliser and detail of the spar tube and doubler.

Helicopter

The helicopter was manufactured in 1978 and had accumulated 6,903.50 hours time in service at the time of the accident. In September 2002, at 6,386.25 hours in service, it sustained major damage following a sudden main rotor stoppage. The helicopter was rebuilt and had since accumulated approximately 517.20 hours in service. It was maintained in accordance with the maintenance requirements applicable at the time and had a valid Maintenance Release. It had flown approximately 31 hours since the last periodic inspection, a 100 hourly inspection completed in May 2004.

Horizontal stabiliser

The operator advised the Australian Transport Safety Bureau (ATSB) that the horizontal stabiliser is not a serial numbered item and its time in service since new was not known. The stabiliser was fitted to the helicopter when it was acquired from its previous owner. During the helicopter rebuild in September 2002, the stabiliser was inspected in accordance with the 100 hourly component inspection guide and the Hiller UH-12E maintenance manual, before being re-installed. The component inspection guide required the stabiliser tube to be inspected for corrosion and security and the maintenance manual detailed the spar tube inspection requirements and wear limits. No anomalies with the horizontal stabiliser were reported at that time.

Spar tube failure

The failed horizontal stabiliser was forwarded to the ATSB for detailed examination. That examination revealed that the stabiliser spar tube fractured as a result of fatigue cracking. The crack initiation occurred at a number of locations where the tubing had been reduced in wall thickness by wear. It was apparent that the wear of the tubing was associated with small scale repeated movement between the stabiliser tube and the doubler attached to the stabiliser's inner rib. There was no evidence of any pre-existing defect present.

A search of the ATSB, United States National Transportation Safety Board and Federal Aviation Administration databases found no reported events involving an in-flight separation of the horizontal stabiliser on similar helicopter types.

The Australian Transport Safety Bureau did not conduct an on-scene investigation of this occurrence.

Communications

The reason the HF operator was unable to establish two-way communication with the crew of IVC could not be determined. A formal procedure for notifying the controller that the HF radio operator could not pass the instruction to the crew of IVC may not have ensured that a separation standard existed, as an incorrect instruction had been issued. However, it may have prompted a more timely reassessment of the situation by the controller. In other circumstances, a formal procedure may alert the controller in sufficient time for action to be taken before separation standards are infringed.

Air Traffic Controller

A review of the controller's training records indicated that he had an adequate understanding of the concept of lateral separation. However, his misapplication of a lateral separation point may have reflected his lack of proficiency due to inexperience in working the particular sector of airspace.

On 16 June 2004, a Boeing 747-400, registered GC-IVC (IVC), was en-route from Melbourne to Singapore on airway A576. The aircraft was approaching the boundary between the Brisbane flight information region (FIR) and the Bali FIR at flight level (FL) 340. Concurrently, a second 747, registered 9V-SPE (SPE), was en-route from Sydney to Singapore on airway G326. That aircraft was also approaching the boundary between the Brisbane FIR and the Bali FIR at FL 340.

The controller recognised that he had to resolve a potential confliction between SPE and a third aircraft. In resolving that confliction, the controller created a new confliction between IVC and SPE. In response, the controller instructed the crew of IVC to reach FL320 by waypoint ATMAP (see Annex A). That requirement was intended to achieve vertical separation between the aircraft prior to any loss of lateral separation¹. However, the required lateral separation point was 94 NM south-east of waypoint ATMAP on airway A576. The incorrect requirement led to an infringement of separation standards.

Following a scan of the air situation display, the controller realised that he had issued an incorrect requirement. The controller then issued a second requirement, via the high frequency (HF) radio operator, for IVC to descend to reach FL320 by 80 NM to the south-east of waypoint ATMAP. The controller reported that this requirement was based on the application of a required navigation performance (RNP) 10 separation standard². This standard was not authorised for use within the Bali FIR. Upon issuing the requirement, the controller handed over to another controller and went on a scheduled break.

Over a period of 10 minutes and 37 seconds, the HF radio operator made 12 unsuccessful attempts to contact the crew of IVC. The oncoming controller realised that the HF radio operator had not been able to make contact with the crew of IVC. He also realised that RNP 10 was not an authorised standard. By that time the authorised separation standards had already been infringed. The oncoming controller then successfully instructed the crew of IVC to descend their aircraft immediately to FL320, via a controller pilot data link communication³ (CPDLC) message to the crew of another aircraft.

The controller involved in this occurrence had completed his field training in April 2004. A review of his training records indicated that he had approximately five weeks' experience on that sector of airspace. According to his check and standardisation supervisor, the controller demonstrated a satisfactory level of competency on completion of his sector-specific training.

The controller was rostered to work an afternoon shift, which commenced at approximately 1500 eastern standard time. During the shift the controller had a break of one hour, returning to the console at approximately 1930. The incident occurred at 2038. There was no evidence that fatigue played a part in the incident.

