B733, vicinity Kosrae Micronesia, 2015

B733, vicinity Kosrae Micronesia, 2015

Summary

On 12 June 2015, a Boeing 737-300 crew forgot to set QNH before commencing a night non-precision approach to Kosrae which was then flown using an over-reading altimeter. EGPWS Alerts occurred due to this mis-setting but were initially assessed as false. The third of these occurred when the eventual go-around was initially misflown and descent to within 200 feet of the sea occurred before climbing. The Investigation noted failure to action the approach checklist, the absence of ATC support and the step-down profile promulgated for the NDB/DME procedure flown as well as the potential effect of fatigue on the Captain.

Event Details
When
12/06/2015
Event Type
CFIT, HF, LOC
Day/Night
Night
Flight Conditions
IMC
Flight Details
Operator
Type of Flight
Public Transport (Passenger)
Intended Destination
Take-off Commenced
Yes
Flight Airborne
Yes
Flight Completed
Yes
Phase of Flight
Missed Approach
Location
Location - Airport
Airport
General
Tag(s)
Approach not stabilised, Altimeter Setting Error, Inadequate Aircraft Operator Procedures, Non Precision Approach
CFIT
Tag(s)
No Visual Reference, Vertical navigation error
HF
Tag(s)
Fatigue, Ineffective Monitoring, Procedural non compliance, Ineffective Monitoring - PIC as PF
LOC
Tag(s)
Flight Management Error, Environmental Factors
Outcome
Damage or injury
Yes
Non-aircraft damage
Yes
Non-occupant Casualties
No
Number of Non-occupant Fatalities
0
Number of Occupant Fatalities
0
Off Airport Landing
Yes
Ditching
Yes
Causal Factor Group(s)
Group(s)
Aircraft Operation
Safety Recommendation(s)
Group(s)
None Made
Investigation Type
Type
Independent

Description

On 12 June 2015, a Nauru Airlines Boeing 737-300 (VH-NLK) being operated on a scheduled international passenger flight from the Marshall Islands to Kosrae and making a non-precision offset approach to runway 05 at destination in night IMC received successive EGPWS terrain proximity alerts. These were initially assessed as false but a go around was eventually commenced due to lack of the visual reference required to commence circling but insufficient thrust was applied and descent to within 200 feet of the sea occurred accompanied by a further EGPWS terrain proximity alert. A climb was then established, the aircraft stabilised and the subsequent approach and landing were uneventful.

Investigation

An Investigation was carried out by the Australian Transport Safety Bureau (ATSB). Data relevant to the Investigation was obtained from the aircraft QAR and the NVM contained in the EGPWS.

It was noted that the Captain, a Nauruan national who had been acting as PF for the investigated flight, had a total of 16,600 flying hours of which almost all - 16,100 hours - were on type. He had positioned from his base in Brisbane the day before the investigated event and was well rested after days off but on arriving at Nauru, he reported having discovered that his mother was very ill. As a consequence, he had spent a significant amount of time overnight at the hospital with his mother, achieving only five to six hours sleep. The Company had not been informed of the situation with his mother and at interview, the Captain believed it might have led to fatigue and stated that with hindsight, it would have been better to remove himself from duty. He had not done so because of his perceived ‘pressure’ to meet the ‘request’ that he should be the Nauruan captain flying a duty which involved operating the inaugural Nauru Airlines scheduled service into the Federated States of Micronesia. The First Officer, also based in Brisbane, had a total of 3,300 flying hours of which 1,600 hours were on type and had also positioned to Nauru the day before the investigated event after two days off.

It was established that the flight involved had originated at Nauru before landing at Tarawa and the Marshall Islands. The investigated flight was the following sector to Kosrae, after which the final destination was Pohnpei, both in the Federated States of Micronesia and was the Operator’s inaugural scheduled public transport service to Kosrae and Pohnpei. Those on board included the Nauruan President, the Nauruan Minister of Aviation, and the Chairman of the Board of Directors of the Operator. In the six weeks prior to this flight, the Operator had flown three charter flights to Kosrae and Pohnpei, two of which had been operated in daylight hours by the Captain on the investigated flight.

