Safety summary

What happened

On the morning of 28 November 2020, a Jetstar Airways Airbus A320, registered VH-VGP (VGP), was conducting an approach to land at Ballina Byron Gateway Airport, New South Wales (NSW). At the same time, a Jabiru J230D, registered 24-7456 (7456), was conducting a private flight from Heck Field, Queensland, to Evans Head, NSW. About 12 NM south-west of Ballina Airport, the flight paths of the two aircraft inadvertently intersected. The crew of VGP received a traffic collision avoidance system (TCAS) traffic advisory alert prior to passing beneath 7456. The vertical separation between the two aircraft reduced to about 600 ft. Both the pilot of 7456 and the flight crew of VGP observed no lateral separation between the two aircraft.

What the ATSB found

The ATSB’s investigation identified that the pilot of 7456 was not aware of the presence of VGP, or that the two aircraft were converging, until having passed above VGP. The flight crew of VGP were also unaware of the presence of 7456 until they were alerted to the impending conflict by the aircraft’s TCAS. The ATSB also found the pilot of 7456 did not set the aircraft’s transponder to broadcast altitude data. Consequently, the TCAS on board VGP was unable to provide the flight crew with the necessary information to positively avoid the potential collision. The flight crew of VGP were unable to sight the aircraft until just before the flight paths intersected. The vertical separation between the two aircraft was influenced by chance alone as the flight crew of VGP and the pilot of 7456 were not aware of the altitude of the opposing aircraft.

The ATSB also found that the most recent regulatory review of the airspace surrounding Ballina Byron Gateway Airport, and subsequent periodic reviews, had not specifically considered the risks associated with aircraft transiting the airspace without taking off or landing at the airport (such as 7456).

What has been done as a result

The Ballina Airport broadcast area was expanded to a radius of 15 NM in January 2021 and an Airservices Australia surveillance flight information service (SFIS) began operating in August 2021. The SFIS provided traffic information to aircraft operating within the broadcast area on the airport’s common traffic advisory frequency.

The Civil Aviation Safety Authority (CASA) has advised that the current Ballina Airport airspace review (due for release in February 2022) utilises data that includes transiting aircraft. Additionally, CASA has developed an airspace risk modelling system (ARMS) that should provide an enhanced capability to consider transiting aircraft. CASA also advised that an initiative by the Australian Government to increase the uptake of automatic dependent surveillance broadcast (ADS-B) equipment in general aviation would result in improved aircraft detection.

While the proposed CASA actions have the potential to address the safety issue, this will largely depend on the conclusions of the current Ballina Airport airspace review and the effectiveness of the new ARMS. As such, the ATSB will monitor and assess their effect on the safety issue.

Safety message

Communication and self-separation in non-controlled airspace is one of the ATSB’s SafetyWatch priorities. Pilots can guard against similar issues to those highlighted by this incident by:

• making the recommended broadcasts when in the vicinity of a non-controlled aerodrome
• actively monitoring the common traffic advisory frequency while maintaining a visual lookout for other aircraft
• ensuring transponders, where fitted, are selected to transmit altitude information.

The occurrence

On the morning of 28 November 2020, a Jetstar Airways Airbus A320-232 aircraft, registered VH‑VGP (VGP) (Figure 1), was conducting a scheduled passenger service from Melbourne Airport, Victoria, to Ballina Byron Gateway Airport (Ballina Airport), New South Wales (NSW). There were two flight crew, five cabin crew and 163 passengers on board. The captain was pilot flying (PF) and the first officer (FO) was pilot monitoring.^[1]

Figure 1: VH-VGP

Source: Supplied

At about 1122 Eastern Daylight-saving Time,^[2] when VGP was approximately 40 NM to the south‑west of Ballina Airport, the FO made a positional broadcast on the Ballina Airport common traffic advisory frequency (CTAF).^[3] This CTAF was also used by two neighbouring airports and several neighbouring aircraft landing areas (ALAs) (see the section titled Common traffic advisory frequency).

On receipt of the broadcast from VGP, the Ballina Airport certified air/ground radio operator (CA/GRO) (see the section titled Certified air/ground radio service) responded. The CA/GRO confirmed that the flight crew were aware of two other passenger services, an Airbus A320 aircraft departing to the south from runway 06 at Ballina Airport, and a Boeing 737 inbound to Ballina Airport from the south.

At about the same time, a Jabiru J230D aircraft, registered 24-7456 (7456) (Figure 2), was conducting a private visual flight rules flight from Heck Field ALA, Queensland, to Evans Head ALA, NSW. There was one pilot and one passenger on board. At 1124:49, the pilot of 7456 made a broadcast on the shared CTAF, addressed to Lismore traffic, advising that the aircraft was 4 NM to the east of Lismore at 5,300 feet and descending. The flight crew of VGP did not respond to (or recall hearing) this broadcast.

Figure 2: 24-7456

Source: Andrei Bezmylov

Meanwhile, VGP continued tracking towards Ballina Airport via the waypoint^[4] OPESO, descending to an altitude of about 3,200 feet in preparation for the required navigation performance^[5] approach for runway 06. Prior to crossing the OPESO waypoint, the flight crew of VGP received a traffic collision avoidance system (TCAS) proximate traffic alert for an unidentified aircraft at an unspecified altitude in the 11 o’clock^[6] position relative to VGP (see the section titled Traffic collision avoidance system). Unbeknown to the flight crew of VGP, the proximate traffic was 7456 tracking in a southerly direction towards Evans Head (Figure 3).

Figure 3: VH-VGP and 24-7456 tracks

Source: Google Earth, annotated by the ATSB

The pilot of 7456 did not recall hearing the earlier CTAF broadcast from VGP and was unaware that the two aircraft were on converging tracks. 7456 was fitted with a transponder that could transmit the aircraft’s altitude (see the section titled Transponder). However, 7456’s transponder was selected ‘ON’ (not ALT) and was not transmitting the altitude of the aircraft.

Without the altitude information, the TCAS on board VGP could only display the relative bearing and distance of 7456.

The flight crew of VGP attempted, unsuccessfully, to visually acquire the proximate traffic (visibility was greater than 10 km and there was no cloud at the time). The crew later reported experiencing ‘tunnel vision’ while conducting the visual search. They did not attempt to contact the traffic on the CTAF.

At 1128:18, VGP’s flight crew received a TCAS traffic advisory. The flight crew maintained their visual scan and continued with the approach to runway 06.

At 1128:38, the flight crew of VGP made a broadcast on the shared CTAF and advised Ballina traffic, and the Boeing 737 aircraft in the vicinity, that VGP had just passed waypoint OPESO. The pilot of 7456 did not respond to, or recall hearing, this broadcast.

The data obtained from VGP’s quick access recorder and the OzRunways program used by the pilot of 7456, indicated that, at 11:28:41 and 12 NM south-west of Ballina Airport, the tracks of VGP and 7456 intersected, with vertical separation between the two aircraft reducing to about 600 ft. The flight crew of VGP sighted 7456 just prior to passing below the aircraft. The pilot of 7456 sighted VGP shortly after passing above the aircraft. Both the pilot of 7456 and the flight crew of VGP observed no lateral separation between the two aircraft (Figure 4).

Figure 4: : Recorded flight paths of VH-VGP and 24-7456

Source: Google Earth, annotated by the ATSB

At 1128:59, the flight crew of VGP contacted the crew of the Boeing 737 inbound to Ballina Airport on the shared CTAF to advise that they had experienced a traffic advisory, and the involved aircraft was headed in the direction of the Boeing 737.

A short time later, VGP landed at Ballina Airport while 7456 continued on to Evans Head ALA.

__________

1. Pilot Flying (PF) and Pilot Monitoring (PM): procedurally assigned roles with specifically assigned duties at specific stages of a flight. The PF does most of the flying, except in defined circumstances; such as planning for descent, approach and landing. The PM carries out support duties and monitors the PF’s actions and the aircraft’s flight path.
2. Eastern Daylight-saving Time (EDT): Universal Coordinated Time (UTC) +11 hours.
3. A common traffic advisory frequency is a designated frequency on which pilots make positional broadcasts when operating in the vicinity of a non-controlled airport or within a broadcast area.
4. A waypoint is a specified geographical location used to define an area navigation route or the flight path of an aircraft employing area navigation.
5. A statement of the navigation performance necessary for operation within a defined airspace.
6. The position of an object or location relative to the aircraft with 12 o'clock considered the dead-ahead position.

Context

Personnel information

Flight crew VH-VGP

The captain held an Air Transport Pilot Licence (ATPL) (Aeroplane) and had a total flying time of 13,935 hours, having flown 66.5 hours in the previous 90 days. The captain was familiar with Ballina Airport but had not operated there often. The captain’s last flight to Ballina Airport took place in June 2019.

The FO held an ATPL (Aeroplane) and a total flying time of 4,830 hours, having flown 10.5 hours in the previous 90 days. The FO was somewhat familiar with Ballina Airport having operated there twice before the incident flight. The FO’s last flight to Ballina Airport took place the day prior to the incident.

Pilot 24-7456

The pilot held a recreational pilot certificate and a total flying time of 775 hours, having flown 56 hours in the previous 90 days. The pilot was familiar with Ballina Airport and transited the surrounding airspace about eight times a year. The pilot reported being unfamiliar with the radio phraseology commonly used by passenger transport flight crew, including reference to waypoints such as OPESO.

Ballina Byron Gateway Airport

Ballina Byron Gateway Airport is a certified airport situated approximately 3 NM from the city of Ballina, NSW. The airport has an elevation of 7 feet above mean sea level (AMSL) and a single sealed runway orientated in a 062°-242° magnetic direction (Figure 5). The airport had GPS‑based instrument approaches and a non-directional beacon ground-based navigation aid.

Figure 5: Ballina Byron Gateway Airport

Source: Airservices Australia

Airspace and traffic services

Ballina Airport was located within non‑controlled Class G airspace which extended from the ground surface to 8,500 feet AMSL. The airport did not have a control tower and was not supported by air traffic control (a non-controlled airport).

Overlying the non‑controlled airspace was Class C controlled airspace which extended up to flight level (FL) 180,^[7] and controlled Class A airspace above that. An air traffic information and separation service was provided within the Class C airspace and a separation service was provided within the Class A airspace. A restricted area existed approximately 5 NM south of the airport (the aircraft involved in this incident were clear of this area).^[8]

The non‑controlled airspace surrounding Ballina Airport was available for use by aircraft operating under visual flight rules and instrument flight rules. No separation service was provided to aircraft operating in this airspace with pilots responsible for making themselves aware of nearby aircraft and maintaining self‑separation.

The primary method of traffic separation at Ballina Airport was visual and relied on pilots using ‘alerted see-and-avoid’^[9]practices. A broadcast area was in place requiring aircraft to use a radio on the Ballina Airport CTAF when operating within a 10 NM radius of the airport.

Common traffic advisory frequency

The Ballina Airport CTAF was a designated frequency on which pilots made positional broadcasts when operating in the vicinity of the airport. The Ballina Airport CTAF was shared with neighbouring airports and ALAs Casino, Lismore and Evans Head to assist traffic coordination and enhance the situational awareness of pilots operating within the surrounding airspace (Figure 6).

Figure 6: Shared common traffic advisory frequency

Source: Airservices Australia, annotated by the ATSB

When operating within the Ballina Airport broadcast area, pilots were required to make mandatory transmissions when arriving or departing from the airport, and when flying through the broadcast area.

When operating outside of the Ballina Airport broadcast area, but within the vicinity of the other non‑controlled airports on the shared CTAF, pilots were required to make a broadcast whenever it was reasonably necessary to do so to avoid a collision, or the risk of collision, with another aircraft. There were also several recommended positional broadcasts (Table 1).

Table 1: Recommended positional broadcasts in the vicinity of a non-controlled airport

Source: Civil Aviation Advisory Publication 166-01 V4.2 (Feb 2019) with minor amendments by the ATSB

Certified air/ground radio service

At the time of the incident, Ballina Airport was one of two airports in Australia to have a certified air/ground radio service (CA/GRS) in operation.

According to CASA’s guidelines for a CA/GRS, the primary purpose of the service was to enhance the safety of passenger operations by providing all pilots with information to enhance their ability to see-and-avoid potentially conflicting traffic.

The information provided to pilots included:

• frequency confirmation
• traffic information on first call
• airport weather
• other advice to facilitate aeronautical safety and efficiency

The Ballina Airport CA/GRS commenced operations in March 2017 in response to the increasing number of aircraft movements at the airport. The service was delivered by the airport operator, via a third-party contractor, and had formal CASA approval. The service was delivered by a certified air/ground radio operator located at the airport.

At the time of the incident, the service was provided to all aircraft operating within the Ballina Airport broadcast area of 10 NM, during passenger service operations (greater than 30 seats) between the hours of 0800-1800 local time.

The CA/GRS did not provide a traffic separation service to aircraft as would occur at airports located within Class D or higher airspace (controlled airports).

Regulatory oversight

The Airspace Act 2007 assigned the administration and regulation of Australian administered airspace to the Civil Aviation Safety Authority (CASA). As part of this function, CASA was required to undertake regular reviews to determine if:

• existing classifications of ‘volumes of airspace’ were appropriate
• existing air navigation services and facilities provided to volumes of airspace were appropriate
• there was safe and efficient use of airspace, and equitable access to that airspace for all users
• any identifiable risk factors were present

Ballina airport airspace review

At the time of this occurrence, CASA had last completed a review of the airspace surrounding Ballina Airport in July 2015. The purpose of that airspace review was to examine the airspace classification within 20 NM of the airport from ground level to 8,500 feet AMSL.

The objective of the review was to assess the risk levels for passenger transport operations to determine if the airspace classification was appropriate and whether an air traffic service was required. The options considered during the review included upgrading the airspace to Class D or higher classification (a controlled airport).

The Australian Airspace Policy Statement (AAPS) contained airspace review criteria thresholds for volumes of airspace around airports (Table 2). If an airport met or exceeded any of the thresholds for a classification, then CASA was required to undertake a review of the volume of airspace in question. This review was to consider public, industry and agency comments, forecast future traffic levels, and any significant risk mitigators before finalising an airspace determination.

Table 2: Airspace criteria thresholds AAPS 2015

Source: Civil Aviation Safety Authority

The 2015 airspace review found that the Ballina Airport exceeded the AAPS passenger criteria threshold for Class D airspace with 437,940 passenger movements recorded in 2014. However, it also found that both passenger transport aircraft movements and total aircraft movements for 2014 were significantly below the Class D trigger criteria. Therefore, CASA decided not to upgrade Ballina to a controlled airport on the basis that:

Considering the total aircraft movements, total passenger transport aircraft movements, stakeholder feedback and aviation safety incident reports CASA considers Class D or higher would currently be a disproportionate response to the identified airspace issues at Ballina.

The ATSB noted that, in reaching that conclusion, the airspace review had not assessed risks associated with aircraft transiting the airspace surrounding Ballina Airport without taking off or landing at the airport (that is, transiting aircraft such as 24-7456 was in this occurrence). While the review included analysis of reportable events near Ballina Airport from 2009 to 2014, it did not specifically consider the involvement of transiting aircraft in those events or the influence such aircraft may have on future separation events to properly assess the risks involved. Of the 11 separation events in the area that the review identified, 6 involved a passenger transport service. However, while CASA advised that all of these occurrences were considered, the review did not identify that 2 of those 6 also involved a conflict with a transiting aircraft.

The review primarily focused on aircraft movements (the total number of take-offs and landings at an airport) specified in the AAPS threshold criteria. Movement data for Ballina Airport was sourced from Airservices Australia (Airservices) and the Bureau of Infrastructure, Transport and Regional Economics.

While Airservices had some data on transiting aircraft, it was not readily available as computational analysis of this data was required to extract information on aircraft operating under the visual flight rules and/or aircraft operating without a flight plan. Notwithstanding these difficulties, data for transiting aircraft was available, but not obtained or analysed in the 2015 review.

In March 2020, CASA commenced another review of the airspace, but following the expansion of the Ballina Airport broadcast area, decided to defer its finalisation. At the time of writing, a final report had not been released.

Periodic assessments

In addition to the 2015 airspace review, CASA conducted periodic risk assessments of Ballina Airport and the surrounding airspace. The information considered for these assessments included:

• aircraft and passenger movements
• incident reports
• IFR to VFR traffic ratios
• stakeholder feedback
• previous risk assessments.

Records provided by CASA indicated eight documented risk assessments between July 2015 (when the 2015 airspace review was published) and the time of the incident. The last of these assessments was completed in June 2020. The analysis recorded within these assessments primarily focused on aircraft and passenger movements. There was no reference to transiting aircraft, either in terms of occurrences involving such aircraft or overall numbers/movements.

Comments included in this assessment stated that passenger transport aircraft and passenger numbers had declined at the airport, probably due to the impact of COVID-19,^[10] but passenger numbers continued to exceed the AAPS 2018 threshold by 19 per cent. The assessment indicated three reported incidents had occurred in the vicinity of the airport but provided no detail about those incidents. The last periodic assessment concluded that the risk level was lower than that of the previous assessment.

Previous events

A search of the ATSB database identified that between 1 January 2010 and 28 November 2020, there were 20 separation events involving passenger transport services (including this incident) within a radius of approximately 20 NM of Ballina Airport below 8,500 feet AMSL (Table 3). Of those 20 events, seven involved a conflict between an aircraft transiting the airspace and a passenger transport service approaching or departing from Ballina Airport (Table 4). Two of these events occurred prior to the publication of the 2015 CASA airspace review.

Table 3: Ballina separation events involving passenger transport services

Source: ATSB

Table 4: Separation events involving passenger transport service and transiting aircraft

Source: ATSB

Aircraft systems

Traffic collision avoidance system

A traffic collision avoidance system (TCAS), as fitted to VGP, interrogates the transponders (see the section titled Transponder) of nearby aircraft and uses this information to calculate the relative range and altitude of this traffic. The system provides a visual representation of this information to the flight crew as well as issuing alerts should a traffic issue be identified. These alerts include:

• Proximate traffic – an alert issued when an aircraft is within a range of less than 6 NM and 1,200 ft, or a range of 6 NM if the traffic is not transmitting altitude information. Depicted as a white filled diamond on the navigation display (ND)
• Traffic advisory (TA) – an alert issued when the detected traffic may result in a conflict (the closest point of separation is about 40 seconds away on the current projected flight paths). Depicted as an amber filled circle on the ND and an aural alert. Pilots are expected to initiate a visual search for the traffic causing the TA (the operator’s procedures required flight crew not to perform a manoeuvre based solely on a TA).
• Resolution advisory (RA) – a manoeuvre, or a manoeuvre restriction, calculated by the TCAS to avoid a collision (the closest point of separation is approximately 25 seconds away or less). Depicted as a red filled square on the ND and vertical speed orders on the primary flight display. A series of aural alerts will also sound. Pilots are expected to respond immediately to an RA.

Due to its method of operation, a TCAS cannot detect aircraft that are not equipped with a transponder (or switched off). Additionally, the system is unable to issue an RA for traffic that is not fitted with an altitude reporting transponder (mode C or S), or in circumstances where the mode C or S transponder on board the conflicting traffic is not transmitting altitude information (as was the case with 7456).

Transponder

A transponder is a receiver/transmitter which transmits an automatic reply upon receiving an interrogation request. A manual ‘ident’ transmission can also be initiated by the pilot. The information transmitted by a transponder is dependent on the ‘mode’ of equipment fitted and the mode of transmission selected by the pilot.

The transponder fitted to 24-7456 was capable of operating in mode 3A and 3C. In mode 3A (ON) the equipment would transmit the configured transponder code only. In mode 3C (ALT) the equipment would transmit the aircraft’s altitude in addition to the configured transponder code.

The pilot of 24-7456 had elected to set the transponder to mode A only so the altitude of the aircraft was not being transmitted. The pilot incorrectly believed there were no requirements relating to the use of modes 3A and 3C.

The Aeronautical Information Publication (AIP) contained information on the operation of aircraft transponders. AIP ENR 1.6 paragraph 7.1.2 stated:

Unless advised otherwise by ATC, pilots of Mode 3A or Mode S transponder equipped aircraft operating in Australian airspace must activate their transponders, and where a Mode C capability is also available it must be activated simultaneously with Mode 3A.

Note: Pilots must ensure that transponders and ADS-B transmitters are activated and the altitude function is selected as:

a. primary radar coverage only exists within 50NM of major airports and the remainder of the ATS surveillance system relies on SSR transponder and ADS-B transmitter information, and b. TCAS relies on transponder information for its pilot alerting and collision avoidance functions.

AIP GEN 1.5 paragraph 7.1.2 stated:

Pilots of transponder-equipped aircraft should ensure their transponder is switched to ON/ALT (Mode C) at all times.

Human factors

The ATSB investigation considered a range of human factors that could have influenced the decisions and actions of the pilots involved. No indicators that increased the risk of any of the individuals experiencing a level of fatigue known to influence performance were found. The following factors, however, were likely to have had an influence.

• cognitive tunnelling
• human visual system limitations.

Cognitive tunnelling is an inattentional blindness where an individual becomes overly‑focused on some variable other than the present environment (Mack & Rock 1998). Cognitive tunnelling may also impact an individual’s decision-making processes (Bell et al 2005). The comments by VGP’s flight crew about ‘tunnel vision’ after they received the TCAS TA was a reference to this factor.

Limitations associated with the human visual system include:

• empty field myopia – in an empty field, such as blue sky, the eye will focus at around 56 cm which may reduce the chance of identifying a distant object
• a lack of relative motion when on a collision course – the human visual system is less effective at detecting stationary objects than moving objects. Because of the geometry of collision flightpaths, from each pilot’s point of view, the converging aircraft will grow in size while remaining fixed at a particular point in the windscreen
• visual angle – an approaching aircraft at high speed will present a small visual angle until a short time before impact. Limitations associated with visual acuity mean this small visual angle may make it impossible for a pilot to detect the aircraft in time to take evasive action (Hobbs 1991).

All these limitations impair the effectiveness of ‘see-and-avoid’ practices, the primary method of self-separation in uncontrolled airspace, which requires pilots to conduct a visual search to ‘see-and-avoid’ potentially conflicting traffic.

An ‘unalerted’ search is one where reliance is entirely on the pilot searching for, and sighting, another aircraft without prior knowledge of its presence. On the other hand, an ‘alerted’ search is one where the pilot is alerted to another aircraft’s presence, typically via radio communications or aircraft based alerting systems. An alerted search is likely to be eight times more effective than an unalerted search (Hobbs 1991).

__________

7. Flight level: at altitudes above 10,000 ft in Australia, an aircraft’s height above mean sea level is referred to as a flight level (FL). FL 180 equates to 18,000 ft.
8. The restricted area was activated by a notice to airmen when military jet aircraft were operating within the area and/or live-firing exercises were taking place.
9. Pilots are responsible for sighting conflicting traffic, and avoiding a collision, having been alerted to the presence of traffic in their immediate vicinity. This is principally achieved via radio communications.
10. Coronavirus disease (COVID-19) was an infectious disease caused by a newly discovered coronavirus. The World Health Organization (WHO) first learned of this new virus on 31 December 2019. International and domestic responses to manage the pandemic included reducing aviation activity internationally and domestically.

Safety analysis

The incident

On 28 November 2020, the flights paths of Airbus A320, VH-VGP, and Jabiru J230D, 24-7456, intersected about 12 NM south-west of Ballina Byron Gateway Airport, with the vertical separation between them reducing to about 600 ft. The aircraft were outside the airport’s broadcast area and within non‑controlled Class G airspace. That meant that the pilots were responsible for safe separation using radio communications and collision avoidance aids to support ‘see-and-avoid’ practices.

Communications and collision avoidance aids

The flight crew of VGP did not recall hearing the broadcast from 7456 prefixed with ‘Lismore traffic’ on the common traffic advisory frequency (CTAF). That prefix (a different airport to their destination, Ballina) probably did not trigger their auditory attention. In any case, they were not aware of 7456 until alerted to a conflict by the aircraft’s traffic collision avoidance system (TCAS).

Similarly, the pilot of 7456 did not recall hearing broadcasts from VGP on the CTAF. Additionally, the pilot was not familiar with phraseology commonly used in passenger transport operations, including reference to instrument approach waypoints such as OPESO. That probably contributed to the pilot not registering those broadcasts and being unaware of the presence of VGP before the incident.

The pilot of 7456 incorrectly selected mode 3A (instead of the required 3C) on the aircraft’s transponder resulting in altitude data not being transmitted. As the aircraft approached VGP, its TCAS issued the conflict alert and, subsequently, a traffic advisory alert with the relative bearing and distance of 7456. However, the TCAS could not indicate the approaching aircraft’s altitude or provide a resolution advisory, significantly disadvantaging the flight crew in managing the situation.

See-and-avoid

The circumstances and the restrictions imposed on the available electronic aids, particularly TCAS functionality, were impediments to effectively applying see-and-avoid practices.

Since the pilot of 7456 was not aware of VGP, seeing and avoiding it depended on the success of ‘unalerted’ visual searches by the pilot. Although visibility was greater than 10 km with no cloud in the area, any searches were unsuccessful as the pilot only sighted VGP after passing above it.

On the other hand, VGP’s TCAS alerted the flight crew to 7456. However, their ‘alerted’ search was unsuccessful in part due to the limitations associated with the human visual system and the absence of altitude data. Additionally, the effects of cognitive tunnelling, together with the rapid sequence of events following the TCAS alerts, possibly resulted in them not considering options such as contacting the approaching aircraft via radio on the CTAF. With no TCAS resolution advisory manoeuvre available, they continued visually searching, sighting 7456 moments before it passed and too late to take any action to avoid a potential collision.

Therefore, the vertical separation of about 600 ft when the aircraft passed was entirely fortuitous.

This incident, together with previous events involving transiting aircraft shows that separation occurrences, with potentially serious consequences, can and do occur (Table 4) in the airspace encompassing the approach and departure flight paths of large transport aircraft operating at Ballina Airport.

Airspace classification

The objective of the 2015 Civil Aviation Safety Authority (CASA) review of the airspace surrounding Ballina Byron Gateway Airport was to assess the level of risk posed to passenger transport services and determine if the airspace classification was appropriate. The review considered total aircraft movements (including passenger transport aircraft) at the airport, some reportable events near the airport and stakeholder feedback and determined that a Class D or higher airspace classification was not appropriate. Significantly, transiting aircraft movements were not considered, primarily due to difficulty in extracting this data.

However, the review had not specifically considered the risks associated with aircraft transiting the airspace without taking off or landing at Ballina Airport. Recognising that such a consideration at the time of the review would have only identified 2 occurrences associated with transiting aircraft, additional occurrences in the intervening years indicate that the risk of a proximity event in these circumstances may be significant. However, opportunities to consider such risk events in the periodic risk assessments of the airspace since the 2015 review had not been taken.

Capturing these tangible risks through the airspace review and periodic assessment mechanisms would result in a more accurate and realistic risk assessment. Such a risk assessment would, in turn, better inform an assessment of the appropriate airspace classification.

When operating in non-controlled airspace (such as the current Class G airspace around Ballina), whether under the instrument or visual flight rules, pilots hold responsibility for separation from other aircraft. A review of past occurrences indicates that self-separation using broadcast traffic advice has been a largely reliable procedure.

The ATSB does however note that the effectiveness of the current pilot-separation method relies on individual pilots:

• recognising a potentially unsafe situation
• formulating an effective separation plan that often requires coordination with the occupants of the other involved aircraft.

While on this occasion one of the involved aircraft was equipped with TCAS, this process is almost exclusively reliant on individual human actions without other mechanisms potentially acting as a safeguard and/or safety redundancy, and as such subject to human error, even when it involves experienced pilots. Furthermore, such errors often increase under high workload associated with, for example, instrument flying approach procedures, low experience or a busy airspace environment.

Of note, the airspace surrounding Ballina Airport accommodates a complex mix of aircraft types and operations, including high-capacity passenger transport, while also being located close to several other non‑controlled airports.

In that context, while the available evidence in this investigation does not support a conclusion that the present self‑separation system is unsafe, there is an opportunity to potentially reduce safety risk further. 

The ATSB therefore supports systemic enhancements to the overall air traffic system that have been assessed by regulatory and air traffic specialists, in keeping with their obligations as providing a net overall safety increase. Key examples of such enhancements include:

• the increased use of controlled airspace and ADS‑B aircraft surveillance data (both by air traffic services and in‑cockpit)
• improved monitoring of air traffic movements (both quantity and complexity) to assist the identification of increasing risk areas.

With respect to this occurrence, had the aircraft been operating in controlled airspace the crew of VGP would have, at a minimum, received traffic information from air traffic control on 7456 and may have been positively separated.

Findings

From the evidence available, the following findings are made with respect to the separation occurrence involving Airbus A320-232, VH-VGP and Jabiru J230D, 24-7456 about 12 NM south‑west Ballina Byron Gateway Airport, New South Wales on 28 November 2020.

Contributing factors

• The mode of the transponder on board 24-7456 was not selected to transmit altitude data, which resulted in VH-VGP’s traffic collision avoidance system (TCAS) not indicating the approaching aircraft’s altitude or providing a resolution advisory.
• The pilot of 24-7456 did not recall hearing broadcasts from VH-VGP and remained unaware of the other aircraft until passing above it.
• The flight crew of VH-VGP did not recall hearing the broadcast from 24-7456 when it was near Lismore and remained unaware of the aircraft until receiving a TCAS alert. However, they did not know the approaching aircraft’s altitude, did not attempt to make radio contact and only sighted it moments before the aircraft passed.
• The vertical separation between the aircraft when their flight paths inadvertently intersected reduced to about 600 ft, which was entirely fortuitous as the pilots of neither aircraft had been able to manage aircraft separation in the circumstances.

Other factors that increased risk

• The Civil Aviation Safety Authority review and periodic risk assessments of the airspace surrounding Ballina Byron Gateway Airport did not include data for aircraft transiting the airspace without using the airport. Therefore, the risk associated with occurrences such as this one were not specifically considered when assessing the appropriate airspace classification. (Safety issue)

Safety issues and actions

Ballina airport safety review

Safety issue number: AO-2020-062-S1-01

Safety issue description: The Civil Aviation Safety Authority review of the airspace surrounding Ballina Byron Gateway Airport did not include data for aircraft transiting the airspace without using the airport. Therefore, the risk associated with occurrences such as this one were not specifically considered when assessing the appropriate airspace classification.

Glossary

AIP                   Aviation information publication

ALA                  Aircraft landing area

AMSL               Above mean sea level

ATC                 Air traffic control

ATSB               Australian Transport Safety Bureau

CA/GRS           Certified air/ground radio service

CASA               Civil Aviation Safety Authority

CTAF               Common traffic advisory frequency

EDT                 Eastern daylight-saving time

FDR                 Flight data recorder

FL                    Flight level

FO                   First officer

NM                   Nautical mile

NSW                New South Wales

PF                    Pilot flying

PM                   Pilot monitoring

RA                   Resolution advisory

SFIS                Surveillance flight information service

TA                    Traffic advisory

TCAS               Traffic collision avoidance system

UTC                 Coordinated universal time

Sources and submissions

Sources of information

The sources of information during the investigation included:

• the flight crew of VH-VGP and pilot of 24‑7456
• the CA/GRO and CA/GRS service provider
• Jetstar Airways
• Ballina Byron Gateway Airport
• Avdata
• Bureau of Infrastructure and Transport Research Economics
• Civil Aviation Safety Authority
• Airservices Australia

References

Bell, M, Facci, E, & Nayeem, R 2005, Cognitive Tunnelling, Aircraft-Pilot Coupling Design Issues and Scenario Interpretation Under Stress in Recent Airline Accidents, 2005 International Symposium on Aviation Psychology, 45-49

Hobbs A 1991, Limitations of the See-and-Avoid Principle, Australian Transport Safety Bureau  

Mack A & Rock I 1998, Inattentional blindness, MIT Press Cambridge MA

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to:

• the crew of VH-VGP and pilot of 24-7456
• Jetstar Airways
• Ballina Byron Gateway Airport
• Civil Aviation Safety Authority
• Airservices Australia.

Submissions were received from:

• the crew of VH-VGP
• Civil Aviation Safety Authority
• Airservices Australia

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations & publishing information

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2022 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

Safety summary

What happened

On the afternoon of 2 December 2020, a Robinson Helicopter Company (RHC) R44 Raven I, registered VH-HGU, departed Goulburn Airport, New South Wales with a student pilot and instructor on board. The helicopter flew east, and the last recorded automatic dependent surveillance broadcast (ADS-B) detected it descending into a valley in the Bungonia State Conservation Area. A search commenced when the helicopter did not return as expected, and the wreckage of VH-HGU was found in a valley, approximately 4 km north-west of its last ADS-B transmission. Both pilots were fatally injured, and the helicopter was destroyed.

What the ATSB found

Wreckage examination indicated that while flying in the vicinity of the valley, the helicopter entered a low-G condition due to turbulence, inappropriate control inputs, or a combination of both. This condition, probably in combination with inappropriate recovery control inputs resulted in extreme teetering of the main rotor. A mast bump occurred as a result, and the helicopter subsequently broke up in flight.

An intense post-impact fire prevented a complete examination of the wreckage. However, the evidence available gave no indication that the helicopter was operating abnormally prior to the in‑flight break-up.

The circumstances leading to in-flight break-ups from mast bumping and extreme teetering are usually not identified. While the fire would likely have prevented data recovery in this case, the inclusion of readily available cockpit video recorders on helicopters with semi-rigid rotor heads would provide valuable insights into low-G mast bumping events, which could help to prevent future occurrences.

What has been done as a result

The operator has introduced a limit of 6 consecutive training days. In addition, an online flight and duty monitoring system will be used to assess and adjust duty hours for instructors and students.

Since the occurrence, RHC has introduced cockpit video/audio recorders as standard equipment on new R44 helicopters, having previously been available on R66 models. The recorders are optional on R22 models, but will be standard equipment in 2023, with retrofit kits made available.

Safety message

Low-G conditions can be catastrophic for helicopters with semi-rigid rotor heads. A pilot’s ability to recover from low-G remains uncertain, though all parties agree that it is dependent on airspeed and time available. Pilots must therefore avoid low-G situations, and take the following actions to mitigate risk:

• Make a careful study of the terrain, forecasts and observations applicable to the proposed flight to identify significant weather such as fronts or mechanical turbulence and avoid flying in those conditions.
• Reduce airspeed when encountering significant turbulence.
• Avoid flight downwind of hills, ridges or other potential sources of turbulence, particularly during changing or unpredictable weather conditions.
• Allow plenty of altitude so that there is enough time for recovery from unexpected flight conditions.
• Use slow and small control inputs.

Recording devices have long been recognised as an invaluable tool for investigators in identifying the factors behind an accident, and their contribution to aviation safety is irrefutable. While not required by regulations, operators should consider the benefits of installing such devices.

The occurrence

On the morning of 2 December 2020, an instructor and student pilot began a day of flight training operations in a Robinson R44 Raven I, registered VH-HGU. The pilots conducted a training area familiarisation from Goulburn Airport, New South Wales, after which the student pilot’s first solo flight was conducted.

Over the course of the morning and early afternoon, the student completed two sets of solo circuits and a solo navigation flight around the edge of the training area. Immediately after the navigation flight, the instructor joined the student pilot on board VH-HGU, and at 1558 Eastern Daylight‑saving Time,^[1] they departed Goulburn Airport toward the east.

At 1846, when the helicopter had not returned to Goulburn as expected, the training operator’s chief pilot contacted the Australian Maritime Safety Authority’s Joint Rescue Coordination Centre (JRCC). The chief pilot then began an aerial search of the training area in another helicopter.

During the search, JRCC provided the chief pilot with the helicopter’s last known location—the last recorded automatic dependent surveillance broadcast (ADS-B)—which had detected it descending into a valley in the Bungonia State Conservation Area at 1611. At about 2000, the chief pilot located the helicopter’s wreckage 4 km north‑west of its last ADS-B transmission, which was in a valley approximately 31 km east of Goulburn Airport (Figure 1). The pilot then provided the wreckage location to emergency services. The helicopter’s engine and fuselage had been exposed to a fire at the main wreckage site, which had self‑extinguished before emergency services arrived. Both pilots had been fatally injured.

Figure 1: Flight path and wreckage location of VH-HGU

The aircraft track was calculated from ADS-B data.

Source: Google Earth, FlightAware, annotated by the ATSB

__________

1. Eastern Daylight saving Time (EDT): Coordinated Universal Time (UTC) +11 hours

Context

Pilot information

Instructor

The instructor held a commercial helicopter pilot licence and a Grade 3 helicopter flight instructor rating. At the time of the occurrence, the instructor had accrued a total flying time of 690 hours. This included 684 hours in the Robinson R44, of which 103.1 hours had been in the past 90 days and 64.3 hours in the previous 30 days. The pilot held a Class 1 Aviation Medical Certificate valid until 7 December 2021.

Student

The student was 12 days into an intensive flight training course for the issue of a private helicopter pilot licence (PPL), having not flown previously. Over that time, the student had accrued a total of 37.9 hours of dual flying, all on VH‑HGU. Prior to the accident, the student pilot had accrued 3 hours of solo flying. The student pilot held a Class 2 Aviation Medical Certificate valid to 23 September 2022.

Fatigue considerations

During the 12 days of flying training, the work patterns of the instructor and the student identified some factors that placed them both at risk of their performance being affected by fatigue. Some research (Powell and others, 2007) conducted on short-haul airline pilots indicated that the most important influences on fatigue were the number of sectors conducted and the duty period. This study indicated that fatigue increased with each additional sector studied, up to a maximum of 5 sectors studied. On 30 November, the pilots conducted four flights totalling 4.2 hours flying time. The instructor also conducted a fifth flight with another student, 1.5 hours long. On 1 December, the day before the accident, the pilots conducted five flights with 5.5 hours total flying time. On 2 December, prior to the accident flight, the student pilot had conducted three solo flights totalling 3 hours, and two dual flights totalling 1.3 hours.

The nature of the work also could have contributed to fatigue—the Civil Aviation Safety Authority has stated that from a workload perspective, training and monitoring another flight crew member is often more fatiguing than regular operations^[2].

In the days prior to the accident, the student pilot was reportedly working into the night studying PPL theory. In conjunction with the flight operations and other training activities conducted during training days, this could have resulted in duty limits^[3] being exceeded. However, it was not possible to obtain an accurate record of the amount of time either pilot spent on ground operations, including PPL study.

Helicopter information

The Robinson Helicopter Company (RHC) R44 Raven I is a four-seat, piston-engine helicopter, powered by a Lycoming O-540-F series six-cylinder carburetted engine. At the end of 2020, there were 566 R44 helicopters on the Australian civil aircraft register.

The R44 Raven I helicopter involved in the accident, serial number 2615, was built in May 2020 in the United States. The helicopter was first placed on the Australian register as VH-HGU on 10 July 2020.

Rotor system

The Robinson R44 main rotor hub assembly design is a semi-rigid rotor head, otherwise known as a teetering rotor head (Figure 2). Bolts secure the blades to the hub at the coning hinges. During stopping and starting of the main rotor, when rpm is low, the blade tusks rest against the droop stops, restricting teetering and preventing the blades from drooping. As the main rotor rpm increases, the blades become rigid and straighten due to rotational forces, and the tusks shift off the droop stop as the blades lift. During normal flight, the rotor is free to teeter and flap around its designed flight axis via the teeter hinge, while polyurethane teeter stops limit the degree of teetering.

Figure 2: R44 main rotor hub assembly

Source: Robinson Helicopters, annotated by the ATSB

Under certain specific flight conditions, semi-rigid rotor systems are susceptible to extreme teetering where the blades teeter beyond their normal operational range, resulting in a what is commonly known as ‘mast bumping’. Mast bumping is the act of the inboard end of the blade (the spindle) contacting the main rotor shaft. In R44 helicopters, this can generally be identified by extensive damage to the teeter stops and varying degrees of damage to the main rotor shaft. Extreme teetering will also result in failure of one or both pitch links,^[4] allowing uncontrolled pitching of the blade(s). This allows the blade(s) to deviate from their normal path of rotation.

Maintenance

The operator was using the RHC R44 manual suite to maintain VH-HGU. The most recent maintenance release was not recovered from the wreckage, but the helicopter had an estimated 150 hours total time in service.

According to RHC, it had received reports of burned intake valves on engines with serial numbers ending in ‘40E’ and less than 500 hours’ time-in-service, which included VH-HGU. It noted that a burned valve can result in partial or complete loss of power. On 14 October 2020, RHC issued a safety alert for engine intake valves installed on O-540-F1B5 engines. The alert stated:

Pilots should observe the following precautions:

          1. Perform a complete run up and stabilized hover check prior to every flight. Do not initiate flight if there is any indication of engine roughness or sudden yaw.

          2. If engine roughness or a sudden yaw occurs in flight, land as soon as practical and be prepared to land immediately.

The intake valves were inspected, and a compression check was conducted at a recent periodic inspection (about 101 hours total time in service), with no issues identified.

Meteorological information

Weather forecast

The Goulburn aerodrome forecast (TAF)^[5] for the day of the accident predicted clear conditions up until a change at about 1600, which was expected to bring 13 kt easterly winds and broken cloud with a base 2,000 ft above ground level (AGL). The cloud was expected to descend to 1,000 ft AGL over the next few hours following the change.

The flight took place on the borders of several Graphical Area Forecasts^[6] (GAF) with different predicted conditions. Forecast conditions issued at 1515 and valid from 1600 indicated some areas of:

• greater than 10 km visibility
• isolated drizzle, rain and thunderstorms reducing visibility, with associated broken cumulous and stratocumulus clouds as low as 2,000 ft above mean sea level (AMSL)
• isolated drizzle with broken stratus clouds between 1,000 and 2,000 ft AMSL.

Reported weather conditions

The Bureau of Meteorology (BoM) provided METAR^[7] and SPECI^[8] data from Goulburn and Moss Vale aerodromes, located 31 km west and 43 km north-east of the accident site, respectively. These were the two automatic weather stations closest to the accident site, but indicated significantly different weather conditions—at the time of the accident and 1 hour prior.

Between 1530 and 1630, the METARs for Goulburn aerodrome recorded the wind varying in direction from 290° to 340°, up to 9 kt. The visibility remained greater than 10 km, and the temperature was 25 °C. There was no other significant weather observed. The 1530 METAR at Moss Vale recorded the wind from 080° at 14 kt, and a layer of overcast^[9] cloud at 1,800 ft AGL and 2,600 ft AGL.

Based on the Moss Vale observations, a SPECI was first issued at 1538, due to a significant deterioration of cloud conditions in the area. Conditions at this time included a layer of broken cloud at 1,300 ft AGL and overcast layer at 2,000 ft AGL. The SPECI issued at 1600 included a layer of broken cloud at 1,100 ft AGL an overcast layer at 1,800 ft AGL. These conditions continued to deteriorate for several hours.

Bureau of Meteorology analysis

The BoM provided an analysis of the likely weather conditions based on nearby observations and the geographical features of the accident location, such as the valley. It estimated that the weather change would have arrived at the accident site between 1540 and 1610. It would then have taken 30 minutes or more for the change to pass.

Before the change, the BoM determined that conditions would probably have been fine and clear with light to moderate winds. Low cloud and turbulence would not have been expected. Similarly, after the change, conditions would likely have been clear, with cloud probably not developing until approximately 1630.

At the time of the change, however, the BoM predicted the possibility of a short period of moderate or greater low-level windshear and turbulence. As the change moved across the area, it noted that low-level vertical windshear of 20 kt or more would be expected, with possible turbulence and rotor winds^[10] induced by the local terrain.

Recorded information

The ADS-B transmissions detected during the accident flight provided the helicopter’s track after departing Goulburn towards the Bungonia State Conservation Area. The last recorded transmission at 1611 indicated that the helicopter was descending into a system of valleys, approximately 4 km southeast of where it later impacted terrain.

The training operation regularly used OzRunways^[11] for flight planning and tracking, however there was no data recorded for the accident flight. There were no other recording devices fitted to the aircraft, and data from personal electronic devices found at the accident site could not be recovered.

In early 2021, RHC introduced cockpit video cameras, standard on the R66 and optional for the R22 and R44. The forward-facing camera records video (encompassing view through the windshield, pilot controls and the instrument panel), intercom audio, radio transmission and GPS data. RHC advised the recordings (up to 10 hours) can be used as a training tool, maintenance aid, or aerial-tour souvenir. VH-HGU did not have a cockpit video camera installed.

Purpose of the flight

There was nothing in either pilot’s logbook or the training syllabus to indicate the purpose of the accident flight. The chief pilot—who was also the owner of the training operation—reported that the flight was likely to provide the student with area familiarisation. Previous students of the instructor were asked about any similar flights during their training—none had any recollection of such a flight. However, these students trained in regions without any substantial mountain ranges or valley systems nearby.

Without any documentation or recorded information to indicate the purpose of the flight, it could not be determined which pilot might have been flying the helicopter at the time of the accident.

Wreckage and impact information

Accident site and distribution of wreckage

The accident site was located about 31 km east of Goulburn Airport, near the bottom of a valley in the Bungonia State Conservation Area. The valley was about 1 km wide and 500 m deep.

The wreckage trail was approximately 275 m long on a south-easterly heading, parallel to the river. Figure 3 shows the location of the wreckage and calculated direction of flight prior to impact, as well as the helicopter’s last detected ADS-B transmission.

Figure 3: Wreckage trail and final ADS-B transmission from VH-HGU

Orange arrows show the aircraft heading at its last transmission point and prior to the break-up. Heading at its last transmission point was determined using ADS-B data, while the heading prior to the break-up was estimated based on the direction of wreckage and damage to trees around the accident site.

Source: Google earth, annotated by the ATSB

The start of the wreckage trail consisted of:

• multiple pieces of windshield Perspex
• fragments of the forward left door
• a small piece of main rotor blade trailing edge skin
• left landing gear skid toe cap
• covers for the left and forward sections of the landing gear.

Extensive examination of surrounding vegetation identified no evidence of a tree strike at the start of the wreckage trail.

Figure 4 shows the locations of some of the components identified in the wreckage trail. An outboard section of one main rotor blade was located about halfway along the debris trail, along with liberated pieces of landing gear skids and struts. Damage to vegetation commenced about 20 m prior to the impact zone and was consistent with a relatively shallow descent profile. Impact forces likely ruptured the crash-resistant fuel tanks and an intense fire ensued, destroying the fuselage, engine and main gearbox. The fire zone was about 10 m in diameter. The tail cone assembly was situated a few metres to the west of the fire zone and was unaffected by the fire. All extremities of the helicopter were identified.

Figure 4: Location of aircraft components following in-flight break-up

Source: Google earth, annotated by the ATSB

Wreckage examination

Fuselage

The fuselage was upright, but almost entirely destroyed by the fire. The few flight instruments that were identifiable did not exhibit any witness marks that could assist the investigation. All of the seat belt buckles were found in the secure (closed) configuration, but the belt webbing had perished in the fire.

Main rotor drive and hydraulics

The main rotor drive assembly was primarily intact, however, the intense heat of the post-impact fire had reduced the integrity of the mounts, and the assembly had collapsed. Examination of the main gearbox did not identify any pre-existing condition that could have affected its operation.

Damage due to impact forces and fire also prevented a complete examination of the hydraulic system. However, a hydraulic system failure should not result in a loss of control or in-flight break‑up.

Engine

Examination of the severely fire damaged engine and clutch assembly identified no pre-existing issues that might have affected operations. The cooling fan exhibited some rotational scoring, consistent with the engine operating up to the point of impact. Based on the presence of fire and a fuel odour present during the on-site examination, there was a considerable amount of fuel on board at the time of the break-up

Various engine components—including the throttle and magnetos—were identified, however, the fire damage prevented complete examination. The engine was retained for further examination (see section titled Engine inspection).

Flight controls

Most of the flight control tubes had melted in the fire. Those that were recovered showed fractures consistent with overstress due to the impact. Without all flight control tubes, examination of the bellcranks was used to confirm flight control continuity from the main rotor hub through to the tail rotor—The steel rod ends were securely attached despite the control rod having melted. Damage to the pilot dual flight controls did not provide any indication as to whether the left or right seat pilot was in control.

Main rotor assembly

The main rotor hub was secured to the mast, and both spindles—including an inboard portion of each main rotor blade—were attached. Both main rotor pitch links had failed in overstress at the upper rod end thread, probably as a result of extreme teetering. The pitch link rod ends were secured to their respective pitch horns. The remainder of the pitch links were secured to the swashplate. Both spindles could be rotated freely about their axis of operation.

The coning hinge bolts were removed to allow removal of the spindles and remains of the main rotor blades (designated red and blue) from the main rotor hub (MRH). While the MRH attachment point on the red rotor blade was undamaged, the blue rotor blade tusk had been severely deformed (bent downwards nearly 90°), and the coning hinge bolt bore was split open as a result. This damage is shown in Figure 5, with the undamaged tusk from the red rotor blade for comparison. Following removal of the MRH, the teeter stops were seen to exhibit damage consistent with severe impact from the blade spindles, shown in Figure 6.

Figure 5: Blue blade tusk and coning hinge bolt bore

Source: ATSB

Figure 6: Teeter stop damage

Source: ATSB

The underside of the blue main rotor blade exhibited impact damage and a witness mark consistent with the dimensions of a landing gear strut. In addition, the forward left landing strut showed some scrapes and scuff marks. The location of the damage on the blade was consistent with the measured distance of the left forward strut from the main rotor hub. Some yellow paint transfer was noted on the forward section of the left skid, consistent with a main rotor strike.

Figure 7: Damage to the blue main rotor blade and the forward left landing strut

Source: ATSB

Tail cone

On-site examination of the separated tail cone assembly, including tail rotor drive components and flight control surfaces identified no pre-existing anomalies that could have contributed to a loss of control. In addition, black paint transfer was noted on the tail cone, just aft of where it had separated from the fuselage. The red main rotor blade had impact marks consistent with the rivet pattern at that location, including comparative distance from the main rotor hub. The paint transfer and rivet marks were indicative of tail cone separation due to a main rotor blade strike, likely just prior to the collision with terrain. The black paint was from the underside of the main rotor blade, indicating that it was inverted due to the fractured pitch link.

Engine inspection

• catastrophic mechanical failure
• oil starvation to the crankshaft journals and conrod bearings
• damage to the valves, valve faces or cylinder bore.

Summary

Distribution of the wreckage, damage to the main rotor system, and damage to the fuselage and skids are all consistent with a main rotor blade striking the left forward fuselage at the commencement of an in-flight break-up.

Examination of the wreckage did not identify any pre-existing issues with the helicopter that could have contributed to:

• loss of control
• engine power loss
• in-flight break-up.

Comments from the manufacturer

According to RHC, the damage observed was consistent with an extreme teetering event leading to mast bumping. It described the two different scenarios below to explain further:

Low rotor rpm mast bumping

In a low rotor rpm situation, the main rotor disc can start to cone, and subsequent mast bumping events tend to result in the main rotor striking the tail cone, often leading to separation of the tail.

Low-G mast bumping

RHC stated that extreme teetering and subsequent mast bumping can result from the pilot attempting to recover from an uncommanded right roll while in a low-G condition. This condition is expanded in the section below (see the section titled The low-G condition). RHC stated that inappropriate recovery control inputs often resulted in the main rotor blade impacting the forward left side of the fuselage.

In the case of VH-HGU, RHC concluded from the available evidence that mast bumping was the result of an extreme cyclic input while in a low-G condition. However, there was insufficient evidence to determine whether low-G was induced by a pilot input, turbulence or a combination of both.

The low-G condition

‘G’ or ‘g’ is an abbreviation for the acceleration due to the earth’s gravity. Positive G is necessary for helicopters to respond to pilot control inputs. In a low-G condition (that is, approaching the feeling of weightlessness), the pilot’s ability to control the attitude of the helicopter is greatly reduced. A low-G condition could be induced by an unusual attitude (for example, if the helicopter was inverted due to spatial disorientation). However, during normal flight, low-G is typically induced either by a cyclic pushover, or by turbulence.

Low-G due to cyclic pushover

The FAA Helicopter Flying Handbook^[12] provided a description of how cyclic pushover can induce low-G and ultimately result in a mast bump:

During a pushover from moderate or high airspeed, as the helicopter noses over, it enters a low-G condition. Thrust is reduced, and the pilot has lost control of fuselage attitude but may not immediately realize it. Tail rotor thrust or other aerodynamic factors will often induce a roll. The pilot still has control of the rotor disk, and may instinctively try to correct the roll, but the fuselage does not respond due to the lack of thrust. If the fuselage is rolling right, and the pilot puts in left cyclic to correct, the combination of fuselage angle to the right and rotor disk angle to the left becomes quite large and may exceed the clearances built into the rotor hub. This results in the hub contacting the rotor mast, which is known as mast bumping.

RHC Safety Notice SN-11^[13] provided a descriptive warning about the cyclic pushover manoeuvre described above. It also stated that severe in-flight mast bumping usually results in main rotor shaft separation and/or rotor blade contact with the fuselage. The following warning was included:

Never attempt to demonstrate or experiment with low-G manoeuvres, regardless of your skill or experience level. Even highly experienced test pilots have been killed investigating the low-G flight condition. Always use great care to avoid any manoeuvre which could result in a low-G condition. Low-G mast bumping accidents are almost always fatal.

Low-G due to turbulence

The FAA Helicopter Flying Handbook stated that ‘Turbulence, especially severe downdrafts, can also cause a low-G condition and, when combined with high airspeed, may lead to mast bumping.’ RHC Safety Notice SN-32^[14] referred to flying in high winds or turbulence, firstly stating that it should be avoided. It continued:

A pilot’s improper application of control inputs in response to turbulence can increase the likelihood of a mast bumping accident.

The following were among RHC’s recommendations when encountering turbulence:

          - Reduce power and use a slower than normal cruise speed. Mast bumping is less likely at lower airspeeds.

          - Avoid flying on the downwind side of hills, ridges, or tall buildings where turbulence will likely be most severe.

In November 2016, RHC released a safety alert^[15] reiterating the dangers of cyclic pushovers and the low-G condition, warning against over-reacting or flying too fast when in turbulence. If pilots found themselves flying in the low-G condition, RHC recommended the application of gentle aft cyclic in order to reload the rotor.

Previous occurrences

There have been a number of fatal mast bumping accidents involving helicopters with semi-rigid rotor heads worldwide. A review of occurrences in Australia, New Zealand and the United States identified the following accidents, dating back to 1983:

• In New Zealand, there have been 12 fatal mast bumping accidents, all involving RHC helicopters, 6 of which have been deemed low-G accidents.
• In the United States, there have been 18 fatal accidents where mast bumping was the primary occurrence. Eleven of those accidents involved RHC helicopters.
• In Australia, there has been one other known fatal mast bumping accident.^[16] It was determined to be induced by low rpm rather than low-G.

Many of the low-G fatal mast bumping accidents in New Zealand and the United States share other similarities with the in-flight break-up of VH-HGU. Two such accident investigations by the New Zealand Transport Accident Investigation Commission (TAIC) are summarised below:

Loss of control, R44 helicopter ZK-HTB, New Zealand (AO-2018-006)

The helicopter was travelling to Upper Estuary Burn Valley when it departed controlled fight and crashed into Lake Wanaka, fatally injuring the pilot who was the sole occupant. Wreckage examination found signs of mast bumping, including teeter stop damage and evidence of the main rotor striking the canopy, however it was not conclusively determined to be an initiator of the occurrence. With regard to turbulence, the report stated:

Turbulence is a known contributor to mast bumping. Large, sudden upward or downward gusts can upset a helicopter and cause the blades to flap up or down excessively, or cause a low-G situation. Inappropriate or inadvertent pilot inputs or over-controlling by the pilot can further exacerbate the effects of the turbulence.

The investigation found that the helicopter probably encountered turbulence that was strong enough to have resulted in the in-flight break-up of the helicopter. The report also made note of the lack of available data regarding mast bumping accidents in RHC helicopters

In-flight break-up, R44 helicopter ZK-IPY, New Zealand (AO-2015-002)

The helicopter was returning to Queenstown from a training flight when it broke up in mid-air and crashed near the Lochy River, fatally injuring the instructor and student. The break-up occurred after a main rotor blade struck the cabin due to mast bumping. The investigation report stated that mast bumping was typically caused by one or a combination of the following factors:

• low main rotor rpm
• the helicopter entering a low-G condition
• turbulence
• the pilot making large and abrupt movements with the helicopter controls.

The report noted that a low-G scenario could have been induced by a combination of airspeed and the pilot’s response to turbulence. In a safety recommendation, TAIC stated:

The uncertainty around the circumstances of this accident are not unique. The nature of mast bump accidents is that they are usually fatal, leaving no one to explain what was happening at the time. In-flight break-ups are destructive, making it difficult to determine with certainty whether mechanical failures of some kind could have initiated the mast bumps.

The report included a recommendation emphasising ‘…the need for cockpit video recorders and/or other forms of data capture in the cockpits of certain classes of helicopter to address this safety issue [mast bumping]’. The report identified the following key lesson:

Helicopter pilots must be fully aware that a condition of low-G (feeling of lightness or weightlessness) can result in: a rapid right roll; mast bumping; and in-flight break-up before even the most experienced pilot can react and recover the situation. Pilots need to fly in a manner that avoids low-G conditions rather than allow them to develop and then expect that they can recover from them.

__________

2. CAAP 48-01 (2020), paragraph 5.3.6.
3. Regulatory duty limits manage the effects of fatigue on a pilot, such as Civil Aviation Order 48.1. /a>
4. Pitch links are rods that adjust main rotor blade angles based on pilot control inputs.
5. Aerodrome Forecast (TAF): a statement of meteorological conditions expected for a specific period of time in the airspace within a radius of 5 NM (9 km) of the aerodrome reference point. The heights referenced in TAFs are heights above the aerodrome reference point (ground)
6. Graphical Area Forecast (GAF) provides information on weather, cloud, visibility, icing, turbulence and freezing level in a graphical layout with supporting text. These are produced for 10 areas across Australia, broadly State-based.
7. METAR: A meteorological report for an aerodrome at a routine time (half hourly) when conditions are better than specified thresholds
8. SPECI: A weather report for an aerodrome or significant location issued whenever weather conditions fluctuate below specified criteria.
9. Cloud cover: in aviation, cloud cover is reported using words that denote the extent of the cover— ‘scattered’ indicates that 3 to 4 eights of the sky in covered by cloud, ‘broken’ indicates that more than half to almost all the sky is covered, and ‘overcast’ indicates that all the sky is covered.
10. Rotor winds are a type of turbulence associated with mountain waves—a weather phenomenon involving winds flowing over mountain ranges.
11. The OzRunways application is a portable information system for flight deck crew members which allows storing, updating, delivering, displaying and/or computing digital data to support flight operations or duties. It provides the option for live flight tracking by transmitting the device’s position and altitude.
12. FAA Helicopter Flying Handbook, Chapter 11: Helicopter Emergencies and Hazards
13. RHC Safety Notice SN-11: Low-G Pushovers – Extremely Dangerous
14. RHC Safety Notice SN-32: High Winds or Turbulence
15. RHC Safety Alert: Low-G Mast Bumping Accidents
16. AO-2016-156: In-flight break-up involving Robinson R44, VH-ZNZ, 41 km NW Mossman, Queensland, on 18 November 2016

Safety analysis

Introduction

While flying in the vicinity of a valley in the Bungonia State Conservation Area on the afternoon of 2 December 2020, the R44 helicopter, VH-HGU, experienced an in-flight break-up. As a result, both pilots were fatally injured, and the helicopter was destroyed.

Although fire and impact damage had destroyed some parts of the helicopter, the ATSB was able to use available evidence to make certain important findings. Components recovered near the beginning of the wreckage trail indicated that a main rotor blade had struck the left side of the fuselage at the beginning of the break-up sequence. No evidence was found to indicate pre-existing mechanical defects or issues that could have prevented normal engine operation. While a transient condition such as a partial or complete power loss could not be ruled out, such an event should not have resulted in an in-flight break-up.

The investigation considered the possibility of clouds or reduced visibility forcing a descent. However, a Bureau of Meteorology (BoM) analysis of the weather conditions indicated that low cloud would not have created difficulties until after the time of the accident. Further, any unfavourable weather coming with the easterly change would not have prevented the pilots from safely returning to Goulburn at any point.

As there were no survivors and no available records, the purpose of the flight could not be determined, nor which pilot had been flying the helicopter. For the same reason, their working hours that day and in the preceding days could not be ascertained. While the nature of their work and work patterns in terms of overall duty hours and time spent on ground operations suggested that both pilots could have been at risk of fatigue, there was insufficient evidence to determine whether or not fatigue was a factor in this accident.

The following analysis focuses on factors that probably contributed to the in-flight break-up, which include the low-G condition in helicopters with semi-rigid rotor heads, the effects of turbulence, and the influence of control inputs (pilot actions) in such conditions.

Helicopter entered low-G

The broken teeter stops, failed pitch links, bent tusk and blade impact damage were consistent with the low-G mast bumping scenario described by the Robinson Helicopter Company (RHC). The damage was also similar to many other in-flight break-up incidents that were the result of extreme teetering after entering a low-G condition.

There have been several fatal mast bumping accidents where turbulence has probably contributed to inducing a low-G scenario. While not a necessary condition to enter low-G, the Federal Aviation Administration has stated that turbulence can induce a low-G condition. RHC has also stated that turbulence can contribute to a low-G mast bumping event and encourages pilots to avoid the severe turbulence associated with flying on downwind side of hills and ridges.

Based on the weather and the geographical conditions at the accident location, the BoM analysis noted the potential for windshear and moderate or greater turbulence at the time of the accident, dependent on the arrival of a weather change. It is likely but not certain that turbulence was present in the area at the time of the break-up. The various weather forecasts showed differing conditions, but it could not be determined which, if any, the pilots checked before departure nor the conditions that they were expecting.

While the evidence shows that VH-HGU entered a low-G condition, it was not possible to determine the contribution that turbulence or incorrect/inappropriate pilot control inputs (for example, initiating a cyclic pushover) had in triggering the condition. However, one or both of these factors must have been present.  

In-flight break-up following low-G

While it has been well established that a low-G condition can result in mast bumping and an in-flight break-up, there remains some uncertainty regarding a pilot’s ability to recover from a low-G situation, and whether inappropriate recovery control inputs are necessary for an accident to occur.

Both RHC and New Zealand’s Transport Accident Investigation Commission (TAIC) have acknowledged that turbulence and airspeed can contribute to a low-G mast bump and subsequent in-flight break-up. The latter has identified turbulence as a sufficient condition for low-G mast bumping to occur and stated that certain low-G situations can lead to mast bumping and in-flight break-up before a pilot can reasonably react. Conversely, RHC has stated that turbulence only results in mast bumping when pilots react with inappropriate control inputs. However, it should be noted that according to RHC, even highly experienced test pilots have sometimes failed to recover from a low-G condition.

In this accident, once the helicopter entered a low-G condition, inappropriate recovery control inputs probably contributed to the mast bumping and subsequent in-flight break-up. Following the right roll that can be induced by low-G, either pilot could have reacted by instinctively applying left cyclic, rather than the gentle aft cyclic recommended by RHC. Extreme teetering, potentially in combination with broken pitch links resulted in the main rotor striking the fuselage and the helicopter breaking up in flight.

Video recorders

Throughout the accident sequence, the relative contribution of turbulence and pilot actions could not be determined due to the lack of available evidence. This lack of evidence is common to many previous low-G mast bumping accidents. While the post-impact fire in this accident would probably have destroyed recorded data, including readily available cockpit video recorders on helicopters with semi-rigid rotor heads would provide valuable insights into low-G mast bumping events, and help to prevent future occurrences.

Findings

From the evidence available, the following findings are made with respect to the in-flight break-up of Robinson R44 VH-HGU on 2 December 2020.

Contributing factors

• While flying in the vicinity of the valley, the helicopter entered a low-G condition due to turbulence, inappropriate control inputs, or a combination of both.
• The low-G condition, probably in combination with inappropriate recovery control inputs resulted in extreme teetering of the main rotor and subsequent in-flight break-up.

Other findings

• No evidence was found to indicate that the helicopter was not capable of normal operation prior to the in-flight break-up.
• In helicopters with semi-rigid rotor heads, the circumstances leading to in-flight break-ups from mast bumping and extreme teetering are not well understood. Recorded cockpit imagery could provide valuable insight for understanding the effect that weather conditions and pilot input has in these occurrences.

Safety action

Safety action by Hughes Helicopters

Following this accident, Hughes Helicopters placed a limit of 6 consecutive training days on the intensive flight training course. An online flight and duty system has been introduced so that the head of operations can monitor and adjust the duty periods and hours flown by students and instructors. Instructors will be continuously monitoring student fatigue levels and adjusting the training program accordingly.

Safety action by the Robinson Helicopter Company

In 2022, video and audio recorders that were already equipped on R66 models became standard equipment on all new production R44 models and optional on R22 models. In 2023, the recorders will be standard on all three models and retrofit kits will be available.

Glossary

Sources and submissions

Sources of information

The sources of information during the investigation included the:

• chief pilot
• Bureau of Meteorology
• Robinson Helicopter Company
• OzRunways
• maintenance organisation for VH-HGU
• former trainee pilots
• Civil Aviation Safety Authority
• New South Wales Police Force
• New Zealand Transport Accident Investigation Commission
• United States National Transportation Safety Board
• United States Federal Aviation Administration.

References

Civil Aviation Safety Authority (2020); CAAP 48-01 v3.2 Fatigue management for flight crew members. D20/103350. Civil Aviation Safety Authority, Canberra, Australia. https://www.casa.gov.au/sites/default/files/2021-08/caap-48-01-fatigue-management-for-flight-crew-members.pdf . Accessed on 16 December 2020

Powell, DMC; Spencer, MB; Holland, D; Broadbent, E and Petrie KJ (2007) Pilot fatigue in short-haul operations; effects of number of sectors, duty length and time of day. Aviation Space and Environmental Medicine 2007; 78:698-701.

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• the operator
• the Robinson Helicopter Company
• the Civil Aviation Safety Authority
• the United States National Transportation Safety Board.

Submissions were received from:

• the operator
• Robinson Helicopter Company
• Civil Aviation Safety Authority.

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations & publishing information

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2022 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

Safety summary

What happened

At around 1736 on 23 November 2020, V/Line loco-hauled passenger service 8239 from Southern Cross to Melton passed signal SST535 at stop, about 700 m north-west of Southern Cross Station. The signal was at stop to protect the movement of V/Line passenger service 8156 (Wendouree to Southern Cross) that was to cross ahead of train 8239. Train 8239 passed the signal at about 23 km/h and continued for about 200 m before stopping across the junction through which train 8156 would pass. The driver of train 8156 brought their train to a stop about 100 m prior to that junction. The drivers of both services had stopped their trains following an emergency broadcast from the controlling signaller. 

What the ATSB found

When approaching signal SST535 at stop, the driver of train 8239 was probably distracted by task unrelated thoughts. They did not respond to the indication of the signal and probably looked past to another signal further along the track that they incorrectly believed was the signal where they were to stop.

Once signal SST535 had been passed, the risk control to reduce the likelihood of a collision was primarily the action of the signaller to respond to system alarms. In this instance, the signaller responded by making an emergency broadcast to train 8156 and this was sufficient to stop that train. The driver of train 8239 also overheard this broadcast and stopped their train.

Some signals on the V/Line network were fitted with Train Protection and Warning System (TPWS) equipment that automatically initiated a brake application when a signal was passed at danger (SPAD). Signal SST535 was not fitted with TPWS and its absence at this signal increased the potential consequences of a SPAD. A 2014 V/Line risk assessment of signal SST535 had not considered a head-on or side-on collision as a credible scenario. Such a consideration would have increased the risk rating of this signal and probably led to the fitting of TPWS.

Train 8156 was approaching signal SST578 as train 8239 passed signal SST535 at danger. However, signal SST578 continued to show a proceed aspect and did not restore to danger until train 8156 had passed that signal. There were no further signals in front of train 8156 to warn that the junction ahead was occupied by train 8239. The V/Line signalling standards did not include specific reference to a design that would require signal SST578 to restore to danger in this scenario (flank-track protection).

What has been done as a result

V/Line Corporation has advised that a funding submission to fit TPWS to signal SST535 and other similar signals in the Southern Cross area was approved in July 2021. Plans are underway to complete this installation during 2022.

The Rail Industry Safety and Standards Board (RISSB) advised that flank - track protection has been recorded for inclusion when the signalling principles standard AS 7711 is next under review.

Safety message

This occurrence has highlighted the importance for passenger rail networks to have engineering controls in place to detect SPAD events and prevent potential consequences such as collision. In determining applicable SPAD risk controls, rail operators should consider all SPAD precursors and potential collision scenarios.

The occurrence

Prior to the event

The driver of V/Line Corporation^[1] (V/Line) loco-hauled passenger service 8239 had signed on for duty at 1505 on the day of the incident and moved the train from the yard at Southern Cross Station to platform 3 at approximately 1600. At 1633 the conductor of train 8239 informed the driver that platform work had been completed. At the same time, a passenger approached the locomotive at the front of the train. The driver then turned on the headlights and revved the train engine to warn that the train was departing, and the train left without the passenger boarding.

Figure 1: Map showing location of incident

Source: PASS Assets (Department of Transport, Victoria) adapted by CITS

Journey of Train 8239

Train 8239 was a loco-hauled passenger train scheduled to travel from Southern Cross Station to Melton. Closed circuit television (CCTV) showed train 8239 departed Southern Cross Station on-time at 16:33:58, and passed signal SST537 at 16:34:53. This signal was located at the northern end of the station and displayed a red aspect over a yellow aspect,^[2] which advised the driver to proceed at a limit of medium speed^[3] but be ready to stop at the next signal. The train then approached the LaTrobe Street road-over-rail bridge and diverged onto the dual gauge track. On passing under the LaTrobe Street bridge, train 8239 passed signal SST535 showing a stop (red aspect over red aspect) indication and occupied the track circuit^[4] past this signal at 16:35:40. Train 8239 passed the signal at 23 km/h. The speed limit for this section of track was 25 km/h.

At 16:35:50, train 8239 progressed onto the next track circuit that was located 53 m after signal SST535. At 16:36:03 the on-board event recorder^[5] on train 8239 recorded the application of the train brakes, and at 16:36:13, with train 8239 having stopped approximately 228 m past signal SST535. This placed train 8239 over a junction that was on the intended route of train 8156. The driver of train 8239 reported having stopped on seeing train 8156 on the flyover approach to Southern Cross Station and hearing a radio communication from the signaller requesting train 8156 to stop.

The signaller had observed CCTV of train 8239 approaching signal SST535 and reported that based on their observations, anticipated that the train would not stop at the red signal. At approximately the same time, the signaller reported hearing and seeing a SPAD alarm warning from the train control system. Consequently, the signaller commenced a radio broadcast to the driver of train 8156, requesting that they stop owing to a train having passed a signal at danger.

Figure 2: Simplified layout of the relevant tracks and signals at Docklands (not to scale)

Source: CITS

Journey of Train 8156

Train 8156 was a Diesel Multiple Unit (DMU) passenger train scheduled to run from Wendouree to Southern Cross Station. The on-board event recorder showed the train departed Footscray Station, the station prior to Southern Cross Station, on time at 1629. A recording of the signalling display^[6] showed that the train was routed via the ‘Up Regional Rail Link (RRL) dual gauge’^[7] track (Figure 2).

On approaching Southern Cross Station, the driver received the radio communication from the signaller instructing the train to be stopped, and the on-board event recorder showed that the driver responded by making a brake application at 16:35:47. At this point the train was travelling at 27 km/h. At 16:35:57 the signal interlocking recorded that train 8156 passed signal SST578 at a proceed aspect and occupied track circuit SST584T. The on-board event recorder showed that the emergency brake was applied at 16:35:58 and that the train came to a stop at 16:36:01, approximately 20 m past signal SST578 and about 100 m from train 8239.

__________

1. V/Line was a Victorian Government state-owned enterprise operating as a not-for-profit corporation.
2. A red over yellow signal in Victorian three aspect speed signalling is referred to as a medium speed warning signal. Speed signalling indicates the speed at which a train may travel, and of any speed change required.
3. Medium speed at this location was 40 km/h. However, a permanent speed restriction of 25 km/h applied to this section.
4. An electric circuit where current is carried through the rails and used to detect the presence of trains. Track circuits are used in the operation and control of points and signalling equipment.
5. The station CCTV showed that the timings of the CCTV, signal interlocking, and on-board event recorders were aligned.
6. A video recording of the signalling display, as seen at the time of the incident by the Southern Cross signaller.
7. The Up Regional Rail Link (RRL) dual gauge is the name given to the track by V/Line. It indicates that the track is for trains heading in the direction of Southern Cross Station (up), was installed as part of the Regional Rail Link project and accommodates both broad-gauge and standard gauge trains.

Context

Location

The SPAD occurred on a section of track that had been constructed in 2014 as part of the Regional Rail Link project (RRL). The RRL project included extensive track and signalling works for accessing platforms 1 to 8 at Southern Cross Station. V/Line provided both the infrastructure management and train control functions for this section of track, under a lease arrangement with VicTrack.^[8]

Train and driver information

Details of trains

Train 8239 was a scheduled passenger service operated by V/Line. It consisted of an N-class locomotive and six carriages, with an overall length of 136 m. Train 8156 was a three carriage VLocity DMU with a length of 76 m. Both trains were operated by solo drivers. 

Driver of train 8239

The driver commenced with V/Line as a trainee driver in 2017 and qualified as a V/Line driver in August 2019. A V/Line train driver safety re-accreditation audit was conducted in September 2020 and no non-conformances were recorded. Medical records provided by V/Line showed the driver had undertaken a medical assessment (rail category 1 – high-level safety worker) in 2017 and was assessed as fit for duty. The fitness certificate was still valid at the time of the incident. V/Line information did not record any safety incidents related to the driver during the 36 months prior to the SPAD, and reported the driver was compliant with the V/Line driver fatigue policy. The driver advised that they were feeling physically well and had a good night’s sleep prior to the incident. The driver was familiar with the route taken by train 8239, having last driven the same service over the same route two weeks earlier.

The driver had signed on for duty at 1505 on the day of the incident and following some train preparation duties, train 8239 was their first passenger service of the day. The driver reported being distracted by a personal issue on the afternoon of the incident and that they were reminded of this issue by a passenger who had just missed the departure of the train from Southern Cross Station. The passenger event, that was captured on station CCTV, was not reported to Train Control and was not required to be reported. At the time of the SPAD, the driver was not in conversation with the train’s conductor and there was no record of the driver using a mobile phone at the time.

The driver reported thinking about the passenger they had left behind when passing signal SST537 on the departure from Southern Cross Station. This signal displayed a red indication over a yellow indication, which advised the driver to be ready to stop at the next signal. The driver reported seeing signal SST537 and acknowledging it by repeating the aspect to themselves. While the driver reported acknowledging this warning signal, they also advised that they did not at that time think about the next signal, SST535, which was 351 m ahead of signal SST537. The driver reported they looked past ground-mounted signal SST535 at LaTrobe St bridge and saw gantry-mounted signal SST991 at Dudley St showing a red indication (Figure 3). The driver reported they erroneously matched stop signal SST991 to the warning (to be ready to stop at the next signal) provided by SST537.

The driver saw train 8156 and around the same time heard the signaller on the train radio requesting train 8156 to stop. The driver of train 8239 then applied emergency braking.

Following the incident, testing of the driver returned a zero blood alcohol result and nil presence of drugs.

Figure 3: Signals SST535 and SST991 at stop

Source: CITS. Photo taken from the cab of an N-Class locomotive post-incident.

Driver of train 8156

The driver was qualified for the operation of this train on this route and was medically fit for duty.

Recordings of the train radio and the on-board event recorder show that the driver promptly stopped the train on being requested to do so by the signaller.

Environmental conditions

Meteorological conditions recorded at Melbourne for the afternoon indicated mild and cloudy conditions, 7 oktas^[9] of cloud cover, and light south westerly winds.^[10] Ambient temperature was recorded as 17.8 °C at 1500. The cloud cover recordings were consistent with CCTV recordings from Southern Cross Station. It was determined that the environmental conditions did not contribute to the incident.

Train control and signalling

Overview

The signalling of the V/Line infrastructure at Southern Cross Station was operated from Southern Cross Number 1 signal box. Absolute Block Signalling rules applied at this location.^[11] A computer-based interlocking controlled points and signals to prevent conflicting movements of signals, and to make sure routes were set correctly.

In relation to the real-time monitoring and management of field signalling equipment, the train control system provided the signaller with a video display unit (VDU) and interface to enter control requests. The signaller was responsible for the day-to-day operational management of the rail corridor for V/Line trains departing and entering Southern Cross Station.

Signal, points, track, and train movement data were captured by a computer-based interlocking event logger. Voice communication between train drivers and the signallers at the Southern Cross Number 1 signal box was via channel one of the Local Radio System (LRS), which was recorded. The radio channel was an open system and communications from the signaller could be heard by the drivers of all train in the area.

Signalling system playback

The status of signals and the movement of trains were captured by the Southern Cross train control system (TCS) event logger. Figure 4 shows the position of trains, the status of signals and points, and routes set at 16:35:31 (9 seconds prior to the SPAD event), as were shown on the signaller’s train control system display. Red indicated the presence of a train on a section of track and green indicated the track was clear and the route was set. Train 8239 approached signal SST535 at stop. A route had been set for train 8156 to pass in front of train 8239 and onto the ‘Through Country’ track, with signals SST986 and SST578 both at proceed.

Figure 4: TCS playback showing train 8239 approaching SST535 at stop

Source: V/Line Southern Cross Train Control System, annotated by CITS.

Figure 5 shows that at 16:35:47, train 8239 had passed signal SST535 at stop, and train 8156 was approaching signal SST578 at proceed. It is estimated that train 8156 was between 50 m and 80 m from signal SST578 when train 8239 passed signal SST535 at stop.

Figure 5: TCS playback showing train 8239 having passed SST535 at stop

Source: V/Line Southern Cross Train Control System, annotated by CITS

Figure 6 shows that at 16:36:03 train 8239 was continuing to travel towards the route set for train 8156. The system playback shows train 8156 had passed SST578 when it showed a proceed indication, and this signal had subsequently restored to danger once the train had moved onto the track circuit past the signal. Both trains had come to rest by 16:36:13.

Figure 6: TCS playback showing train 8156 having passed SST578 showing proceed

Source: V/Line Southern Cross Train Control System, annotated by CITS

Signaller

The signaller advised that they commenced work as a signaller at Southern Cross in 2018, having worked previously for another rail infrastructure manager for 25 years in various safeworking roles. The signaller was qualified to operate the signalling control panel at Number 1 signal box at Southern Cross.

On 23 November, the signaller had signed on for duty at 1300 to commence an eight-hour shift. The signaller reported observing CCTV of train 8239 approaching signal SST535 and based on their observations, anticipated that the train would not stop at the red signal. Consequently, the signaller commenced a radio broadcast directed to the driver of train 8156, requesting that they stop owing to a train having passed a signal at danger. At approximately the same time, the signaller reported hearing and seeing a SPAD alarm warning from the train control system. Following the communication with the driver 8156, the signaller directed a broadcast to the driver of train 8239.

The signaller was subjected to post-incident drug and alcohol testing returning negative results for both tests.

Signaller radio communications

The following communications took place between the signaller and the driver of train 8156 (Table 1). The time log on the radio communication showed that the initial call from the signaller to the driver of train 8156 was followed 8 seconds later with a second ‘urgent’ call. The driver of train 8156 replied immediately after this second call.

Table 1: Radio communications between signaller and driver of train 8156

The following communications then took place between the signaller and the driver of train 8239 (Table 2).

Table 2: Radio communications between signaller and driver of train 8239

Signal STT535 information

Overview

Signal SST535 (Figure 7) was commissioned in 2014 as part of the RRL project as a ground mounted mainline three-position home signal. The main signal aspects were displayed as light-emitting diode (LED) colour light combinations on the “A” and “B” lights. The “C” light could display a yellow low speed aspect when both A and B lights were red. As the signal was located on a dual gauge line, it was provided with “V” and “S” indicators. A “V” indicator displayed in conjunction with another proceed aspect authorised a broad-gauge train to pass the signal, and an “S” indicator displayed in conjunction with another proceed aspect authorised a standard gauge train to pass the signal. The signal was not equipped with Train Protection and Warning System (TPWS)^[12] or other train protection equipment.

Post-incident observations from the driver’s cab showed that signal SST535 was visible to the driver for approximately 60 seconds when approached in compliance at the allowable track speed from platforms 1 to 5 at Southern Cross Station.

Risk of SPAD at signal SST535

In May 2014, the V/Line RRL Management of Change and Risk Team undertook a risk review to determine which signals should be fitted with TPWS. At that time, the Rail Safety National Law (RSNL), the V/Line Safety Management System (SMS), and the RRL Scope and Technical Requirements required risk assessments to be undertaken to assess the risk of train-to-train collision and determine whether additional signal enforcement mitigation measures were required. This work assessed a total of 54 signals. Twenty-six of these signals were in the Southern Cross Station to West Footscray area, which was defined as work package B of the RRL project.

Signals that had experienced a SPAD event^[13] on more than one occasion were chosen to be fitted with TPWS, as were those that V/Line ranked as ‘Potentially Severe’. Seventeen signals were ranked as ‘Potentially Significant’ and subjected to a cost-benefit analysis to determine if fitment of TPWS was cost effective. The ranking was determined using a ‘SPAD Ranked Risk Tool’ (SRRT), and a ranking of ‘Potentially Significant’ was evaluated to be equivalent to a ‘medium risk’ in accordance with the rankings used by the V/Line Enterprise-Wide Risk Model (EWRM).^[14]

The SRRT was made up of three elements: an initial collision potential assessment, an accident vulnerability ranking, and a final risk ranking score. For signal SST535, the analysis of the SRRT considered a potential collision type of ‘Plain line rear-on or buffer’.

V/Line used the Signal Passed at Danger Assessment Model (SPADAM) to assist in the ongoing assessment and control of risks associated with trains passing signals at danger. An assessment using the model that was dated 1 December 2020 was provided by V/Line as being that applicable to SST535 at the time of the SPAD on 23 November 2020. The signal had been ranked as the 50^th highest for risk. Approximately 250 signals had been assessed using the SPADAM tool by around the time of the incident. Signal SST535 had obtained a semi-quantitative risk exposure score of 1940, which fell below the normal V/Line threshold of 2500 for increasing the priority of introducing further risk controls.

Figure 7: Signal SST535 at stop

Source: CITS, observed post-incident

Other SPADs at or near SST535

Signal SST535 was located on the ‘Down Dual Gauge’ line, one of five bi-directional lines leading to platforms 1-5 at Southern Cross Station. Parallel to SST535 were three other mainline signals (SST533, SST527 and SST507). All were located at ground level and under the LaTrobe St bridge, as shown in Figure 3.

On 22 December 2017, a SPAD at signal SST533 was reported. The driver of train 8921 had received a medium speed warning aspect on the preceding signal. Train 8921 passed signal SST533 at stop by approximately 255 m. The driver was reportedly distracted by a person walking near the track. The South Geelong to Southern Cross service (train 8742) was waiting for a proceed aspect on signal SST578, which briefly displayed a proceed aspect before reverting to stop. The driver of train 8742 then saw train 8921 approaching in head-to-head conflict and flashed the headlight and sounded the whistle. Train 8921 stopped approximately 60 m from train 8742.

Eight SPADs^[15] at signal SST507 involving shunting moves were reported between 11 December 2014 and 16 January 2021. A SPAD at signal SST535 involving a shunt move was also reported on 23 December 2019.

Risk management

Overview

The V/line Safety Management System (SMS)^[16] had risk controls in place to manage the risk of a train-to-train collision because of a SPAD. Additional risk controls had been identified but not adopted at the time of the incident.

Procedural SPAD controls

V/Line applied the requirements of the 1994 Book of Rules and Operating Procedures for train operations. Section 2 rule 14b stated that no train must pass a home signal showing a stop aspect except under certain circumstances, which did not apply for this incident.

Engineering controls

SPAD Detection (in conjunction with radio communication)

The train control system at Southern Cross was equipped with SPAD detection such that an alarm was activated at the signaller’s workstation in the signal box if a train passed a controlled signal^[17] at stop. Therefore, the system had the potential to mitigate the consequences of a SPAD occurrence by the signaller making a radio broadcast requesting trains to stop (as was the case in this instance).

The V/Line Book of Rules and Operating Procedures (revision 7) referred to a SPAD alarm in section 36, rule 14, part g. This rule defined the steps a controlling signaller must take on observing a SPAD alarm in relation to a SPAD at a TPWS location. SST535 was not a TPWS location and the V/Line Book of Rules and Operating Procedures did not define actions to be taken on observing a SPAD alarm for non-TPWS locations.

Train Protection and Warning System

Trains 8239 and 8156 were equipped with TPWS equipment that would activate a brake application where a train passed a signal at stop and the signal was linked to track-mounted TPWS transmitters. The system was also capable of enforcing medium speed signal aspects and train speeds through junctions at chosen locations. The V/Line passenger train fleet was equipped with this system to facilitate operation on the Regional Fast Rail (RFR) network which began operations with a maximum speed of 160 km/h in 2005. TPWS was extensively utilised on the V/Line network however signal SST535 was not equipped with TPWS at the time of the incident.

Flank protection

Flank protection is defined as protection from overrunning movements approaching on converging tracks, usually by additional point interlocking or train detection.

The V/Line standard for signal principles (NIST-12.0, revision 3, dated 10/8/2020) referenced catch protection as providing physical protection of main line moves from the unauthorised movement of vehicles from sidings, yards, or maintenance facilities. It is described in the standard as a form of flank protection^[18] which is provided to prevent a train from being hit in the side by another train. No other references to flank protection were provided in the standard.

Flank protection is often provided through the setting of signals and points. In Figure 8: for example, a route from signal SST578 is at proceed for train A. Points 421 are set to the normal (straight through) position so that if train B proceeds past signal SST572, it is not routed into the side of train A. That is, the flank of train A is protected by points 421 set normal and routing train B straight ahead rather than towards the junction.

Figure 8: Simplified track layout for the relevant tracks at Docklands

Image shows a simplified representation of the track layout at Docklands with a route called from signal SST578 (green).

Source: ATSB and CITS

However, for the SPAD that occurred on 23 November 2020, similar protection was not available for an unauthorised movement past signal SST535 since there were no facing points to route the movement of train 8239 away from the junction.

Flank-track protection

Where flank protection cannot be provided by setting flank points, the provision of flank-track protection can be considered. If a route is set over a junction, flank-track protection requires the tracks between a signal protecting a converging route and the junction to be clear before the signal will display a proceed aspect.

Figure 9 provides an example of a route called from signal SST578 for train A. However, if train B passes signal SST535 at stop (SPAD) and occupies a track circuit between the signal and the junction (tracks 419T and/or 401T), a potential conflict with train A would be imminent. If flank-track protection was provided, any unauthorised occupancy of track circuits 419T or 401T would revert signal SST578 to a stop aspect. This would provide the opportunity for the driver of train A to react and stop prior to the junction and the potential conflict with train B (braking distances permitting and assuming train A had not already passed signal SST578).

Figure 9: Simplified track layout for the relevant tracks at Docklands

Image shows a simplified representation of the track layout at Docklands with a flank-track occupied by train B. While a route has been called from signal SST578 for train A, flank-track protection would place signal SST578 at stop.

Source: ATSB and CITS

However, the risk control of flank-track protection was not provided for the section of track where the SPAD occurred on 23 November 2020.

Train Vigilance System

Trains 8239 and 8156 were both equipped with vigilance control systems that verified the driver was not incapacitated by monitoring task-linked activities. In the absence of any such activities, the systems would provide intervention through alarms and subsequently by applying the train’s brakes. In this instance, recordings from both trains indicated that the drivers were providing active task-linked inputs and there were no activation of the trains’ brakes by the vigilance systems.

__________

8. VicTrack is the trading name of Victorian Rail Track Corporation, a Victorian Government state-owned enterprise which owns the track where the incident occurred.
9. Cloud cover is measured in eighths or oktas.
10. Cloud observations are from Melbourne Airport, temperature and wind are from Melbourne Olympic Park.
11. A system of train operation that prevents more than one train being in the block section at any one time.
12. TPWS stops a train by automatically initiating a brake demand, where TPWS track and train-borne equipment is fitted, if the train passes a signal at danger without authority.
13. Some signal locations were retained at Southern Cross during the RRL project.
14. The output from the SPAD Risk Ranking Tool ranked the risks as ‘severe’, ‘significant’ or ‘no significant risk’, which mapped onto the V/Line risk ranking of ‘high’, ‘medium’ and ‘low’ respectively.
15. Reported as either driver misjudged, completely missed, or limit of authority missed by train crew, in accordance with the reporting requirements of the Office of the National Rail Safety Regulator (ONRSR) at the time of reporting.
16. It is a requirement of Rail Safety National Law for accredited rail transport operators have an appropriate SMS in place.
17. A signal that is, or may be, controlled or operated by a signaller, network control officer or competent worker.
18. RISSB AS 7711:2018 Signalling Principles, listed flank protection as a means of mitigating the risk of collision owing to rolling stock exceeding its limit of authority. Flank protection was defined in the RISSB glossary as protection from overrunning movements approaching on converging tracks, usually by additional point interlocking or train detection.

Safety analysis

Driver distraction

Interview comments revealed that the driver was distracted by task unrelated thoughts on the approach to signal SST535. While there is diversity in the definition of distraction, a version commonly referenced is:

the diversion of attention away from activities critical for safe driving toward a competing activity.^[19]

Attention is conceived as a focusing response to a stimulus or task that reflects a state of arousal or concentration.^[20] Studies indicate that attention paid to a particular stimulus or task generally occurs in the context of competition among multiple stimuli or tasks for limited processing capacity.^{[21] [22]} Multiple stimuli or tasks that make simultaneous demands on an individual’s central processing mechanism will tend to interfere with each other. These demands can come from a range of factors either inside or outside a vehicle, that draws on the limited physical, visual, and cognitive resources, resulting in a degradation of the driver’s performance. Thought intrusions, or mind-wandering, can be a type of distraction. In train driving, the inherent propensity to mind-wandering can generate an unacceptable level of operational risk.

Should one or more of these competing demands be of sufficient magnitude to interfere with or divert attention from the original focus of attention, then the individual becomes distracted.^[23] It is within the context of attention that the process of distraction occurs. According to Nelson et al. (1993) this involves (i) a primary task (ii) a secondary or distracting stimulus or task (i.e., distractor) and (iii) the diversion of attention in response to the secondary task.

The driver of train 8239 reported being distracted by a personal issue on the afternoon of the incident and that they were reminded of this by a passenger who had just missed the departure of the train from Southern Cross Station

Distraction is strongly linked to decrements in driving performance and a higher risk of accident by motor vehicle drivers.^{[24] [25]} While rates of distraction amongst train drivers in Australia are not available, research from the U.K.^[26] of 1021 reported SPADs between 2006 and 2009 on Network Rail managed infrastructure found approximately one third of SPADs in this period were associated with some form of distraction or inattention.^[27]

The driver of train 8239 reported they sighted another stop signal further down the track (signal SST991 at Dudley St) and associated this signal with the warning provided by SST537 that the next signal would be at stop. Attention plays an important role in visual inspection strategy, especially in planning eye movements either toward locations preselected by expectations or toward an object that automatically attracts attention.^[28] Research has found that mind-wandering is associated with horizontal narrowing of motor vehicle drivers’ visual scanning process.^{[29] [30]}

During their training, V/Line trainee train drivers were made aware of the importance of controlling distraction and advised of factors which can lead to distraction. Drivers were also made aware of techniques to manage distraction, and to refocus after distraction. This information was also published by V/Line in their Professional Driving Booklet. While these techniques to control distraction are not mandatory, drivers were encouraged to determine or develop the ones that worked best for themselves.

In this incident the distraction was at a critical time after departure and when the train was approaching signal SST535. While the driver was familiar with techniques to manage distraction and reported having acknowledged the warning signal before signal SST535, this was insufficient in preventing the SPAD.

Reading through signal SST535

Reading through or reading across to another signal was listed as a common cause of passing a signal at stop in the V/Line Professional Driving Booklet. Often, the scenario considered is not stopping at a signal due to sighting a proceed signal beyond the stop signal and incorrectly acting in response to the proceed signal. However, just as relevant is the scenario that existed in this case. The driver advised not observing SST535 and intending to stop at another signal beyond.

To assist in the assessment and control of risks associated with trains passing a signal a danger, V/Line used the Signal Passed at Danger Assessment Model (SPADAM). For the signal being assessed, information relating to the signal location, signal characteristics, rail traffic volumes, consequence factors (such as track speed), and control measures were inputs to the model. As an output, the model provided a ‘likelihood score’, ‘exposure score’, and ‘consequence score’ which were combined to provide an overall risk exposure score for the signal.

In relation to the input of ‘signal sighting’, there were 13 factors considered, including ‘read across (adjacent line/track)’ and ‘read through (same line/track). Signal SST535 had been assessed against the model on a number of occasions. For the assessment of January 2020, ‘read across’ had been assigned a risk score of five, corresponding to ‘definite potential to read across’. Signal SST535 had been assigned a risk score of zero for ‘read through’, corresponding to there being no potential to read through to another signal. Had the signal been assessed as having a ‘definite potential to read through’, then the risk score for the signal would have increased from 1940 to 2116 and would have still been below the normal V/Line threshold of 2500 for increasing the priority of introducing further risk controls.

Based on the risk score calculated for signal SST535, TPWS was identified as a potential control for further investigation. The model identified ‘adequate length of signal overlap’ and ‘SPAD alarm and signaller response’ as existing signal location control measures.

Flank and flank-track protection

At junctions, flank protection is a means of preventing collisions of trains by setting points to prevent unauthorised movement of another train from also occupying the junction. That is, points are set to route the unauthorised movement away from the junction. In the SPAD incident of 23 November 2020 such flank protection was not available for an unauthorised movement past signal SST535 since there were no facing points to route the unauthorised movement of train 8239 away from the junction.

Where flank protection cannot be provided, flank-track protection can be considered. If a route is set over a junction, flank-track protection requires the tracks between a signal protecting a converging route^[31] and the junction to be clear before the signal will display a proceed aspect.

When train 8239 passed signal SST535 at stop and continued towards the junction, signal SST578 continued to display a proceed aspect for train 8156. Consequently, both trains were on a converging path and the risk of collision was significantly increased. Had flank-track protection been included in the signalling design, any SPAD at signal SST535 would have restored signal SST578 to stop. In this instance, train 8156 was relatively close to signal SST578 when the SPAD occurred at signal SST535, so there may have been insufficient time to stop prior to passing the signal. However, the installation of flank-track protection at this location would have reduced the risk of collision.

Infrastructure standards pertinent to this event did not specifically reference the potential application of flank-track protection. V/Line signalling standards did not call for flank-track protection to be considered. 

Although not published at the time of the 2014 commissioning of new infrastructure at this location, subsequent RISSB standards^[32] also did not reference the potential control of flank-track protection,^[33] although the control of flank protection was referenced in RISSB AS 7711:2018 Signalling Principles, and RISSB AS 7724:2020 Unauthorized movement protection - Operational requirements.

Flank-track protection was referenced by some infrastructure managers in Australia. Signalling principles^[34] for the Adelaide network required flank protection to be provided where a risk assessment determined the track layout warrants additional protection. The standard stated:

Where it is not practicable to provide flank protection by setting flank or trap points, and a significant safety benefit would arise, the provision of flank track section overrun detection at vulnerable signals should be considered. It may be initiated by overlap track section, or treadle, occupied without signal having cleared, or by sequential operation of track sections (e.g. overlap track occupied after berth track occupied). It may effect automatic replacement of conflicting signals…

An inquiry^{[35] [36]} into the Ladbroke Grove (U.K.) incident on 5 October 1999 considered the availability of automatic replacement of a signal at danger where a SPAD had occurred, and the layout was such that there was a significant danger of collision. A recommendation of the inquiry resulted in a Railway Group Standard being produced for U.K. railway infrastructure managers^[37] which stated:

Overrun detection shall usually be provided so that where a train overruns a stop signal which protects an area of conflict, other stop signals protecting the same area of conflict are automatically placed or maintained at danger. The replacement of signals to danger shall take place as quickly as possible after an overrun occurs.

Authority-overrun protection at SST535

Risk review for the initial installation in 2014

The risk review to determine fitment of TPWS considered a potential collision type of ‘Plain line rear-on or buffer’ for a SPAD at signal SST535. A potential collision input for the model of a head-on or side-on collision, that may have been experienced on 23 November 2020, was not considered for the 2014 analysis of a SPAD at signal ST535. Using the semi-quantitative input rankings of the model, a ‘head-on’ collision type would have resulted in a higher overall risk ranking score for signal SST535. If a ‘head-on’ potential collision type had been attributed to signal SST535, the resulting consequence potential score and final risk ranking score would have placed signal SST535 in the ‘Potentially Severe’ category, and the signal would probably have been equipped with TPWS.^[38]

TPWS proposal of 2020

V/Line provided documentation to the investigation that showed that in January 2020 V/Line costed a series of projects for installing TPWS, one of which was for the Southern Cross area and included signal SST535. The proposal was based on Southern Cross being identified as a high risk SPAD location and included input from the SPADAM model.^[39] V/Line had commenced the funding process, but it had not been finalised at the time of the SPAD in November 2020.

Radio communications

Instructions for radio communications during an emergency were provided in documents SAPR-78 Verbal Safety Critical Communications Protocol and train control training unit NCT5, titled Communications. Both documents allowed for an emergency message to be sent when the passage of a train is endangered and require an emergency message to be answered immediately. Both documents stated that an emergency message is commenced with the phrase “Emergency, Emergency, Emergency.”

The V/Line Book of Rules and Operating Procedures (revision 7), Safety Critical Communications Protocol, and the training unit provided generic advice and allowed for a broad range of scenarios. They did not provide specific instructions on communicating with multiple trains in an emergency. The V/Line Verbal Safety Critical Communications Protocol stated that Rail Safety Workers must use the principle of safety critical communications being accurate, brief, and clear.

The signaller promptly commenced a radio broadcast to the driver of train 8156 on observing CCTV of train 8239 approaching signal SST535 and anticipating that the train would not stop at the red signal. This broadcast was heard by train 8239 and resulted in the trains receiving enough warning time to stop short of each other.

Other options were available to the signaller including a broadcast directed at both trains simultaneously, or a broadcast for all trains to stop. V/Line procedures do not provide specific advice on the preferred option in such a scenario.

The signaller stated later at interview that they were aware of the requirements concerning an emergency message, though during the incident used the word ‘urgent’ rather than ‘emergency’ to warn the driver of train 8156.

__________

19. Lee JD, Young KL, and Regan, MA (2008) Defining driver distraction. In: Regan MA, Lee JD, Young .L (Eds.) Driver Distraction: Theory, Effects, and Mitigation, CRC Press Taylor & Francis Group, Boca Raton, FL, USA.
20. Berlyne DE (1960) Conflict, arousal and curiosity, McGraw-Hill, New York.
21. Broadbent DE (1958) Perception and communication, Pergamon Press, New York.
22. Kahneman D (1973). Attention and effort, Prentice Hall, New Jersey.
23. Nelson JE, Duncan CP & Kiecker PL (1993) ‘Toward an understanding of the distraction construct in marketing’, Journal of Business Research, 26, 201-221
24. Klauer SG, Dingus TA, Neale VL, Sudweeks JD, and Ramsey DJ (2006) The Impact of Driver Inattention on Near-Crash/Crash Risk: An Analysis Using the 100-Car Naturalistic Driving Study Data, Report No. DOT HS 810 594. National Highway Traffic Safety Administration, Washington DC.
25. Galera C, Orriols L, M’Bailara K, Laborey M, Contrand B, Ribéreau-Gayon R, Mason F, Bakiri S, Gadbaude C, Fort A, Maury B, Lemercier C, Cours M, Bouvard M, Lagarde E (2012) ‘Mind wandering and driving: responsibility case-control study’, BMJ, 345.
26. Luke T, Heavisides J, and Basacik D (2013) Management of Distraction Risk from Mobile Phones in the UK Rail Industry. Driver distraction and inattention. Advances in Research and Countermeasures. Ashgate, UK.
27. In comparison, during 2020, 563 SPADs where the limit of authority was passed by a train crew were reported to ONRSR by train operators in Australia. However, the number because of distraction were not available as this information is not required to be reported. Sighted at Office of the National Rail Safety Regulator (n.d.) national-safety-data, ONRSR website, accessed 2 July 2021.
28. Henderson JM (1993) ‘Visual Attention and Saccadic Eye Movements in Complex Visual Tasks’, Behavioural and Brain Sciences, 16(3), 579-580.
29. He J, Becic E, Lee Y, and McCarley JS (2009) ‘Identifying Mind-wandering Behind the Wheel’, Proceedings of the Human Factors and Ergonomics Society 53rd annual meeting.
30. Recarte M, Nunes L (2000) ‘Effects of Verbal and Spatial-Imagery Tasks on Eye Fixations While Driving’, Journal of Experimental Psychology: Applied, Vol. 6, No. 1,31-43.
31. A signal protecting a converging route, or an opposing train movement, may also be referred to as a conflicting signal.
32. RISSB is accredited as a Standards Development Organisation, and all new standards commenced by RISSB after 31 July 2007 were published as Australian Standards.
33. AS 7711 does refer to the case of overlap protection on a single line, stating that the overlap should extend from the home signal as far as the opposing main and loop starting signals controlling the entrance to the single line block and should incorporate loop and flank protection. This could be considered a specific and basic case of flank-track protection, but it is not referred to as such.
34. As stated in Government of South Australia Rail Commissioner (2017) South Australia Signalling Principles and Practices for the Adelaide Metropolitan Passenger Rail Network, Engineering Standard, Adelaide.
35. Cullen Rt. Hon Lord (2000a) The Ladbroke Grove Rail Inquiry: Volume 1, Health and Safety Commission, London.
36. Cullen Rt. Hon Lord (2000b) The Ladbroke Grove Rail Inquiry: Volume 2, Health and Safety Commission, London.
37. Rail Safety Standards Board (2000), Railway Group Standard RGS GK/RT/0064 Issue 1, Provision of overlaps, flank protection, and trapping, London.
38. Rail Safety Standards Board (2000), Railway Group Standard RGS GK/RT/0064 Issue 1, Provision of overlaps, flank protection, and trapping, London.
39. This V/Line SPADAM analysis considered the risk of a side-on collision.

Findings

From the evidence available, the following findings are made with respect to passenger train 8239 passing signal SST535 at danger and near collision with another passenger train at Docklands, Victoria, on 23 November 2020.

Contributing factors

• The driver of train 8239 was probably distracted when approaching ground-mounted signal SST535 and did not notice it displaying a stop indication. The driver probably noticed a gantry-mounted stop signal located further along the track and continued past signal SST535.
• The absence of authority-overrun protection (such as TPWS) at signal SST535 increased the potential consequences of a SPAD. (Safety issue)
• The V/Line risk assessment for signal SST535 that was conducted in 2014 did not consider a head-on or side-on collision as a credible scenario. As a result, TPWS was not considered a necessary control at the time of commissioning.

Other factors that increased risk

• Signal SST578 did not restore to danger when train 8239 SPAD signal SST535. With signal SST578 at proceed, the opportunity for the signalling to provide a warning to train 8156 of a potential conflict condition was not available.
• V/Line signalling standards did not identify flank-track protection as a control to prevent collision because of a SPAD, and none was installed at the incident location. The absence of flank-track protection increased the risk of side-on collision at the junction of track SST416T.
• RISSB signalling standards did not identify flank-track protection as a potential control to prevent collision because of a SPAD.

Other findings

• The signaller promptly broadcast a warning to the driver of train 8156 to stop their train, preventing a potential collision.
• There was no specific advice in the V/Line radio communication directions for a signaller to direct multiple trains in an emergency.

Safety issues and actions

Absence of authority-overrun protection

Safety issue number: RO-2020-019-SI-01

Safety issue description: The absence of authority-overrun protection (such as TPWS) at signal SST535 increased the potential consequences of a SPAD.

Additional safety actions

Additional safety action by V/Line Corporation

Coaching and a return-to-work plan has been implemented for the driver of train 8239, and the signaller has undertaken additional communications training.

Additional safety action by RISSB

RISSB advised that flank - track protection has been recorded for inclusion when the signalling principles standard AS 7711 is next under review.

Glossary

AS                    Australian Standard

CCTV               Closed-circuit television

DMU                Diesel Multiple Unit

EWRM             Enterprise-Wide Risk Model

LED                 Light-emitting diode

LRS                 Local Radio System

ONRSR            The Office of the National Rail Safety Regulator

RFR                 Regional Fast Rail

RISSB              Rail Industry Safety and Standards Board

RRL                 Regional Rail Link

SMS                 Safety Management System

SPAD               Signal Passed at Danger

SPADAM          Signal Passed at Danger Assessment Model

SRRT               SPAD Ranked Risk Tool

TCS                 Train Control System

TPWS              Train Protection and Warning System

UK                   United Kingdom of Great Britain and Northern Ireland

VDU                 Video Display Unit

Sources and submissions

Sources of information

The sources of information during the investigation included the:

• V/Line
• V/Line personnel
• Telstra
• Event recorders from trains 8156 and 8239
• Signalling event logger
• Radio broadcast recordings
• Bureau of Meteorology
• CCTV footage from Southern Cross Station
• Photographs taken on the day of the incident

References

Berlyne DE (1960) Conflict, arousal and curiosity, McGraw-Hill, New York.

Broadbent DE (1958) Perception and communication, Pergamon Press, New York.

Cullen Rt. Hon Lord (2000a) The Ladbroke Grove Rail Inquiry: Volume 1, Health and Safety Commission, London.

Cullen, Rt. Hon Lord (2000) The Ladbroke Grove Rail Inquiry: Volume 2, Health and Safety Commission, London.

Galera C, Orriols L, M’Bailara K, Laborey M, Contrand B, Ribéreau-Gayon R, Mason F, Bakiri S, Gadbaude C, Fort A, Maury B, Lemercier C, Cours M, Bouvard M, Lagarde E (2012) ‘Mind wandering and driving: responsibility case-control study’, BMJ, 345, doi: https://doi.org/10.1136/bmj.e8105.

Government of South Australia Rail Commissioner (2017) South Australia Signalling Principles and Practices for the Adelaide Metropolitan Passenger Rail Network, Engineering Standard, Adelaide.

He J,  Becic E, Lee Y, and McCarley JS (2009) ‘Identifying Mind-wandering Behind the Wheel’,  Proceedings of the Human Factors and Ergonomics Society 53rd annual meeting, doi:10.1177/0018720810391530

Henderson JM (1993) ‘Visual Attention and Saccadic Eye Movements in Complex Visual Tasks’, Behavioural and Brain Sciences, 16(3), 579-580. doi:10.1017/S0140525X00031721

Kahneman D (1973). Attention and effort, Prentice Hall, New Jersey.

Klauer SG, Dingus TA, Neale VL, Sudweeks JD, and Ramsey DJ (2006) The Impact of Driver Inattention on Near-Crash/Crash Risk: An Analysis Using the 100-Car Naturalistic Driving Study Data, Report No. DOT HS 810 594. National Highway Traffic Safety Administration, Washington DC.

Lee JD, Young KL, and Regan, MA (2008) Defining driver distraction. In: Regan MA, Lee JD, Young .L (Eds.) Driver Distraction: Theory, Effects, and Mitigation, CRC Press Taylor & Francis Group, Boca Raton, FL, USA.

Luke T, Heavisides J, and Basacik D (2013) Management of Distraction Risk from Mobile Phones in the UK Rail Industry. Driver distraction and inattention. Advances in Research and Countermeasures. Ashgate, UK.

Nelson JE, Duncan CP & Kiecker  P L (1993) ‘Toward an understanding of the distraction construct in marketing’, Journal of Business Research, 26, 201-221,  https://doi.org/10.1016/0148-2963(93)90032-K

Office of the National Rail Safety Regulator (n.d.) national-safety-data, ONRSR website, accessed 2 July 2021.

Rail Safety Standards Board (2000), Railway Group Standard RGS GK/RT/0064 Issue 1, Provision of overlaps, flank protection, and trapping, London.

Rail Safety National Law (SA) Act 2012, part III division 1.

Rail Industry Safety and Standards Board (2018), AS 7711:2018 Signalling Principles, Rail Industry Safety and Standards Board, Queensland.

Recarte M, Nunes L (2000) ‘Effects of Verbal and Spatial-Imagery Tasks on Eye Fixations While Driving’, Journal of Experimental Psychology: Applied, Vol. 6, No. 1,31-43.

V/Line (2017) SAPR-78 Verbal Safety Critical Communications Protocol.

V/Line (2108) Train control training unit NCT5: Communications, TLI42215 Certificate IV in Rail Network Control.

V/Line (2020) 1994 Book of Rules and Operating Procedures (revision 7).

V/Line (2020) Signalling principles, NIST-12.0.

V/Line (n.d.) Professional Driving Booklet MSR01345

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• V/Line
• Driver of train 8239
• V/Line signaller on duty at Southern Cross at the time of the incident
• ONRSR
• RISSB

Any submissions from those parties will be reviewed and, where considered appropriate, the text of the draft report will be amended accordingly.

Purpose of safety investigations & publishing information

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2021 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

Safety summary

What happened

On 27 October 2020, train 2PM9, operated by SCT Logistics, departed Perth, Western Australia for Melbourne, Victoria. Near Koolyanobbing, Western Australia, a wheel fractured and disintegrated on the 58th vehicle on the consist, wagon ARFY2253s, resulting in a derailment. The derailed wagon and those trailing separated from the train, but no other wagons derailed. There were no injuries.

What the ATSB found

Detailed material examination identified that a transverse fatigue crack initiated in the vicinity of a spalling defect in the wheel tread prior to the incident journey. The crack propagated into the rim and flange of the wheel. On the day of the incident, a skidding event at the fatigue crack likely induced rapid brittle cracking in the wheel. This resulted in its fragmentation and subsequent derailment.

In the 12 months prior to the occurrence, ARFY2253s experienced various issues with brakes and overheated wheels, although it is not clear whether any of these issues contributed to the derailment.

The failed wheel was recently reprofiled, but the maintenance provider’s work instruction did not require non-destructive examination beyond a visual inspection to identify any remaining defects.

What has been done as a result

The operator and maintenance provider have developed a process for monitoring wagons that experience repeated issues, such as with braking. Inspection procedures for thermal cracks and overheated wheels have been refined, and non-destructive testing is being investigated for use in certain wheel inspections.

Safety message

Skids and hotspots may be repairable on wheels in otherwise good condition, however, they can induce wheel failure if cracks are present, as in this case. Diligent inspections and non-destructive testing can be useful for detecting and monitoring wheel cracks, which is particularly important as these cracks approach the wheel wear condemning limit.

The investigation

The occurrence

On the morning of 27 October 2020, train 2PM9, operated by SCT Logistics, departed Perth, Western Australia for Melbourne, Victoria. The train was 1,658 metres long, and consisted of 2 locomotives hauling 68 wagons, with a trailing load of 3,690 tonnes. At 0816 Western Standard Time,^[1] the train was approaching a restricted speed zone and the driver applied the automatic brakes. At 0841, while approaching Koolyanobbing, the trailing left wheel on wagon ARFY2253s fragmented, and two large pieces of the wheel were ejected. ARFY2253s was the 58th wagon in the consist. The wagon’s braking system used a WF2 triple valve, and the wagon was unloaded with a tare mass of 30.84 tonnes. The train continued without incident, and the driver applied the dynamic brake in preparation for an upcoming turn. At 0843, approximately 3.5 km beyond the initial wheel fragmentation, a third piece of the wheel was ejected, and the wagon derailed (Figure 1).

Figure 1: The derailed wheelset on ARFY2253s

Source: SCT Logistics

At this point, the crew observed dust towards the rear of the train, and applied both the dynamic and automatic brakes until the train came to rest at 0846. Upon inspection it was found that ARFY2253s and the trailing ten wagons had separated from the train. No other wagons derailed. Track damage was observed over a distance of 2.4 km between the wagon derailment and its stopping point.

Context

Wheel fracture examination

The derailed wheelset and the ejected wheel pieces were recovered from the site by Gemco Rail (Gemco), SCT Logistics’ primary maintenance provider. Gemco contracted independent consultants Bureau Veritas to conduct a metallurgical examination of the derailed wheelset in order to determine that nature of the failure, and any aspects that may have contributed.

Figure 2 shows the recovered fragments of the rim, labelled 1-3, and the three fractures that separated the different pieces, labelled A‑C. Fragments 1 and 2 were the first to be ejected from the wheel during the incident journey, with fragment 3 ejected after the brakes were applied.

Figure 2: Reassembled wheel fragments (left) and wheel hub (right)

Source: Bureau Veritas, annotated by ATSB

The Bureau Veritas analysis identified a fatigue crack at the surface of the wheel, within fracture A (Figure 3). The fatigue crack presented as a darkened region on the fracture surface, with distinct beachmarks indicating crack propagation into the wheel rim and flange from the initiation point on the surface. The fatigue appeared to have propagated through approximately 40 per cent of the wheel rim’s cross section. Bureau Veritas made no comment on the age of the fatigue crack.

Figure 3: Fracture A on fragment 1 of the wheel

Source: Bureau Veritas, modified by ATSB

The Bureau Veritas report concluded that the wheel failure originated from this fatigue crack. At some point, the fracture mode changed from fatigue to brittle cracking, indicated by the colour and surface texture change between the fatigue crack and the rest of the wheel cross section. The brittle cracking progressed towards the wheel hub. It then propagated circumferentially around the wheel, branching out to the wheel surface, resulting in the ejection of the three wheel fragments.

A heat tint mark was also present on the fracture surface. Heat tinting such as this is indicative of oxidation, a result of freshly exposed steel that is subjected to heat and oxygen. The Bureau Veritas examination also identified skid marks on the wheel surface near fracture A.

In addition to skid marks, the wheel surface near the fatigue crack had multiple spalls and gouges (see Figure 4). Fluorescent magnetic particle inspection also revealed multiple transverse and longitudinal cracks around these surface defects. Several cracks were found that extended below the wear condemning limit^[2] of the wheel rim.

Figure 4: Wheel surface adjacent to fracture A

The left image shows the various spalling and gouging observed on the wheel surface. The right image shows numerous surface cracks under fluorescent magnetic particle inspection. Source: Bureau Veritas, modified by ATSB

Hardness testing was performed on several sections of the wheel. The heat-affected regions showed some surface hardening outside of the specification for Class B^[3] forged steel wheels such as these, although this would be expected in the event of a skid. Some areas within the wheels were slightly under the minimum specified hardness, but the report did not find this to have contributed in any way to the wheel failure, and since these regions were below the wear condemning limit, this was permitted by the wheel specification.

Wheel maintenance and inspections

In the 12 months before the derailment, wagon ARFY2253s experienced a variety of issues requiring maintenance, including brake defects and uneven wheel wear. Wheelsets were replaced twice as a result. The most recent wheelset change was 4 months before the derailment, when the incident wheelset 92709 was installed on the wagon due to overheated wheels.

Prior to installation on the wagon, wheelset 92709 was re-profiled. The rim thickness was reduced to 34 mm from 44 mm to remove hollowing and spalls that were observed on the wheel tread. Gemco’s wheel reprofiling work instruction was based on the Association of American Railroads Specifications Manual. Regarding the removal of defects such as spalling and shelling, Gemco required a visual examination following wheel reprofiling, but neither document required any non‑destructive testing, such as fluorescent magnetic particle inspection.

The Australian Standard for wheel maintenance current at the time of writing (AS 7514:2018) referred to the Rail Industry Safety and Standards Board’s Code of Practice – Wheel Defects. Regarding the reprofiling of spalled wheels, it stated:

Rectify spalling or shelling by machining to a sufficient depth (3 mm minimum) to completely eliminate it, and carry out crack detection after machining.

Crack detection was not explicitly defined in the document, but it included visual inspection, magnetic particle inspection and dye penetrant inspection as examples.

Following the installation of wheelset 92709, dragging brakes were reported on the wagon on two occasions. The wheels were inspected on the second occasion and no defects were noted.

SCT Logistics investigation

As a result of the derailment, SCT Logistics contracted an independent investigator to determine any potential factors that led to the derailment. In the resulting investigation report it was noted that a wheel flat was present at fracture A, and the flat’s worn edges indicated that it may have occurred some time ago, on a previous journey. The report examined readings from wayside detectors on the day of the derailment and from previous journeys, but found no evidence of a wheel flat being detected.

The independent investigator’s report found that the fatigue crack at fracture A resulted in the wagon’s derailment. It determined that the crack originated from a spall on the wheel’s surface. The report also stated that the brittle fracture of the wheel was likely initiated by the braking event prior to entering the restricted speed zone during the incident journey.

The report stated that wagon and bogie maintenance were ‘…considered to be a major contributory factor in this derailment event.’ It highlighted previous brake-related maintenance, and suggested that it was indicative of unresolved braking issues. The report attributed the observed wheel defects, including flat-spotting, skids, spalls, and the fatigue crack to these braking issues.

The investigation also looked at train handling and crew performance during the incident journey, and found all actions to be in accordance with accepted procedures and good driving practice.

Related occurrences

ATSB Investigation (RO-2019-001)

On 6 January 2019, SCT Logistics freight train 6MP9 derailed near Cook, South Australia after a wheel fragmented on one of the wagons. Inspection of the failed wheel found that thermal damage resulted in the initiation of a fatigue crack that propagated into the wheel, ultimately leading to its fragmentation.

At the last inspection—carried out by Gemco—the flange crack was likely observable but was either not detected, or was deemed acceptable under the provided work instruction. This work instruction was less conservative than the Australian Standard, but it was not possible to establish whether compliance with the standard would have prevented the occurrence.

Following the derailment, SCT Logistics worked with Gemco to improve inspection processes for wheels with brake issues. SCT Logistics also instructed Gemco to only install Class B wheelsets in future, after the wheel manufacturer stated that Class C wheelsets—such as the one that failed—were more prone to thermal issues.

Safety analysis

Visual examination of the wheel fracture surfaces found that a fatigue crack initiated at the surface of the wheel tread, in the vicinity of a spalling defect. The fatigue crack propagated into the wheel tread and the flange, consuming approximately 40 per cent of the wheel rim cross section prior to failure. The dark colour of the fatigue region was indicative of corrosion in a low‑oxygen environment—before the crack opened at wheel failure. This suggested that the fatigue crack had existed for some time, and was certainly present before the incident journey. However, it was not possible to determine whether or not this crack existed in some form the last time the wheel was reprofiled or examined by maintenance personnel. Given that the fatigue crack might have been present when the wheel was last reprofiled (removing visible spalls), then the nearby spalling defect may have been unrelated.

The skid marks, flat spot and heat-affected regions at fracture A (Figure 2) all indicated that a skidding event occurred at this location, probably when the automatic brakes were applied before the restricted speed zone. This skidding event would have generated considerable heat, and the heat tint observed on the fracture surface is evidence that the brittle cracking occurred at the same time as this heat generation. It is therefore likely that the skidding event induced the brittle cracking at fracture A. Once the brittle cracking commenced, it quickly propagated through the wheel, resulting in the failure and subsequent derailment.

In SCT Logistics’ independent report, it was suggested that the flat spotting might have occurred on a previous journey, gone undetected by wayside monitoring, and even contributed to the fatigue crack. However, if this was the case, a second skidding event would had to have occurred at the same location during the incident run in order to generate the heat tint seen on the fracture surface.

Wagon ARFY2253s had a number of issues with brakes and overheated wheels in the 12 months leading up to the derailment. It is not clear whether these were resolved through maintenance actions, so it is possible that they contributed to the development of the fatigue crack as well as the skid during the incident journey. However, without additional evidence, it was not possible to determine the degree of contribution, if any.

Findings

From the evidence available, the following findings are made with respect to the derailment of SCT Logistics freight train 2PM9, near Koolyanobbing, Western Australia on 27 October 2020.

Contributing factor

• Prior to the incident journey, a transverse fatigue crack initiated in the vicinity of a spalling defect on the tread surface, and propagated into the wheel rim and flange.
• During the incident journey, a skidding event likely induced rapid crack growth from a pre‑existing fatigue crack. This resulted in fragmentation of the trailing left wheel and subsequent wagon derailment.

Other factor that increased risk

• In the 12 months prior to the occurrence, wagon ARFY2253s experienced various issues with brakes and overheated wheels, which can increase the risk of wheel failure.

Safety action

The operator and maintenance provider

As a result of this occurrence, SCT Logistics and Gemco have developed a process for monitoring wagons that experience repeated issues, such as with braking. Gemco staff are completing refresher courses on the field inspection of wheels, and inspection procedures for thermal cracks and overheated wheels have been refined. Non-destructive crack detection is being investigated for use on inspections following overheated or skidded wheels.

Sources and submissions

Sources of information

The sources of information during the investigation included the:

• SCT Logistics
• Gemco Rail
• Bureau Veritas
• the Rail Industry Safety and Standards Board.

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• the Office of the National Rail Safety Regulator
• SCT Logistics
• Gemco Rail
• Bureau Veritas
• the wheel manufacturer.

Submissions were received from:

• the Office of the National Rail Safety Regulator
• SCT Logistics
• Gemco Rail
• Bureau Veritas
• the wheel manufacturer.

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2021 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

__________

1. Western Standard Time (WST): Coordinated Universal Time (UTC) + 8 hours.
2. The wear condemning limit marks the maximum allowable wear in a wheel before it must be replaced
3. Class B is a specification defined by the Association of American Railroads. Wheel hardness and ductility varies with class. Class B wheels are softer than Class C wheels, with a higher fracture toughness.

Executive summary

What happened

In the afternoon of 4 November 2020, a training flight was conducted in an Aquila AT01, registered VH‑OIS (OIS) which departed Bankstown Airport with a Grade 1 flight instructor and a student onboard. The flight was conducted to assess the readiness of the student to complete the commercial pilot licence flight test for aeroplanes. The flight continued normally, proceeding to Coombing Park Aeroplane Landing Area (ALA), where the student demonstrated the conduct of a precautionary search, an expected exercise during this flight. However, following the precautionary search, a touch‑and‑go was performed. During the subsequent climb away from the runway, the aircraft collided with terrain. Both pilots received fatal injuries.

What the ATSB found

The ATSB found that pre-flight planning was likely not performed to identify if Coombing Park ALA was suitable for flight training operations. This placed more importance on the conduct of the precautionary search to identify the rising terrain hazard in the overshoot area of the runway. The precautionary search was conducted at a height and position that likely made assessing the hazard less effective, leading to the pilots deciding to conduct a touch-and-go landing and take-off toward the rising terrain beyond the runway end. Further, the take-off was conducted on an uphill slope with a probable tailwind. A standing take-off in the more favourable reciprocal direction would likely have cleared all obstacles and terrain.

Although the touch-and-go was typical of those conducted by the student, it likely used more runway than if a standing take-off was conducted, leading to the aircraft becoming airborne further along the runway, and closer to the rising terrain. Consequently, although the aircraft was likely performing normally, the aircraft had insufficient performance to outclimb the rising terrain in any direction once established on the initial climb after take-off.

The ATSB also found that the operations manual used for flight training that was based on the CASA sample operations manual allowed the use of any aerodrome (including ALAs) in the Enroute Supplement Australia to be used, although these were not assured to any operational standard. Further, the recommendations contained in CASA guidance CAAP 92-1(1) (although subsequently replaced by AC 91-02 as noted below) did not provide assurance that an aircraft would be able to outclimb rising terrain after take-off more than 900 metres from the runway end.

What has been done as a result

CASA advised that legislation and guidance relating to obstacle clearance in the approach and take-off areas of ALAs was introduced on 2 December 2021. Included in these changes were the replacement of CAAP 92-1(1) with Advisory Circular AC 91-02 titled ‘Guidelines for aeroplanes with MTOW not exceeding 5,700 kg – suitable places to take-off and land’ (the AC).

In contrast to CAAP 92-1(1), the AC included guidance for a pilot to consider and be aware of ‘lateral transition areas’ and ‘obstacles in the approach and climb-out flight paths’, in deciding if an aerodrome is suitable for operations. Further, the introduced regulations require that pilots ensure that the aeroplane has the performance to clear all obstacles by a safe margin until at the minimum height for flight.

Safety message

This investigation shows that the lift-off location following a touch-and-go is more variable and complex to predict than a standing take-off. In this case, the touch-and-goes conducted throughout the flight consistently used more runway than a standing take‑off. This contributed to the aircraft lifting off from a point where the available climb performance of the aircraft was not sufficient to outclimb the rising terrain. This characteristic is particularly important for flights to unfamiliar aerodromes, where the aerodrome characteristics should be considered to confirm if a touch‑and‑go is feasible, particularly for uncertified aerodromes. Additionally, as was the case here, the runway slope, wind and rising terrain at one runway end may mean that the optimal landing and take-off directions are in opposing directions, which cannot be achieved when conducting a touch-and-go.

The investigation also highlights the value of following operator’s procedures as an important risk control to aid planning and in-flight decision making. For this accident, the operator’s procedures did not allow landings to be conducted at the aerodrome where the accident occurred. This meant that the execution of the precautionary search was one of the last remaining defences to identify the rising terrain hazard during take-off, which was ineffective.

Operators should also be aware that aerodromes meeting the recommendations in the CASA guidance publication CAAP 92-1(1) (now obsolete) are not assured that an aircraft will be able to successfully climb away after take-off more than 900 metres past the runway end. The new performance-based recommendations of AC 91-02 now require operators to consider obstacle clearance beyond 900 metres. Flight training operators should also note that there are no standards for ALAs (listed in the ERSA as uncertified aerodromes). The published data for these uncertified aerodromes are potentially incomplete or inaccurate, including obstacle information. This means that for take-off from these aerodromes, the new guidance requires pilots and operators to know the climb gradient needed to clear all obstacles by a safe margin until the aeroplane reaches the minimum height for flight.

Summary video

Pre-flight and departure from Bankstown

In the afternoon of 4 November 2020, a training flight was conducted in an Aquila AT-01 aeroplane, registered VH‑OIS (OIS) between Bankstown Airport and Orange Airport, New South Wales (Figure 1). On board were a Grade 1 flight instructor (the instructor) employed by Soar Aviation (Soar), and a student (the student) of Soar. This flight was required by Soar to assess the readiness of the student to complete the Commercial Pilot Licence (Aeroplane) (CPL(A)) flight test administered by the Civil Aviation Safety Authority (CASA).

Figure 1: Flight path of VH-OIS and accident location

Source: Google Earth and Geoscience Australia, annotated by ATSB.

Automatic Dependent Surveillance Broadcast (ADS-B) data transmitted by OIS (the flight data)[1] indicated the aircraft started to taxi to a run-up[2] bay adjacent to Bankstown Airport runway 29R at 1419. The aircraft remained in the bay for about 5 minutes then, following engine run‑ups, the pilots[3] taxied OIS toward the runway 29R threshold, with take-off commencing at 1431 followed by an upwind[4] departure.

After departing the Bankstown Airport airspace, OIS initially tracked toward the west‑north‑west before turning left onto a west-south-westerly track about 19 km from Bankstown Airport, towards the planned track to Greenethorpe Township, about 220 km away. OIS continued tracking toward Greenethorpe until 1507, when the pilots turned right, tracking north-west toward a grassed area adjacent to the western side of the Blue Mountains.

Diversion to Orange

Flight data indicated that, once overhead the grassed area, the pilots conducted 4 left orbits[5] (blue track in lower centre of Figure 1) at an altitude between 6,000 and 7,000 ft above mean sea level (AMSL). After completion of the orbits, OIS was established on a track in the general direction of Orange township. At 1526, about 6 minutes after setting course toward Orange, the pilots of OIS contacted air traffic control and requested a flight plan amendment from the original track (to a location near Greenethorpe Township) to Orange Airport and back to Bankstown Airport, with no change to the SARTIME. The flight plan amendment and change in track was consistent with the instructor requiring the student to demonstrate an inflight diversion, as labelled in Figure 1.

At 1553, the pilots of OIS arrived overhead Orange Airport, and joined the crosswind leg of the circuit for runway 29 for the first of 4 touch-and-go landings and circuits, as shown in Figure 2. Flight data (shown by the green line in Figure 2) and eyewitnesses indicated that after the completion of the fourth touch‑and‑go on runway 29, as OIS was on the late crosswind circuit leg, a descending left turn was conducted, with the aircraft manoeuvred to align with the reciprocal runway 11. A touch-and-go landing was then conducted on runway 11 followed by a left circuit (dark blue line in Figure 2) and full stop landing on this runway at 1726. The pilots taxied to the main parking area at Orange, with flight data indicating the aircraft was stationary for about 1 minute. During this time, photographs taken by an eyewitness and CCTV footage showed the canopy partially opened and the student looking at a map, with other eyewitnesses indicating that the pilots spoke with one another.

Figure 2: Flight path of VH-OIS at Orange Airport

Source: Google Earth, annotated by the ATSB

After a total of 8 minutes on the ground, OIS entered the main runway from the eastern taxiway and backtracked toward the runway 11 threshold. At 1637, the pilots broadcast that they were rolling on runway 11 and an upwind departure was conducted. Flight data indicated that during the initial climb from the runway, OIS maintained a nominal track on the extended runway centreline for about 9 km.

Audio recordings of transmissions from the Orange common traffic advisory frequency (CTAF) indicated that during this time, both pilots of OIS were in communication with an inbound passenger carrying flight. Once at about 9 km from the runway, the pilots of OIS broadcast on the CTAF that they were on a heading of 130° magnetic and levelling out at 5,000 ft AMSL, consistent with the flight data. OIS turned right briefly toward the township of Blayney and then left to proceed on a south-easterly track for about 2 minutes before turning right onto a south-westerly heading in the general direction of the township of Carcoar and Coombing Park aeroplane landing area (ALA).

Operations at Coombing Park

At 1653 a descent was commenced when OIS was abeam Coombing Park ALA, approximately 3 km to the north of the main east-west runway (runways 07 and 25) as shown in Figure 3. The pilots conducted 3 nominally rectangular patterns around Coombing Park runway 07.

At 1704, while on the easterly leg of the third pattern, the instructor of OIS made a phone call to the property manager of Coombing Park that lasted 24 seconds. The property manager recalled the instructor identifying themselves as an instructor from Soar Aviation and requesting permission to conduct a touch-and-go at Coombing Park. The property manager reported that they thought OIS was on the downwind leg for runway 25, and on this basis approved the touch‑and-go.

The pilots positioned OIS to land on runway 07, with flight data showing the aircraft on a descending profile on the base, and final approach circuit legs. The property manager described the final approach as stable and observed OIS descending to very near the ground at the runway 07 threshold. OIS was not visible to the property manager after this time due to their vantage point being slightly below the runway, however the manager, who was also a pilot, indicated that the projected touch-down point would have been very close to the runway 07 threshold based on the observed approach.

Flight data indicated that OIS passed over the runway 07 threshold at 1708, touched down and rolled along the runway surface. About 30 seconds after passing over the runway threshold, OIS become airborne again and commenced a climb, passing through 50 ft AGL approximately 900 m from the start of runway 07.

Figure 3: Flight path of OIS while operating at Coombing Park ALA

Source: Google Earth and Aireon, modified and annotated by the ATSB

During the initial climb, OIS tracked in a straight line angled slightly left of the runway centreline. Flight data indicated that, about 20 seconds later as OIS climbed through 200 ft above the lift-off point, a climbing left turn was conducted by the pilots before flying straight again in the direction of a small dam, beyond which was a less wooded area. OIS passed over the small dam and collided with an embankment on the far side of the dam. A witness heard the sound of the impact, although they did not realise it was an aircraft accident at the time, and there were no eyewitnesses to the accident.

At 1800, air traffic services commenced trying to contact the pilots of OIS via radio. At 1832, after being unsuccessful in contacting the pilots, an INCERFA[6] was declared with the search and rescue response being transferred to the Joint Rescue Coordination Centre. Local authorities were notified, and a search was commenced in the area. At 2157, the wreckage of OIS was located, and it was confirmed that both pilots had sustained fatal injuries.

Context

Pilot information

Flight instructor

Qualifications

The instructor held a valid commercial pilot licence (aeroplane) issued in March 2009, and a flight instructor rating issued in March 2010. The instructor also held a grade 1 training endorsement to instruct in single- and multi-engine fixed wing aeroplanes.

The instructor was also endorsed to conduct training for the issue of instrument and night VFR ratings. They passed a proficiency check for the flight instructor rating on 13 January 2020 that was valid at the time of the accident. Additionally, the instructor’s flight crew licence indicated that they had completed single- and multi-engine aeroplane flight reviews on 13 January 2020 and 28 July 2019 respectively that were also valid.

Experience

A review of the instructor’s logbooks, staff files and flight data showed that at the time of the accident they had accumulated a total flying experience of approximately 3,169 hours. About 16 of those hours were in the previous 30 days and about 56 hours were in the last 3 months. Most of the instructor’s flying experience was in single-engine aeroplanes (3,002 hours), with the instructor recording approximately 2,584 hours of total experience as a CASA‑qualified flight instructor. Staff records indicated that the instructor had accumulated 2,421 hours while working for the operator. The instructor had approximately 569 hours in the AT-01, the same type flown on the day of the accident, with about 33 hours in the last 3 months on that type.

Medical information

The instructor held a Class 1 aviation medical certificate that was valid until 12 June 2021 with no restrictions. The instructor was reported to be fit and active and in a good state of mind in the days before the flight and was likely well-rested. The instructor was not reported to be taking any prescription medications and had no reported medical condition that could have affected their ability to operate an aircraft that day. A post-mortem examination identified no significant background natural disease, which could have contributed to the accident.

Student

Qualifications

The student held a recreational pilot licence for single engine aeroplanes issued in December 2018, with a navigation rating for this licence issued in September 2019. The student also held recreational licence ratings for operations in controlled airspace, operations at controlled aerodromes and for the operation of flight radio. Additionally, the student was endorsed to operate aircraft with manual propeller pitch control. The student’s licence did not have an expiry date and was valid at the time of the flight.

Experience

A review of the student’s logbooks, student files and flight data showed that at the time of the accident the student had accumulated a total flying experience of approximately 259 hours. About 10 of those hours were in the previous 30 days and about 13 hours were in the last 3 months. Operator’s records indicated that the student had accumulated 174 hours while training with the operator, with the remaining 85 hours acquired previously at another flying school. The student had approximately 72 hours in the AT-01, the same type flown on the day of the accident, with all 13 hours flown in the last 3 months being on that type.