Coordination of high frequency radio communications

The controller issued the requirement to the crew of IVC to reach FL320 by waypoint ATMAP while that aircraft was within reliable very high frequency (VHF) radio range. However, when the controller realised the error in the requirement, that aircraft had passed outside VHF radio range. The controller attempted to issue the amended requirement to the crew of IVC through a HF radio operator, because the crew of IVC had not nominated CPDLC on their flight plan as a means of communication with ATC.

The HF radio operator was unable to establish two-way communication with the crew of IVC, to pass the amended requirement, despite repeated attempts. The controller was not aware that the HF radio operator was unable to issue the instruction to the crew of IVC.

The controller did not confirm with the HF radio operator that the instruction had been passed to the crew of IVC, and there was no published procedure requiring him to do so. Although there was a requirement for the HF radio operator to notify the controller that the instruction was not passed to the crew of IVC, there was no formal procedure to facilitate that notification.

¹ Lateral separation is considered to exist when there is at least a 1 NM buffer between the possible positions of two aircraft (ICAO PANS-ATM, Chapter 5 in CASA Manual of Standards Part 172 10.8.1.1).
² For RNP10, the approval process must show that the total navigation system error in each dimension must not exceed +/- 10 NM for 95 per cent of the flight time on any portion of any single flight:
a) the true position of the aircraft must be within 10 NM of the programmed route centre line; and
b) the true distance to way-points must be within 10 NM of the displayed distance to waypoints.' (International Civil Aviation Organization, 1999, Manual on required navigation performance [RNP], second edition, p. 6).
³ Controller Pilot Data Link Communications (CPDLC): A means of communications between a controller and pilot using text-based messages via an ATC data link (Manual of Air Traffic Services, part 10, effective 10 June 2004).

It was likely that, in lowering the right wing while passing the airstrip and golf course, the pilot 'slipped' the aircraft in order to counter the aircraft tendency to turn. However, the period of any uncoordinated flight could not be determined. Although the quantity of fuel in the selected tank exceeded the '1/4 tanks or less' quoted in the Owner's Manual caution about uncoordinated flight, there was still a risk of fuel starvation and engine stoppage with 1/3 tank capacity. Given the absence of any evidence of pre-accident aircraft defects or engine mis-handling, it is possible that the engine failure was due to uncovering of the right fuel tank outlets and introduction of air into the fuel system.

While the previous power loss event was of interest to the investigation the lack of specific information about that event meant that a link with the engine failure could not be established.

The pilot's response to the engine failure was based on a generic procedure that was inconsistent with the in-flight engine restarting procedures produced by the aircraft manufacturer. Although the investigation could not determine the status of the auxiliary fuel pump wiring, the information available indicated that the LO function of the fuel pump was capable of producing significant fuel flow. It was likely that sustained use of the auxiliary fuel pump, instead of the momentary use specified by the engine manufacturer, provided fuel flow that exceeded the engine's requirements and prevented a restart. Had the pilot referred to the fuel flow gauge, he could have ascertained the amount of fuel being supplied to the engine, and responded accordingly. The pilot's lack of awareness of the manufacturer's procedures could be attributed in part to the absence of the applicable placard and procedure card in the aircraft and the absence of training in type-specific emergency procedures.

The pilot delayed the search for a specific landing site until the aircraft had descended to about 750 ft, because he was initially focussing on restarting the engine. This was primarily due to the pilot's perception that the engine failure was similar to the previous engine power loss event and his expectation that the engine would eventually restart. Although the pilot managed to reach a clear area, an earlier diversion to a specific landing area after the engine failure would have reduced the risk of a forced landing in a less favourable location.

The tailwind during the glide approach, and the limited amount of flap extension had a positive effect on the aircraft's glide range. However, those two factors contributed to a relatively higher touchdown speed, which increased the risk of aircraft damage and occupant injury. Due to the prevailing wind conditions and the track to the landing area, a tailwind during landing was unavoidable. The spring-loaded flap switch combined with high pilot workload made it difficult for the pilot to further extend flap in the late stage of the approach.

On 4 June 2004, the pilot of a Cessna Aircraft Company U206A planned to conduct a private flight from Lakeside Airpark to Proserpine aerodrome and return, a total distance of 40 km. The pilot was the only occupant on the flight to Proserpine which he described as uneventful. Three passengers boarded at Proserpine for the return flight via Laguna Whitsunday Resort, a diversion for sightseeing that added a few minutes flight time.