The departure of the aircraft from Nauru was delayed by an hour after an aircraft change was necessitated by a technical issue with the originally-assigned aircraft. This meant that the 1 hour 20 minute sector from the Marshall Islands to Kosrae was made after nightfall. En route, it became apparent that the weather at Kosrae had deteriorated and during the descent and approach, the Kosrae FISO had provided a considerable number of weather updates. The approach briefed and flown required use of the only instrument approach procedure available to the flight - an offset NDB/DME approach tracking 084° M to the published MDA followed by circling for just over 30° to the right to align with runway 05. There was a straight-in RNAV(GPS) approach procedure to runway 05 but although the Operator was in the process of obtaining regulatory approval for such approaches, this approval was still pending and in any case, the particular aircraft involved was not one of the ones that had already been modified. The Captain stated that during the briefing, special mention had been made regarding the unusually low transition level of FL 55 given that at most airports they operated into, the transition level was between FL 110 and FL 130. It was decided that they would make a maximum of two approaches and then divert back to Nauru, the nominated alternate, if these were unsuccessful.

By the time the aircraft descended through the Transition Level the crew had not - and did not thereafter - complete the Approach Checklist, which consisted of only one item which was to set the altimeters to the local QNH and cross check them. Since the Kosrae QNH was 1007 hPa, this oversight meant that the altimeters were over-reading by about 180 feet. The whole approach was flown without the crew recognising their error. The AP and A/T remained engaged for what was charted as a ‘legacy’ step-down or ‘dive and drive’ procedure wholly over the sea with the final approach beginning from 1500 feet QNH. An altitude of 900 feet was required at the IAF which was at 5 DME (the DME was at the airport), the MDA was 500 feet and the missed approach point was at 2.9 DME. The missed approach procedure required a left turn to pick up and track the 300° outbound radial from the NDB climbing to 3,000 feet. The procedure is shown below.

An annotated copy of the Kosrae NDB/DME-A procedure used for the approach. [Reproduced from the Official Report]

The final descent from an indicated 900 feet began prior to the IAF and as the indicated 500 feet MDA was reached with reportedly intermittent contact with the landing runway through breaks in the cloud, ‘ALT HOLD’ was selected. Almost immediately, an EGPWS ‘TOO LOW TERRAIN’ Alert occurred but after 5 seconds, it was inhibited by the crew on the assumption that it was attributable to ‘map shift’ due to FMS dependence on the IRS in the absence of better data sources. However, the same Alert immediately restarted and continued for 12 seconds. The first Alert was found to have occurred at a height of 368 feet and the second at a height of 340 feet.

As the Missed Approach Point was approached, the previously intermittent visual reference with the landing runway was lost and the AP was disconnected and a manual go-around was commenced. However, as only one press of the TOGA switches on the thrust levers was made, only an intermediate thrust setting was commanded and in the absence of immediate corrective action, the commanded 15° pitch angle led to a reduced speed and then, when reduced, to descent. This continued to within 200 feet of the sea (160 feet below the indicated MDA) and a third EGPWS ‘TOO LOW TERRAIN’ Alert occurred, this time lasting 10 seconds. Corrective action was then taken to place the aircraft in a climb, but FDR data showed that from the time the TOGA switches had been initially pressed to the time when the aircraft was finally stabilised on the missed approach track, the pitch angle had varied from -0.35° to +16°. Repositioning for a second approach and a successful landing followed without further event. It was noted that even disregarding the fact that the altimeters had been over-reading, height keeping at the 500 foot MDA prior to the go around had, despite the AP remaining engaged, been poor - a correctly set altimeter would have read about 320 feet by the time the go around was commenced.

The Investigation reviewed a number of operationally-relevant aspects of the event:

  • Although the crew were unaware of the fact, the EGPWS has an internal GPS and so the position it detected was more accurate that assumed by the FMS (and recorded on the FDR) which was reliant on the IRS using the last reliable position. It was suspected that given the absence of precision radio aids en route, the FMS-computed position had probably been IRS-based for the whole flight sector and a comparison of the recorded runway position at the commencement of the takeoff from the Marshall Islands showed that the FMS position was 810 metres southwest of the actual position. The lateral error had increased to 1,065 metres by the time the Kosrae EGPWS Alerts occurred and after the go around the lateral discrepancy appeared to have increased again to 2,090 metres. However FMS position was not relevant to the approach procedure conducted.
  • The Boeing 737 control column TOGA switches require two consecutive presses to command the A/T to set full thrust. One push only leads to a reduced go around thrust, disconnection of the AP if engaged and a 15° nose up pitch command on the FD. It was noted that the Captain had said that he was not used to conducting two engine go-arounds because simulator training always involved single-engine go-arounds and in this case, full thrust is automatically commended using a single TOGA switch press. Despite over 16,000 hours (as well as almost all) of his flying experience being on the 737, he also commented that the only full thrust go arounds he had made had been in VMC.
  • At the time of the investigated event, the Operator had developed a Fatigue Risk Management System (FRMS) but it was only in draft format and had not received regulatory approval so the risk of fatigue was reliant on the prevailing scheme of flight and duty hours management.
  • The three EGPWS Alerts activated were all triggered by the forward looking Terrain Clearance Floor (TCF). It was noted that the first one was triggered by the initial penetration of the TCF alert envelope. The second was triggered because after an initial envelope penetration, the Alert will re-occur for each further 20 per cent loss of height. The data indicates the aircraft had exited the TCF envelope after the second alert, but then re-entered it for a second time after the flight crew had attempted to fly the missed approach manoeuvre which was the trigger for the third alert. At the time of all three alerts, the aircraft had been in a ‘landing configuration’. The First Officer’s instrument panel included the GPWS controls one that was the ‘Terrain Inhibit’ switch for which the default normal position was mechanically guarded. To inhibit terrain/obstacle alerting, as the crew did after the first Alert, this mechanical guard had to be lifted so that such action was only taken with definite intent. The FCOM was noted as stating that the only exception to correcting the flight path, aircraft configuration or airspeed in response to a TCF Alert was that if it occurred in day VMC and “positive visual verification is made that no obstacle or terrain hazard exists” then it may be regarded as cautionary and the approach continued. Crew action after deciding that the Alerts received had not been generated due to terrain proximity was not in accordance with this procedure.
  • There is considerable evidence over a long period of time that there are a number of factors which tend to increase the risk on an approach and that several of these had been present:
    • The absence of an ATC approach service or airport tower service.
    • A non-precision approach, especially one with a step-down procedure or followed by a circling procedure.
    • visual approach in darkness.
    • The presence of hilly or mountainous terrain in the vicinity of an approach or missed approach track.
    • Visibility restrictions such as darkness or instrument meteorological conditions.

The formally-documented Findings of the Investigation were that there were four Contributing Factors as follows:

  • The flight crew did not complete the Approach Checklist before commencing the non-precision NDB approach into Kosrae. As a result, the altimeters' barometric pressure settings remained at the standard setting of 1013 hPa instead of being set to the reported local barometric pressure of 1007 hPa. The flight crew descended the aircraft to the minimum descent altitude of 500 ft as indicated by the altimeters, however, due to the barometric pressure setting not being reset, the aircraft descended to a height significantly below 500 ft.
  • The crew descended the aircraft in IMC and at night below the approach profile for the Kosrae runway, resulting in EGPWS alerts. Terrain clearance assurance was eroded further by the flight crew not correcting the flight profile until the flight crew lost visual contact with the runway.
  • The flight crew's belief that the EGPWS warnings were due to a decreased navigational performance and not terrain proximity led to their decision to inhibit the first EGPWS warning and not correct the flight path.
  • Due to the Captain’s fatigue and the increased workload and stress associated with the inaugural regular public transport flight into Kosrae at night in rapidly deteriorating weather, the crew’s decision making and task execution on the missed approach were affected.

Two Other Factors that increased risk were also formally identified, one of which was classified as a Safety Issue requiring urgent attention:

  • The crew’s recurrent training had not included B737-300 full thrust go-around simulations.
  • The operator commenced regular public transport operations into Kosrae with the only instrument approach available for use being an offset procedure based on a non-precision navigation aid. The risk associated with this type of approach was amplified due to the need to use a 'dive and drive' style technique instead of a stable approach path, and that it required low level circling manoeuvring from the instrument approach to align the aircraft with the runway. Furthermore, there was very high terrain in close proximity to the runway and the airport did not have a manned air traffic control tower. For this occurrence, the risk was further elevated as a result of the approach being conducted at night-time in poor weather conditions. [Safety Issue]

Safety Action taken by Nauru Airlines as a result of the investigated occurrence whilst the Investigation was in progress was noted as having included, but not been limited to the following:

  • The identified ‘Safety Issue’ relating to approaches at Kosrae was addressed by fitting GPS navigation equipment and the training of flight crew in its use, as well as obtaining regulatory authorisation for the use of GPS based instrument approach procedures which has enabled the operator to conduct runway aligned stabilised approaches into Kosrae - and elsewhere.
  • The flight crew recurrent training regime has been changed to two days twice a year in lieu of one day four times a year so as to provide at least two full days of training each year instead of trying to fit training and checking into each quarterly four-hour session.
  • A review of the process facilitating the conduct of descent and approach checklists has led to the introduction of a control column checklist incorporating tactile indicators.
  • Two engine go-arounds have been added to recurrent simulator training for pilots.
  • QNH setting requirements have been re-visited in ground school and a formal reminder that the validity period for a QNH is 15 minutes has been issued.

The Final Report was released on 16 March 2018. No Safety Recommendations were made.

Related Articles

SKYbrary Partners:

Safety knowledge contributed by: