Description

On 13 October 2011, a Bombardier DHC8-100 (P2-MCJ) being operated by Airlines PNG on a scheduled passenger flight from Nadzab to Madang in day Visual Meteorological Conditions (VMC) on an Instrument Flight Rules (IFR) flight plan lost all engine power en route at 10000 feet amsl following a rapid reduction in power in response to an overspeed warning. A 'MAYDAY' was declared and the aircraft was manoeuvred for a forced landing which was accomplished on relatively flat terrain but concealed boulders resulted in the tail and one wing separating from the fuselage and a severe fuel-fed fire started. The aircraft was destroyed and all but one of the 29 passengers died in the fire. The remaining passenger, the flight attendant and both pilots survived, the passenger with severe burns and the others with relatively minor injuries.

Investigation

An Investigation was carried out by the Papua New Guinea Accident Investigation Commission. Recorded data relevant to the Investigation from the Flight Data Recorder (FDR) and Cockpit Voice Recorder (CVR), which were both located in the tail section of the aircraft and not damaged by the impact or the post-crash fire, was successfully downloaded by the Australian Transport Safety Bureau (ATSB).

It was established that although the 64 year old aircraft commander had "extensive prior experience of flying in Papua New Guinea", he had gained only 500 of his 18000 hours on the accident aircraft type having qualified on it 8 months previously. The 40 year old First Officer had 391 hours experience on the DHC8. The DHC8-100 aircraft type had been operated by Airlines PNG since 2003. Pilot training at Airlines PNG was found not to require that the proficiency in the propeller overspeed emergency procedure be demonstrated. It was established that after a 10 minute cruise at 16000 feet during which convective activity had been avoided laterally whilst crossing high terrain, a descent had begun in order to be able to position visually towards Madang and avoid thunderstorm activity which was reported to be in its vicinity. Just under three minutes after the descent from 16,000 feet had begun and with the aircraft passing 10,500 feet at approximately 4,200 fpm in VMC, the aircraft reached VMO and an overspeed warning began to sound. Neither pilot had noticed the increasing airspeed. The PF raised the nose of the aircraft in response to the warning and this reduced the rate of descent to about 2000 fpm but the overspeed warning continued, so he retarded the power levers back "quickly" and shortly after this both propellers oversped simultaneously, exceeding their maximum permitted speed of 1,200 RPM by over 60 % and seriously damaging the left hand engine and rendering both engines unusable.

Villagers on the ground reported hearing a loud „bang‟ as the aircraft passed overhead. The noise in the fight deck became "deafening", rendering communication between the pilots extremely difficult, and internal damage to the engines caused smoke to enter the flight deck and passenger cabin through the air-conditioning system.

The sudden emergency situation "caught both pilots by surprise". "Confusion and shock on the flight deck (was) compounded by the extremely loud noise from the overspeeding propellers" and it was clear that the aircraft commander was, at least initially, unaware what had happened. However, the consequences were soon recognised and both pilots agreed that there was 'nothing' left in terms of useable engine power. At this point, about 40 seconds after the propeller overspeeds began, the left propeller was windmilling and the left engine was no longer producing any power because of the damage caused to it by the overspeed. The right engine was operating at flight idle, although it was not possible to unfeather it and so it could not produce any thrust.

FDR data indicated that when the propeller overspeeds had begun, both engine torque indications had dropped to zero and had remained there for the rest of the flight. It was also found that both propellers had increased in a few seconds to over 60 % above the maximum permitted propeller rpm.

The aircraft commander instructed the First Officer to transmit a MAYDAY call to Madang TWR and this was done. It was found that emergency checklists and procedures had not been followed and the crew had directed their attention instead to where they were going to make a forced landing. The descent continued at a high rate with the windmilling left propeller creating extra drag and the asymmetry between it and the feathered right propeller making the aircraft difficult to control. It was calculated from the recorded data that the rate of descent between the onset of the emergency and arrival at the crash site had averaged 2,500 fpm and had at one point briefly exceeded 6000 fpm accompanied by a recurrence of the earlier overspeed warning which had led indirectly to the emergency. It was calculated that optimum handling at this time could have extended the airborne time following the initial propeller overspeed from approximately 4 minutes to approximately 10 minutes.

As the chosen landing area was approached, the First Officer asked the commander if he should shut both engines down and the latter replied that he should shut "everything" down. Approximately 800 feet above ground level and 72 seconds before impact, the left propeller was feathered and both engines were shut down. The commander stated afterwards that, having initially decided to make the landing on the river bed, he changed his mind when he realised that it contained large boulders. The alternative area then used beside the river bed also contained boulders but they were beneath the vegetation and thus not readily visible from the air.

The aircraft impacted terrain tail first at a relatively shallow angle of approximately 5 degrees at 114 knots with the flaps and the landing gear both retracted. The distance travelled from the first ground impact to the main wreckage site was about 200 metres. During the impact sequence, the left wing and tail became detached and the front of the aircraft fractured behind the flight deck and rotated through 180 degrees, so that it was inverted when it came to rest.

The available evidence indicated that a post-impact fuel-fed fire had began during the impact sequence when the aircraft was about halfway along the wreckage trail. After it finally came to a stop, the fire appeared to have completely consumed much of the wreckage with the exception of the tail section. On the basis of the fuel uplifted at Nadzab, it was estimated that there would have been approximately 1820 litres of fuel on board the aircraft.

The flight deck section of the fuselage had ended up furthest away from the fuel tanks - the source of the post impact fire and had partially separated from the fuselage and had rotated so that it was upside down, blocking the ceiling-mounted cockpit emergency exit. According to the pilots, there was a gap in the side of the fuselage large enough for them and the Flight Attendant to crawl through. When they emerged from the aircraft, they reported noticing that one of the passengers - the only one to survive - was already outside the wreckage. He stated the smoke and fire were very intense and that he had escaped through a gap in the roof above his seat position. It was clear that the other passengers had had very little time to evacuate before the cabin was completely engulfed by the fire. However, the bodies of two other passengers were found located outside the fuselage on the right hand side adjacent to the row 4 emergency exit

Various aspects of the emergency were reviewed:

• In the absence of any identifiable mechanical component failures, movement of the power levers behind the flight idle gate by the Captain is considered to be the only plausible explanation for the simultaneous double propeller overspeed. It was noted that the AFM included an explicit prohibition on the in-flight operation of the power levers aft of the Flight Idle Gate stating that such operation "will cause propeller overspeed, possible engine failure and may result in loss of control".

• Although it is not possible to exclude alternative explanations completely, the evidence suggests the Pilot-in-Command reacted spontaneously to the VMO overspeed warning and that the action of moving the power levers below the Flight Idle gate - facilitated by removing the Flight Idle gate release triggers as done on every flight once on the ground - was probably unintentional.

• The uncomanded feathering of the right engine propeller occurred because of a malfunction in the PCU beta switch system. Had it been possible to unfeather it, it is likely that useful power may have been available. However, expert knowledge of the propeller control system - beyond that which any pilot could be expected to possess – would have been needed to achieve such an unfeathering.

• The malfunction of the PCU beta right engine switch was a fault which could not have remained undetected in normal operation because the propeller would feather immediately the power levers were moved from the ground range to flight range prior to take off. It was concluded that the switch "probably stuck in the closed position as a consequence of the propeller overspeeds, perhaps because of propeller vibration". It was noted that this fault had been linked to incorrect application of installation procedures during routine overhaul and that the PCU involved would have eventually been recalled for checks. However, the extensive thermal damage to the unit precluded any examination or testing to determine the cause of the switch malfunction and it was therefore impossible to exclude this origin as a potential cause for the sticking switch.

• During the propeller overspeed, the beta warning horn, designed to ensure pilots are immediately aware if one or both power levers are inadvertently retarded into the prohibited ground range in flight, did not initially sound, and when it did begin do so, did this only intermittently. If the beta warning horn had functioned normally, the pilots may have recognised what was happening straight away and been able to take action quickly enough to prevent the propeller overspeeds from damaging the engines. However since the commander said afterwards that he "was not sure he had ever heard the beta warning horn before the day of the accident", he may not have realised what it signified even if it had sounded as soon as he lifted one or both of the power lever triggers.

• Some time after the accident, Bombardier identified a problem with the beta warning horn and issued an SB requesting customer feedback. Five of the 91 aircraft covered by these responses were found to have faulty beta warning horn systems and the problem was attributed to "worn micro-switch retaining brackets". Because of the damage caused by the crash landing and post-impact fire, it was not possible to determine why the beta warning horn had malfunctioned on the accident flight.

• One reason the crew gave for not actioning any formal emergency response procedures was that there was insufficient time to do this. However, given the time actually available, it was considered that "their perception of time during the emergency may have been affected by stress".

The Investigation Findings included the following:

• The aircraft commander moved both power levers rearwards below the flight idle gate and into the beta (ground only) range shortly after Vmo was exceeded and the overspeed warning had occurred. This action was only possible after the release latches at the minimum power setting permitted in the air had first been lifted. It led directly to a loss of propeller speed control, to the double propeller overspeed and a loss of forward thrust to the extent that a crash landing became inevitable.
• Prior to the overspeed warning, the aircraft commander had allowed the rate of descent to increase to 4200 fpm.
• The beta warning horn malfunctioned and did not sound immediately when one or both of the flight idle gate release triggers were lifted. When the beta warning horn did sound, it did so intermittently and only after the double propeller overspeed had begun. However, the sound of the beta warning horn was likely to have been masked by the noise accompanying the propeller overspeeds.
• There was an uncommanded feathering of the right hand propeller after the overspeed commenced due to a malfunction within the propeller control beta backup system during the initial stages of the overspeed of that propeller. The MRO where the right hand Propeller Control Unit (PCU) had last been overhauled was found to have incorrectly applied beta switch reassembly procedures specified in an SB relating to the PCU. This quality lapse affected a considerable number of PCUs which had all been identified as subject to recall; the one fitted to the accident aircraft was due for recall which would have occurred had it remained in service. Although FDR data indicated that the PCU fitted to the right hand engine had caused an uncommanded feather, destruction of the PCU in the post crash fire meant that it was not possible to definitely attribute this to the effects of the quality lapse.

• The landing gear and flaps remained retracted during the crash landing which led to a higher landing speed than could have been achieved if the gear and flaps had been extended as well as to increased impact forces on the airframe and its occupants.
• None of the prescribed Quick Reference Handbook (QRH) procedures or their checklists were used by the flight crew after the emergency began and the left engine propeller was not feathered after the engine had failed.

Subsequent to the accident, it was noted that Bombardier had identified a problem in the beta warning horn system which may have led to faults in it not being identified during routine testing of the system.

Three Safety issues were identified by the Investigation as follows:

• A significant number of DHC-8-100, -200, and -300 series aircraft did not have a means of preventing movement - whether intentional or unintentional -of the power levers below the flight idle gate in flight, nor a means to prevent such movement resulting in a loss of propeller speed control.
• The aircraft manufacturer identified a problem in the beta warning horn system that left the system susceptible to failures that may not have been identified during regular and periodic tests of the system.
• After the accident, the facility that overhauled the Propeller Control Unit (PCU) installed on (the right hand engine) at the time of the accident identified a quality (lapse) relating to the use of incorrect reassembly procedures for the installation of the beta switch within the propeller control unit. (This) quality (lapse) may have led to uncommanded (feathering) of the right propeller.

Safety Action taken as a result of the accident findings, and in particular the Safety Issues identified was noted to have been taken by Bombardier, Transport Canada, the PNG CAA, the operator Airlines PNG and Pacific Propeller International, the propeller overhaul facility referred to in one of those Safety Issues.

The Final Report was made publicly available on 15 June 2014. In the light of Safety Action taken during course of the Investigation, no Safety Recommendations were made.

Further Reading

• Loss of Control
• Dangers of Flight with Power Below Flight Idle
• Emergency and Abnormal Checklist
• Situational Awareness
• Technical Knowledge
• Crew Resource Management

Similar event(s) on SKYbrary:

• DH8A, En route near Sørkjosen Norway, 2006 (LOC HF AW): On 21 February 2006, a DHC8-100 on a passenger flight from Tromsø to Sørkjosen experienced a temporary loss of control during descent in night IMC when the power levers were inadvertently selected to a position aft of the Flight Idle gate and propeller overspeed and engine malfunction followed. After recovery and shut down of the right engine, a return to Tromsø was made using the remaining engine without further event.