The pilot described the pre-flight engine run-ups and takeoff at Proserpine as normal and said that he was operating on the right fuel tank as he climbed the aircraft to 1,500 ft above mean sea level. About 4 minutes after departure, the aircraft flew parallel to the resort's airstrip construction site, and then passed the golf course and marina. The pilot reported that, for short periods of less than a minute he had banked the right wing 30 degrees down to optimise the view for his passengers, but he could not recall the extent to which opposite rudder¹ was applied during those times. Shortly after passing the marina, when the aircraft was flying straight and level and was over water at about 1,200 ft, the engine failed.

The pilot selected the right half (coloured yellow) of the auxiliary fuel pump switch to LO and changed the fuel selector position from the right to left tank. There was no response from the engine so he changed the position of the fuel selector a number of times and selected the left half (coloured red) of the fuel pump switch to HI for short periods.

Figure 1: VH-OWA auxiliary fuel pump switch (centre).

The pilot reported that he didn't refer to the fuel flow gauge, but was convinced that fuel wasn't being supplied to the engine. By this stage the altimeter was indicating 750 ft and he began searching for a specific landing area. The pilot glided the aircraft in a north-westerly direction towards a flat area between the marina and the resort golf course. The wind was from the south-east at about 10 kts. He continued with his attempts to restart the engine and managed to transmit a Mayday late in the approach. Flap was not extended beyond 5 degrees and the pilot reported that he was too busy during the final stage of the approach to hold the spring-loaded flap control switch down.

The stall warning activated just before the aircraft landed heavily on a flat area about 20 m before a 1.4 m high embankment. The propeller dug into the bank and the aircraft overturned, resulting in substantial damage. Witnesses and resort staff attended the scene and helped the pilot and passengers out of the aircraft. Emergency services attended from Proserpine and treated the four occupants, who were seriously injured.

The pilot reported that he was not manipulating any engine controls or switches immediately prior to the engine failure and that there were no prior indications of the failure. An extensive examination of the aircraft including the fuel system, ignition system and engine did not reveal any contaminants or defects that would have contributed to the engine failure. The aircraft's fuel tanks contained approximately 50L of fuel per side, which was about 1/3 of each tank's capacity. Laboratory testing of samples from both tanks identified the fuel as aviation gasoline (AVGAS) and did not identify any characteristics that would have contributed to the engine failure.

The Owner's Manual stated that when selected to LO, the auxiliary fuel pump would only operate when the starter was engaged. However, the pilot advised that when priming the engine prior to starter engagement, the auxiliary fuel pump selected to LO consistently produced at least a 12 gal/hr fuel flow. While this indicated that the aircraft's auxiliary fuel pump wiring did not conform to the manufacturer's specifications, damage to the aircraft prevented an assessment of the pump output and exactly how the fuel pump switch was wired.

A Civil Aviation Safety Authority airworthiness directive (AD), issued by the then Australian Civil Aviation Authority in 1979, mandated action in accordance with a service information letter issued by the aircraft manufacturer. The service letter specified provision of a fuel flow stabilisation placard and associated procedure card in that aircraft model. Although the aircraft logbooks indicated compliance in 1985 with the requirements of the AD, neither the placard, nor the procedure card that included in-flight engine restarting procedures, was in the aircraft. The restart procedures specified that the auxiliary fuel pump be selected to ON or HI until the indicated fuel flow was in the green arc, then it should be selected off. The pilot reported that he was not aware of the requirement for the placard and procedure card, nor was he aware of the type-specific in-flight engine restarting procedures. The copy of the owner's manual that was in the aircraft did not include any emergency procedures.

The pilot related that about 8 months prior to the accident the aircraft had sustained a significant power loss while cruising straight and level at 6,500 ft in calm conditions. He had applied the same engine restart procedure that he used during the accident sequence and after a series of engine power fluctuations eventually accomplished a sustained restart at about 3,000 ft. The pilot believed that there had been a vacuum or blockage and the only way he got the engine to run satisfactorily was to rotate the fuel selector between the left and right tanks using a small amount of boost. There was no reason identified for that power loss and the aircraft operated normally for a further 70 hrs. During that time, a periodic inspection was carried out and no aircraft defects that could have contributed to the power loss were identified.

The owner's manual included the following caution in the description of the fuel system.

... with 1/4 tanks or less, prolonged uncoordinated flight such as slips or skids can uncover the fuel tank outlets, causing fuel starvation and engine stoppage. Therefore, with low fuel reserves, do not allow the airplane to remain in uncoordinated flight for periods in excess of 1 minute.

The ATSB recently completed an investigation into an engine failure resulting from fuel starvation that involved a similar aircraft type (Cessna 207, ATSB report 200403210). That investigation found that the in-flight engine restart procedures published by the manufacturer were not followed, but the engine restarted after a significant height loss of about 700 ft.

¹ Application of rudder in the opposite direction to a lowered wing inhibits the development of a turn, resulting in a slip that can allow flight on a straight track. An aircraft that is slipped is considered to be in uncoordinated flight.

3 Conclusions

3.1 Findings

• There was no witness or mechanical evidence to indicate any problems with the flight controls, fuel system or engine prior to the event.
• Failure of the forward flex plate resulted in drive to the main rotor gearbox being partially disconnected.
• The failed forward flex plate coupling punctured one or both the helicopter fuel tanks and the stainless steal engine compartment firewall.
• The helicopter settled heavily and impacted with terrain.
• The helicopter was destroyed by impact forces and a post-impact fire.
• Egress from the helicopter was hampered by the unusual attitude of the cabin which caused extended exposure of the survivor to the post-impact fire.

3.2 Significant factors

One of the two bolted joints linking the flex plate to the main rotor gearbox yoke in the forward flexible coupling was previously assembled incorrectly, resulting in a lack of clamping force and subsequent fatigue failure of the flex plate and loss of drive to the main rotor gearbox.

2 ANALYSIS

2.1 Helicopter

The initiating event in the occurrence sequence was the failure of the flex plate in the forward flexible coupling. That event was the source of the loud noise that the passenger reported hearing. Once the plate failed, the clutch shaft yoke was retained at one end only.  As a result, the diameter of the yoke and flex plate effectively doubled, allowing those rotating components to contact and puncture the fuel tanks and the horizontal stainless steel fire wall, as observed in the wreckage examination. Fuel leaking from the punctured tanks and firewall, would have fed directly into the engine compartment. The evidence indicates that the fuel then contacted either a hot engine component or some other source of ignition, which could have resulted in the helicopter being on fire before impact. The soot pattern on the ground confirmed that fuel from the punctured tanks was being ejected from the helicopter at first impact with the ground. The fuel may have ignited while the helicopter was still airborne or flash back could have occurred after impact.

One of the bolted joints in the forward flexible coupling had been assembled incorrectly, resulting in a lack of clamping force and subsequent fatigue failure of the flex plate in the forward flexible coupling and loss of drive to the main rotor gearbox.

The build records and digital photographs supplied by the manufacturer showed that the assembly of this joint at the factory had been correct.

Examination of the maintenance history of the helicopter revealed that, while maintenance had been performed on the drive system on a number of occasions since factory assembly, there was no recorded documentary evidence found that any subsequent maintenance had been performed on the particular bolted joint where the failure occurred. The stated procedure employed by the engineer who carried out AD/R22/51 was appropriate and consistent with accepted practices and did not require disassembly of the bolted joint that failed.

Examination confirmed that both tail rotor blades failed as a result of contact with the ground during the impact sequence.

At the time of the incident the helicopter was being operated close to the ground with low forward air speed. That would have left the pilot with little time to respond effectively to failure of a critical flight system such as the main rotor drive system.

The manner in which the failure of the forward flex plate occurred resulted in an unusual emergency situation for the pilot. As well as the obvious noise and out of balance caused by the still rotating but partially separated forward flexible coupling, symptoms would have included a failure of the drive to the main rotor gearbox resulting in decreasing main rotor speed and a nose-left yaw. That was consistent with some of the symptoms normally associated with an engine failure. However, in this case the engine had not failed and initially would have presented the pilot with additional and potentially confusing symptoms. Those symptoms would have included an initial indication of engine overspeed, and continuing tail rotor drive.

The normal response to an engine failure, and also relevant in this case, would be for the pilot to lower the collective control in order to recover any loss of main rotor RPM. The pilot would then allow the helicopter to enter an autorotation descent. If there was insufficient height for the descent to stabilise, main rotor RPM would not have recovered before the pilot commenced the touchdown phase of the autorotation. Low main rotor RPM during the touchdown phase would have decreased the possibility for the pilot to reduce the helicopter's rate of descent and carry out a safe touchdown. It is evident that fatigue cracking in the flexplate initiated at the bolthole and propagated under the washers of the bolted joint towards the edge of flexplate. It was also evident from the fracture surface features that crack growth had occurred over a number of flights prior to the accident flight. Because cracking occurred under the washers in the bolted joint, the opportunity to detect cracking by a pre-flight visual inspection would have been limited to the detection of a crack at the edge of the flexplate arm.

The reliability of a general visual inspection (for example, a pre-flight inspection) is affected by lighting, the ability to get close to the component (proximity), and dirt and dust from the operating environment. Reliable detection of a specific defect would require a directed detailed inspection.

2.2 Survival

The crush damage observed to the right side of the cabin structure showed that impact forces were greater on the right side of the helicopter than the left. As a result, the pilot, who occupied the right seat, was subjected to higher impact forces than the passenger in the left seat. The impact forces in this accident were not directly aligned with the helicopters fore-aft and vertical axes. Hence, the level of protection afforded by the seat/harness combination was reduced and probably contributed to the injuries sustained by the pilot. The extent of the post-impact fire precluded any definite conclusions as to what, if any, material may have been carried under the pilot's seat.

When the helicopter came to rest on its right side, the seat in which the passenger was trapped was in an elevated position and was severely affected by the fire. It was also likely that the difficulty the passenger experienced in escaping from the helicopter was due, at least in part, to him being suspended by the safety harness. The time it took the passenger to release himself contributed to the extent of his burn injuries.

The passenger, although seriously injured himself, walked approximately 8 km to summon help. Unfortunately, due to the accident site's remote location, the pilot succumbed to his injuries before medical assistance could reach him.

2.3 Communication equipment

No record of fitment of a fixed Emergency Locator Transmitter (ELT) could be found in the logbooks for the accident helicopter and no ELT unit remnants for either a fixed or personal unit were identified within the main wreckage. Given the intensity of the post-impact fire, it was probable that, had a fixed ELT been fitted, it would have only operated for a very short period of time before being destroyed.

If the pilot was carrying a personal ELT at the time, it is possible that it separated from his person either during the impact or during egress from the wreckage and was destroyed in the subsequent fire.  In addition, if the occupants carried a satellite telephone, it may have reduced the time taken to summon assistance.

1 FACTUAL INFORMATION

1.1 Sequence of events1

At about 1100 Central Standard Time on 30 May 2004, a Robinson R22 Mariner II helicopter, registered VH-MIB, crashed and caught fire while being operated on a fence inspection flight at Tobermorey Station, NT. The pilot sustained fatal injuries, and the passenger was seriously injured.
The passenger reported that the purpose of the flight was to inspect a fence line bordering the property and then to conduct cattle mustering operations. The helicopter was refuelled to full tanks prior to departure. The weather was fine and sunny, with a slight breeze from the south-east.
The passenger reported that while overflying a section of the fence line about 45 km south of the homestead at 30 to 40 ft above ground level, the pilot initiated a turn back to the left to enable a closer look at a particular section of fence. Part way through the turn there was a loud bang from behind the cabin, followed by "horrendous vibration" and the helicopter immediately began to yaw left and descend. The ground marks showed that the tail rotor blades contacted the ground first and then the forward section of the helicopter's right skid struck the ground, disrupting the front section of the cabin. The helicopter then came to rest on its right side and fire rapidly spread to engulf the cabin area.

The pilot was able to free himself from the helicopter through the broken front section of the cabin, but the passenger experienced difficulty undoing his safety harness and remained trapped. He was eventually able to free himself and joined the pilot at some distance from the wreckage. The passenger reported that he assessed the pilot to be badly injured and directed him to a nearby water hole. The passenger then walked to a water bore approximately 8 km from the accident site, where he met other station personnel. The pilot was deceased when medical assistance arrived at the accident site some hours later.

The passenger reported that he had flown in the helicopter several times. During the last few flights before the accident, and during the accident flight itself, he had detected a vibration that he considered abnormal. The passenger advised the investigation that he had conveyed these concerns to the pilot, who advised that he conducted a good check of the helicopter and was satisfied that there were no problems.

1.2 Wreckage examination

Examination of the wreckage by the Australian Transport Safety Bureau (ATSB) investigation team at the accident site (Figure 1) confirmed that the helicopter was yawing left and moving forward when it struck the ground. The tail rotor blades contacted the ground first. The helicopter then contacted the ground slightly nose down and heavily onto the right skid, causing it to separate from the helicopter. The ground impact marks showed that the helicopter continued to yaw left though about 180 degrees after it struck the ground, before coming to rest on its right side. The majority of the cabin and engine bay, including the entire floor area and cabin structure beneath the pilot and passenger seat positions were destroyed by fire.

Figure 1: Helicopter wreckage (arrow indicates bulkhead deformation behind pilot seat position)

The extent of fire damage meant that a complete examination of the helicopter was not possible. Some aluminium components such as tubing in the flight control system had been destroyed. However, all steel components in the control systems for the main and tail rotor were identified and damage to all of these components was consistent with impact forces or fire. The main rotor blades exhibited damage consistent with low rotor energy at impact. Both tail rotor blades had fractured approximately 1/4 span outboard of the rotor hub centre drive (Figure 2). The failed section of one blade was found adjacent to the main wreckage. The failed section of the other blade was found subsequently about 70 m from the main wreckage. Both blade sections were taken to the ATSB laboratories for further examination.

Figure 2: Tail rotor damage

All of the engine drive system components were identified within the wreckage. The flex plate for the forward flexible coupling of the main rotor drive system was fractured at one of the two attachment points to the main rotor gearbox yoke (Figure 3 and 4). The flex plate, including the clutch shaft, were retrieved from the accident site, for further examination. The flex plate for the intermediate flexible coupling was intact and showed evidence of rotational damage consistent with partial drive system power at impact. The rear flex plate and coupling components were also found intact.

Figure 3: Forward flexible coupling as found in wreckage

Figure 4: Components of failed forward flexible coupling

The fuel system and engine ancillaries were destroyed by the fire. There were vertical cuts puncturing the inside wall of the right fuel tank and the horizontal stainless-steel firewall above the engine. The cuts in the right tank and the firewall aligned with the forward flex plate plane of rotation (Figure 5). The left fuel tank was destroyed by fire.

Figure 5:  Flex plate puncture of stainless-steel firewall above engine.

1.3 Personnel information

The pilot held a commercial pilot (helicopter) licence and was appropriately endorsed on the R22. He was issued with a private pilot (aeroplane) licence in 1974 and a commercial pilot (helicopter) licence in 1990. The pilot was issued with a commercial pilot (aeroplane) licence in 1995. He held a stock mustering rating and a valid class 1 medical certificate. He completed a flight review in the occurrence helicopter on 19 April 2004. At the time of the accident, the pilot had approximately 10,400 hours aeronautical experience. He flew 31 hours in the occurrence helicopter between 17 and 30 May 2004.

1.4 Medical and pathological information

Postmortem and pathology reports did not indicate that the pilot was suffering from any condition that might have affected his performance during the flight. The most significant injuries sustained by the pilot were the result of impact forces rather than fire.

1.5 Fire

There was a fire affected area (sooting) on the ground that extended up-slope from the wreckage (Figure 6). The sooting formed a swirl pattern of decreasing radius in the direction the helicopter was yawing when it contacted the ground.

Figure 6:  Sooting pattern adjacent to wreckage (arrow indicates approximate direction of flight at impact)

1.6 Survival aspects

Three-point lap/sash type safety harnesses were fitted to both seating positions in the helicopter. The passenger reported that both he and the pilot had their harnesses fastened during the flight. Fire damage precluded a detailed assessment of the seats and performance of the crush zones beneath them as well as the seat belt harnesses.

Severe crush damage to the lower cabin bulkhead was evident immediately behind the pilot's seat. (Figure 1).

Following the accident, no Emergency Locator Transmitter signal was received (refer section 1.7.6). There was no mobile telephone coverage in the area and the passenger reported that they did not carry any other communications aids, such as a portable satellite telephone.

1.7 Helicopter information

1.7.1 Helicopter data

The helicopter was manufactured in August 2002 as Serial No 3357M. The most recent maintenance release for the helicopter could not be located. It was reported to have been kept in the helicopter. If so, the maintenance release would have been destroyed in the post-impact fire. Based on other maintenance records and information contained in the pilot's personal diary, the total time in service of the helicopter on 30 May 2004 was estimated to have been 506 hours.

1.7.1 Main and tail rotor drive system

In the R22 helicopter, power to drive the main and tail rotors was transmitted from the engine to the rotor drive train via a multiple Vee belt drive and clutch system. A shaft transmitted power forward from the clutch to the main rotor gearbox and aft to the tail rotor gearbox (Figure 7).

Figure 7:  Main components of main and tail rotor drive systems²

A forward flexible coupling, which includes a flex plate, connected the drive shaft to the main rotor gearbox.  The tail rotor drive system also included an intermediate and a rear flexible coupling. Yoke assemblies at the end of each drive shaft section connected the shaft to the flex plate via bolted joints (Figure 8). The purpose of these flex plates was to accommodate small differences in shaft axial alignment during drive shaft rotation. The flex plates and the bolted joints were critical elements in drive system integrity.

Figure 8:  Components of the forward flexible coupling

The helicopter manufacturer published procedures for assembling and aligning the drive system components, including the allowed tolerances. The design loads of components could be exceeded if those tolerances were not met.

1.7.3 Helicopter manufacture

The helicopter arrived in Australia partly disassembled for ease of shipment. The main drive system components within the engine compartment were assembled during manufacture and not subsequently disturbed for this method of international shipment.

In order to assist the investigation, the manufacturer supplied the investigation with the itemised build records, which included digital photographs of the rotor drive system, for the occurrence helicopter. Those records were reviewed as part of the laboratory analysis of the flexible coupling failure. The review of the build records and photographs for the forward flexible coupling in the occurrence helicopter showed that NAS6605-6 bolts had been used, and a spacer washer had been included in each bolted joint and one thin washer had been installed under the nut, with a palnut (locking nut) fitted to each. The build records showed that the build-up of the bolted joint at the time of manufacture of the helicopter was correct and in accordance with the assembly procedures.

The 'Daily or Preflight Checks' section 4-2 and 4-3 of the manufacturers Pilot's Operating Handbook identified the requirement for a visual check of the flex coupling to ensure there are 'No cracks and 'Nuts tight'. Also required is a check of the yoke flanges for cracking. It further advised, in part, that 'During the following inspection, check the general condition of the aircraft and also look for any evidence of leakage, discolouration due to heat, dents, chafing, galling, nicks, corrosion and especially for cracks. Also check for fretting at seams where parts are joined together. Fretting of aluminium parts produce a fine black powder, while steel produces a reddish brown or black residue'.

1.7.4 Maintenance history

Maintenance records indicated that an Australian certificate of airworthiness for the helicopter was issued on 11 October 2002, after assembly in Australia, following manufacture and acceptance flights in the US.  At that time, the total time in service was 5.1 hours. A summary of subsequent maintenance conducted on the helicopter is as follows. All references to drive system adjustments and/or maintenance have been included.

• 11 March 2003. Total time in service 55.1 hours.  50 hourly engine inspection
• 27 March 2003. Total time in service 98.5 hours.  100 hourly inspection.  Maintenance carried out included adjustment of the engine sheave alignment.³
• 30 June 2003. Total time in service 198.2 hours.  100 hourly inspection.  Maintenance carried out included checking and adjustment of the engine sheave alignment and intermediate flex plate shimming to within limits.
• 15 September 2003.  Total time in service 296 hours.  100 hourly inspection.
• 17 February 2004. Total time in service 384.1 hours.  Civil Aviation Safety Authority Airworthiness Directive (AD) R22/51 'Main Rotor Clutch Shaft', dated 12 November 2003, was incorporated. AD/R22/51 was applicable to all R22 helicopters. It required disassembly of the main rotor yoke (A907) to the clutch shaft joint (A166) (see Figure 6) and inspection of the shaft and yoke for damage including fretting⁴ of bolt holes, cracking in the area of the bolt holes, and the presence of an unapproved jointing compound in the mating surfaces. The helicopter maintenance worksheet indicated that no fretting was evident but that the incorrect jointing compound had been used. The worksheet stated that the AD had been complied with and that the clutch shaft and yoke were reassembled in accordance with the maintenance manual. The worksheet also recorded that a duplicate inspection of the clutch shaft installation and the yoke (A907) assembly had been performed. The licensed aircraft maintenance engineer who carried out the AD reported that he disconnected the yoke (A907) from the forward flex plate, but did not disconnect the flex plate from the main rotor gear box yoke (A908). He stated that he did not perform any maintenance on the bolted joints at the connection between the main rotor gear box yoke and the flexible coupling.
• 27 March 2004. Total time in service 396 hours.  100 hourly inspections.  Maintenance carried out included engine sheave alignment.
• 12 May 2004. Total time in service 476.1 hours.  100 hourly inspections.

The documentation showed that maintenance had been performed on 27 March 2003, 30 June 2003, and 17 February 2004 in the vicinity of the forward flexible coupling that, while it did not specifically necessitate bolt removal, provide opportunities for the forward flexible coupling bolts to be disturbed.

1.7.5 Forward flex plate bolted joint component specification

A review of the diagrams contained in the manufacturer's Maintenance Manual and the Illustrated Parts Catalogue (IPC) revealed a difference in the specifications of the parts in the bolted joints. Notes contained within the IPC explained that bolts of different grip lengths and washers of different thickness were to be used in the flex plate bolted joints to expose between two and four threads beyond the end of the nut. A table comparing the different specifications between the Maintenance Manual table and the IPC is provided in section 4.1.3 of the ATSB technical analysis report attached as Appendix 1.

The manufacturer advised that the bolt length identified in the Maintenance Manual was for use in an earlier version of the manual and was out of date. Corrective action to update this information was scheduled by the company for November 2005, but at the time of writing of this report had not been accomplished.

The manufacturer advised that the company did not publish any warning to maintenance organizations about the discrepancy in bolt length between the Maintenance Manual and the Illustrated Parts Catalogue.  The discrepancy was not considered by the manufacturer to be critical in that the use of either a NAS6605-5 or a NAS6605-6 bolt with the appropriate combination of spacer and washers would give the correct clamp up for proper joint integrity. In the few cases where the -5 bolt did not allow proper installation of the B330-16 palnut, the problem would be self-evident. The manufacturer believed that any engineer performing the installation where the bolt was too short to install a palnut would install a longer bolt or make inquiries to resolve the problem.

1.7.6 Emergency locator transmitter

The maintenance records indicate that the helicopter was imported from the US and subsequently operated by various owners without a fixed Emergency Locator Transmitter (ELT) unit being fitted. This fact was noted on the maintenance releases issued at 5.1 airframe hours total time in service (TTIS) on 11 October 2002 and 98.5 hours TTIS on 27 May 2003, which required the pilot to observe the requirements of CAR 252 and carry a personal ELT. No further entries of this nature were found on maintenance releases issued after this date, nor could evidence be found in the aircraft logbooks that an ELT had been fitted. A search for both a fixed and personal ELT within the wreckage and surrounding accident site was conducted but nothing was found.
The passenger stated that the pilot normally carried a personal ELT on him. No personal ELT was identified among the pilots clothing or personal effects and no emergency signal was received by AusSAR from that location on the day.

1.8 Specialist examination of the failed components

The forward flex plate and the broken sections of the tail rotor blades were subject to detailed examination by the ATSB. The report on those examinations and analysis of the failures is attached as Appendix 1.

The metallurgical evidence confirmed that the failure mode of both tail rotor blades was very similar and was the result of contact with the ground during the impact sequence. The rocky material embedded in the blade tips provided clear evidence that the blades had struck the ground while rotating. The blade that was found about 70 m from the wreckage was thrown that distance as the result of tail rotor rotational energy.

The specialist examination found that the flex plate in the forward flexible coupling fractured as a result of the propagation of a fatigue crack at one of the bolted connections between the plate and main rotor gearbox yoke. Final fracture of the flex plate occurred during operation and not as a result of the collision with the ground. No crack growth or wear damage was observed at the three remaining boltholes. Examination of the bolt installed at the failure location revealed that extensive fretting wear had occurred around the entire circumference of the bolt, in the region adjacent to the flex plate and the regions adjacent to the reinforcing plates. Fretting wear was also evident on the washer surface adjacent to the bolt head.

The bolted joint at the flex plate failure location was found to have a single thin washer under the bolt head and nut, and no spacer washer between the yoke and flex plate. This spacer and washer combination was different from that specified by the manufacturer for use with a NAS6605-6 bolt.

1. Only those investigation areas identified by the headings and subheadings were considered to be relevant to the circumstances of the accident.
2. Diagram with permission of Robinson Helicopter Company.
3. Drive Vee belts sometimes stretch when new and adjustments are then necessary to maintain the correct drive system alignment. Engine sheave alignment is part of that adjustment process.
4. The AD defined major fretting as 'any evidence of the machining marks in any of the bolt holes being partly or fully obliterated'.

The crew of the Robinson R22 helicopter were undertaking a fence line inspection at about 30 to 40 ft above ground level. The crew had initiated a turn back along the fence line for a closer look at a particular section of fence. During the turn, a loud bang was heard, and the helicopter began to rotate quickly before striking the ground.

Both occupants were able to exit the helicopter unaided after it came to rest but sustained serious injuries and burns as a result of a post-impact fire. The pilot subsequently died of his injuries.

The investigation found that one of the bolted joints linking the forward flexible coupling flex plate to the main rotor gearbox drive shaft yoke had been assembled incorrectly. This resulted in subsequent fatigue failure of the flex plate and loss of drive to the main gearbox. Control of the helicopter was then lost at a height from which it was difficult to recover.

The crew of the Robinson R22 helicopter were undertaking a fence line inspection at about 30 to 40 ft above ground level. The crew had initiated a turn back to the left, along the fence line for a closer look at a particular section of fence. During the turn, a loud bang was heard, and the helicopter began to yaw quickly before striking the ground.

Both occupants were able to exit the helicopter unaided after it came to rest but sustained serious injuries and burns as a result of a post-impact fire. The pilot subsequently died of his injuries.

Examination of the helicopter wreckage revealed that both tail rotor blades had failed due to contact with the ground.  In addition, the flex plate in the forward flexible coupling of the main rotor drive was found fractured at one of the two attachment points to the main rotor gearbox yoke. The tail rotor blades and several components from the main rotor drive were recovered for detailed analysis in order to resolve the mechanism of fracture and the sequence of failure.

The flex plate in the forward flexible coupling fractured due to the propagation of a fatigue crack at one of the bolted connections between the plate and main rotor gearbox yoke. Final fracture of the flex plate occurred during operation and not because of the collision with the ground. There was no crack growth or wear damage evidence at the three remaining boltholes. Examination of the bolt installed at the failure location revealed that extensive fretting wear had occurred around the entire circumference of the bolt, in the region adjacent to the flex plate and the regions adjacent to the reinforcing washers. Fretting wear was also evident on the washer surface adjacent to the bolt head. This type and degree of wear damage was indicative of operation with insufficient clamping force in the bolted joint.

A review of the manufacturer's original build-sheets for the forward flexible coupling in the occurrence helicopter revealed that NAS 6605-6 bolts were used, and a spacer washer had been included in each bolted joint and one thin washer had been installed under each nut with a palnut (locking nut) fitted to each.

Examination of the forward flexible coupling retrieved from the accident site, found that the bolted joint had been assembled with a washer and spacer combination that was different from that identified by the manufacturer's original build records. These differences indicated that it was likely that the joint had been disassembled and reassembled during a maintenance action subsequent to assembly in the manufacturer's facility.