Medical information

The student also held a class 1 aviation medical certificate that was valid until 20 October 2021 with no restrictions. The student was reported to be well rested in the days before the flight. The student was not reported to be taking any prescription medications and had no reported medical condition that could have affected their ability to operate an aircraft that day. A post-mortem examination identified no significant background natural disease, which could have contributed to the accident.

Pilot fatigue information

Based on reported sleep data alone, it is unlikely that the crew were fatigued. It is possible that the student may have experienced some mental fatigue due to the assessment nature of the flight. However, there was insufficient evidence to make that determination. Further, the instructor was unlikely to be fatigued at the time of the accident.

Aircraft information

Overview

VH-OIS (OIS) (Figure 4) was an Aquila AT01 fixed-wing aircraft manufactured in 2012. The aircraft was first registered in Australia on 14 December 2012 to Soar Aviation Aircraft Holdings Pty Ltd (Soar). The aircraft was a two-seat, low‑wing training aircraft constructed from carbon fibre and fiberglass. Control surfaces were a mixture of push rod and cable operated. The power plant was a Rotax 912S with a two-blade constant speed propeller.

Figure 4: VH-OIS

Figure 4 shows VH-OIS on display at Avalon Airshow

Source: Andrei Bezmylov, used with permission

Maintenance

OIS had a current maintenance release located in the aircraft, issued on 1 September 2020 which was valid for a period of 150 hours or 12 months, whichever was sooner. At the time of the accident the aircraft had accrued 31.6 hours since the maintenance release issue and had a total time in service recorded as 3,540.9 flight hours.

The aircraft was maintained to the airframe and engine manufacturer’s inspection schedules with an operational category of aerial work. At the time of the accident, there were no listed defects or outstanding maintenance endorsed on the maintenance release.

Aircraft weight and balance

The aircraft had a valid load data sheet, with the empty weight and centre of gravity data approved on 30 October 2013. Weight and balance calculations performed by the ATSB found that OIS was below maximum take-off weight and within centre of gravity limits for the duration of the flight.

Flight data

The aircraft was not fitted with a flight data recorder or a cockpit voice recorder, nor was either required by regulations. Some Aquila models were fitted with a KAPI electronics flight data recorder, however, the manufacturer advised that this device was not fitted to OIS at the time of manufacture.

The primary source of flight data for OIS on the day of the accident was from satellite-based ADS‑B receivers operated by Aireon. These receivers are onboard the Iridium satellite constellation.

Aerodrome information

Coombing Park was classified as an ‘uncertified’ aerodrome, otherwise known as an aeroplane/aircraft landing area (ALA) and included one primary grass runway, and another shorter grass runway aligned about 70° from the primary runway (Figure 5). Coombing Park had an ICAO designator of YCPK, an elevation of 2,350 ft and had a common traffic advisory frequency (CTAF) of 119.0, shared with Orange Airport. The primary runway was nominally aligned in an east-west direction, at about 73° magnetic (85°T) (designated runway 07/25), and had a length of just over 1,200 m. The shorter runway was about 400 m long and aligned in a south-westerly and north-easterly direction and was reported as not being available for use.

The primary runway had a cleared area of more than 30 m on each side of the runway centreline that was free of obstacles. The runway at Coombing Park appeared in good condition at the time of the accident, with the surface being firm, the grass relatively short, and the conditions dry. The gradient of the runway matched the contours of the ground, rising in the direction of runway 07. Rising terrain existed to the east of the field, increasing to a local peak about 350 ft above the runway elevation about 1 km from the end of runway 07.

Two windsocks were present, one near the intersection of the 2 runways and the other near the end of runway 07 on the southern side, as shown in Figure 5. Both windsocks appeared complete, free to move, and in good condition. Powerlines were present about 200 m from the end of runway 25 crossing the extended centreline in a north-south direction less than a 5% gradient from the runway 25 threshold.

Figure 5: Coombing Park ALA facing north-east

Imaging showing a grass runway. Labels show location of the accident site, cross strip, touch down area of runway 07 and Carcoar township.

Source: ATSB

Information about Coombing Park ALA was available in the OzRunways application. This application was installed on the flight instructor’s mobile phone and was in use during the flight. At the time of the accident, the application had an additional ‘remarks’ section for Coombing Park that included:

Bird and animal hazard exist.

Rising terrain east of field.

No other procedures were included in this guide, nor were they required by any regulations. However, following the accident on 12 November 2020, additional information about Coombing Park ALA was included in the OzRunways application. The ‘remarks’ section was updated to include:

Rising terrain to East, crosswind departures from [runway] 07 only
 

Terrain at 3,953 within 3NM of field.

No go around on [runway] 07

Power lines at end of [runway] 25

Additionally, ‘procedures’ were added to the OzRunways application in the same update that included:

 

RH circuits on runway 07

Crosswind departures from [runway] 07

Up wind departures from [runway] 07 prohibited

Simulated engine failures prohibited

Engine failure on [runway] 07: make RH turn and glide to creek flat

Coombing Park ALA was not included in the Enroute Supplement Australia (ERSA).

Meteorological information

Weather information used by the pilots

Documents onboard the aircraft indicated that the student prepared a location briefing, including weather information, at 1304 using NAIPS. This briefing included a graphical area forecast (GAF)[7] for the planned area of operations, aerodrome forecasts (TAFs) and reported observations (METARs)[8]. TAFs and METARs were obtained for Bankstown and Young Airports, and METARs for Cowra Airport. Young Airport was located approximately 35 km from the originally‑planned waypoint near Greenethorpe Township. QNH[9] was also requested for the area of operations and was predicted to be 1014 for the duration of the flight.

The forecasts obtained by the student (TAFs and GAF) covered the duration of the planned flight. The TAF released at 1226 for Young Airport contained predictions for CAVOK[10] conditions. Surface winds were forecast at 12 kt from a westerly direction 290° (True), with the temperature ranging from 26‑28 °C. The GAF issued at 0924, and found onboard the aircraft, predicted visibility greater than 10 km and no significant cloud or weather phenomena for the duration and planned area of flight.

Post‑accident weather analysis

GAFs for the area of operations, TAFs and METARs for Bathurst and Orange Airports, and METARs for Cowra Airport were obtained from the Bureau of Meteorology (BoM) covering the duration of the flight. Although not requested by the student, the TAF released at 1234 for Orange Airport predicted similar operational conditions to Young. These were CAVOK, with similar surface winds at 10 kt also from a westerly direction (260° True), with a temperature of 22 °C. All forecast conditions evaluated predicted suitable weather conditions for flight under the visual flight rules.

Observed conditions at Orange Airport during circuits

The METAR for Orange Airport released at 1500, 4 minutes after OIS arrived, indicated similar conditions to those predicted by the TAF released at 1234. Visibility was greater than 10 km with no cloud detected, and surface winds were reported at 10 kt from a westerly direction (250° True), and a temperature of 24 °C and QNH 1016. Similar conditions were reported in the METAR released at 0530, 7 minutes prior to the departure of OIS from Orange. However, the wind direction changed to a south‑westerly direction (210° True) at 10 kt, with the temperature dropping to 23 °C and QNH 1015.

Estimated conditions on arrival at Coombing Park

Weather observations from Bathurst, Orange and Cowra Airports were used by the ATSB and the Bureau of Meteorology (BoM) to estimate conditions at Coombing Park ALA at the time of the accident (1709). These were the nearest aviation meteorological sites surrounding the ALA, with Cowra Airport being 50 km to the south-west, Orange Airport being 28 km north and Bathurst being 55 km to the north-east.

All METARs released at 1700 for these locations showed visibility greater than 10 km and no cloud detected, consistent with eye-witness reports from the time of the accident. QNH was recorded in METARs as 1011 at Cowra, 1015 at Orange, and 1014 at Bathurst, and was estimated as 1015 for Coombing Park, being closest to Orange. Temperature in the same METARs was 29 °C at Cowra, 23 °C at Orange and 25 °C at Bathurst, and was estimated as 24 °C at Coombing Park, based on the aerodrome elevation and proximity to Orange.

Surface wind analysis

The BoM conducted an analysis of surface winds at the accident site based on the surface winds reported in 1-minute intervals at Cowra, Orange, and Bathurst. BoM estimated a surface wind direction of westerly though north-westerly, with an estimated wind speed of 5‑10 kt, possibly up to 15 kt. Surface wind gusts were not expected to exceed 5‑15 kt, although gusts up to 20 kt were possible, consistent with observed wind gusts at Cowra, Orange and Bathurst. Based on this analysis, if the wind was north-westerly, the mean tailwind component on runway 07 was estimated between 3‑6 kt, gusting up to 13 kt.

If the wind was from a westerly direction, the mean tailwind component was estimated between 5‑10 kt and gusting up to 20 kt. The crosswind component would have likely ranged between 2‑8 kt from the left of OIS, and possible crosswind gusts up to 15 kt. Based on the runway direction, BoM reported that it was unlikely that there was a headwind component to the surface wind over the runway at the time of the accident.

Estimation of tailwind

The ATSB also estimated the tailwind component on runway 07, based on the groundspeed of OIS during approach to land. Flight data indicated that the ground speed on late final approach of OIS was about 67 kt. Based on a recommended approach indicated airspeed of 60 kt, as required by the AFM for OIS, the nil wind ground speed or true airspeed at the elevation of Coombing Park for this indicated airspeed was calculated as 62.3 kt. Based on this, the mean tailwind component was estimated to be 5 kt, within the bounds of the probable wind speed provided by the BoM.

Aircraft performance information

Touch-and-go information

Data for the accident flight showed that, following the 2 left precautionary search patterns around Coombing Park ALA, the aircraft flew parallel to the main runway (corresponding to the left downwind leg for runway 07) before a left descending turn was conducted (base leg). At the completion of this turn the aircraft was slightly right of the runway 07 centreline, approximately 1,200 m horizontally from, and 300 ft above, the runway threshold. The approach flight path indicated that a slight left tracking correction was likely performed to align the aircraft with runway 07. An eye-witness, who was a pilot familiar with the strip and situated approximately 300 m to the west of the runway 07 threshold reported that the aircraft flew about 40 ft above the trees in a very flat, low and slow approach.

The eye-witness also reported that the approach appeared to be smooth, well configured and balanced, to have lots of flap out, with the engine sounding normal. At the time the aircraft flew past, the runway was not visible to the eye-witness due to the vantage point being slightly below the runway threshold. However, the eye-witness recalled stating to a friend that they hoped the aircraft didn’t land short of the strip when observing the approach. The eye-witness expected that the aircraft would have touched down ‘right near’ the runway 07 threshold, based on the flight path and slow airspeed observed. There were no witness marks identified by the ATSB on the runway surface that were indicative of a hard or abnormal landing. The touch down area for runway 07 (Figure 5) was slightly less grassed than other parts of the runway, consistent with normal landings in that area, no marks were found that could be attributed to the touch-and-go.

ATSB evaluation of flight data along the runway strip showed that the previously‑observed descent rate during the approach was no longer discernible shortly after the aircraft crossed the runway 07 threshold. However, it was not possible to identify the exact position of touch down due to the precision of this data. This data also indicated that the aircraft crossed the runway threshold at a ground speed of approximately 60 kt, with the aircraft beginning to reduce speed about 5 seconds after this point. Within about 10 seconds, the groundspeed reduced to a minimum speed of about 44 kt approximately 400 m past the runway threshold. The flight data then indicated that OIS began to accelerate, consistent with application of engine power by the pilots. ATSB evaluation of flight data estimated that lift-off and climb to 50 ft likely occurred approximately 929 m from the runway 07 threshold, attaining a stable ground speed of about 63 kt during the initial climb.

Evaluation of touch‑and‑go length

The ATSB performed a comparative analysis between the length (distance from the runway threshold to the take‑off point) of the touch-and-go at Coombing Park compared to the length of 4 touch-and-go landings and the standing take-off conducted at Orange Airport. The purpose of the analysis was to identify any factors that may have increased the length of the touch-and-go at the ALA. Further detail on this analysis, including methodology, additional context and findings of this analysis are described in Appendix A – The influence of touch‑and‑goes to the accident involving VH-OIS, 4 November 2020. A summary of the findings from this analysis are as follows.

The length of the Coombing Park ALA touch-and-go was the third shortest of the flight. The two shortest touch‑and‑go landings conducted at Orange at 1602 and 1615 had lengths of just over 750 m. The fourth shortest touch‑and‑go was conducted at Orange at 1556 (the first approach) and had a length of about 1,000 m.

The ATSB applied corrections for aircraft mass, density altitude and surface type to estimate the length of the Coombing Park touch‑and‑go if it had been conducted at Orange (the corrected touch‑and‑go). The purpose of this correction was to allow a direct comparison between the Coombing Park touch-and-go and all take-offs conducted at Orange. The corrected touch‑and‑go was estimated to have a length between 812 m (with a 25% factor applied for dry grass) and 1,015 m (no grass strip correction) (as stated above the actual length was 929 m). Based on these estimations, it is likely that if the Coombing Park touch‑and‑go had been conducted in the same way at Orange Airport, the length would have been in the same ranked order (that is, third shortest).

The ATSB also evaluated the potential effect of OIS performing a standing take-off instead of the touch‑and‑go at Coombing Park under the same conditions, and this is also detailed in Appendix A.

From this analysis, the standing take-off length at Coombing Park was estimated to be between 630 and 836 m. The estimated zone along Coombing Park runway 07 where OIS would have climbed to 50 ft from a standing take-off is shown in pink in Figure 6. This figure also shows the location of the start of runway 07, the actual take-off for OIS, the upwind track in red-dashed lines, and an indicative track for the start of the left turn. The reference tree is shown for continuity between figures in this report.

Figure 6: Upwind track of OIS following touch‑and‑go at Coombing Park ALA, (view looking south-west)

Image showing upwind track of OIS (red dashed lines), actual take-off position of OIS and estimated take-off zone if a standing take-off was conducted

Source: ATSB

Guidance for determining touch‑and‑go length

There were no details for the calculation of touch‑and‑go length in the aircraft flight manual (AFM), nor were these required. Research by the ATSB identified a limited number of guidance publications for the calculation of a suitable runway length for a touch‑and‑go. Two articles were identified that presented anecdotal techniques for the calculation of required runway length for a touch‑and‑go. As context to the challenges associated with a touch‑and‑go in an article by Aviation Safety magazine, Burnside (2019) stated:

The principal challenge posed by the touch‑and‑go manoeuvre is managing the immediate transition from a landing to a take-off, presenting that touch‑and‑goes are more complicated than just stringing together a landing with a take-off.

Further, in an article released by the Aircraft Owners and Pilots Association (AOPA), Wright (2006) identified some of the challenges that may be encountered by a pilot during a touch‑and‑go after the aircraft has landed:

Once power is added, the pilot is now concerned with achieving a normal take-off rotation speed while ensuring that there is sufficient runway to safely clear any obstacles. Somewhere in all of this there must be a verification that the flaps really did retract, the engine instruments are giving normal readings, and all required checklist items are accomplished.

At some undefined point, the landing roll ends, and the take-off roll begins. If the runway is relatively short, it may leave you with little to no room to execute a safe rejected take-off. Likewise, floating too long during the landing or being slow in retracting flaps may bring on the same situation.

The articles by Wright (2006) and Burnside (2019) each presented a method to estimate a safe length for a touch‑and‑go to be conducted. Wright (2006) stated:

A good rule of thumb is to take the landing distance over a 50-foot obstacle and the take-off distance over a 50-foot obstacle, add them, and double (or triple) the result to get a minimum runway length.

Burnside (2019) provided a sample calculation based on 3 elements. In summary, this sample calculation involved calculating:

• The average of the landing roll and landing distance required to clear a 50-foot obstacle;
• the time on the ground to reconfigure the aircraft, and;
• the average of the take-off ground roll and take-off distance required.

In this article, Burnside (2019) identified several reasons to be more conservative than this, citing that pilot technique, a less than perfect engine, a dragging brake or a crosswind requiring drag-producing aileron input can change those numbers.

Significantly, both methods estimated that a touch-and-go length used more runway than a standing take‑off. Additionally, the analysis conducted by the ATSB also identified that all touch-and-go manoeuvres conducted by the student were likely to use more runway than a standing take-off.

Specifically, when corrected for density altitude, a standing take-off length at Coombing Park ALA was estimated by the ATSB to be between 630‑836 m for the atmospheric and weather conditions at the time of the accident, notably shorter than the actual Coombing Park ALA touch‑and‑go length of 929 m.

Initial climb performance

The ATSB evaluated the potential influence of a standing take-off on the height of OIS during the initial climb. The analysis indicated the relative height difference if a standing take-off was conducted instead of the actual touch-and-go. The climb profiles are projected in straight lines, along an extended upwind track, although these results also indicated the relative differences between a standing take-off and the touch-and-go on curved flight paths. The distance from the start of Coombing Park ALA runway 07 was used as the datum for the analysis, and this is shown on the horizontal axis of Figure 7. For reference, these distance markers are reproduced over an image of the runway area in Figure 6 and the upwind area in Figure 9.

This analysis used the estimated standing take-off distances described in the section above and is illustrated in Figure 7. The vertical axis shows the height above the estimated point of lift off and the horizontal axis shows the distance from the start of runway 07 at Coombing Park ALA, and also corresponds to the markers shown in Figure 6.

Terrain elevation is shown by the solid dark green area, with typical tree heights indicated by the translucent dark green (minimum 33 ft (10 m)) and light green (maximum 66 ft (20 m)) areas.

Figure 7: Height of initial climb profile for OIS at Coombing Park ALA, including estimations for climb profiles following standing take-off and estimated terrain profile

Figure showing plot of height vs distance travelled of OIS during the initial climb from Coombing Park ALA runway 07 compared to terrain and estimates for climb if a standing take-off was conducted.

Source: ATSB

The actual initial climb profile of OIS conducted at 1709 is shown in Figure 7 by the series with red lines and blue circles. The last data point of this series indicates the relative height of OIS immediately prior to the left turn. The red dotted series is the line of best fit for the 1709 initial climb and shows an estimated projection of the climb if the climb rate was maintained and OIS tracked in a straight line instead of turning left. The top right corner of Figure 7 shows the projected climb passing within about 45 ft (14 m) of the terrain, below the maximum nominal height.

The likely climb profiles following theoretical standing take-offs at Coombing Park ALA are shown by the blue band (labelled ‘Standing take-off climb profile’ in Figure 7). The standing take-off climb estimates show the result of translating the actual climb of OIS at 1709 to start from take-off locations of theoretical standing take-offs at Coombing Park ALA, as described in the section ‘Estimation of standing take-off distances’ in Appendix A.

The dashed lines above and below the blue band represent the limits of the analysis, with the best-case shown in orange and the worst-case shown in dark blue. The best-case scenario is derived from the shortest theoretical standing take-off (630 m), and the worst-case is derived from the longest theoretical standing take-off (836 m), as described above. The top right corner of Figure 7 shows all climbs following hypothetical standing take-offs as passing close to, but above the maximum nominal tree height at the local terrain peak, with the worst case projected to pass within about 67 ft of the terrain.

The best- and worst-case standing take-off estimates equated to OIS being between 22‑70 ft higher at each point during the initial climb if a standing take-off was conducted instead of a touch‑and-go, as shown by the height differences between these lines and the dotted red-line in Figure 7. In contrast to the initial climb after the actual touch-and-go projecting below the maximum tree height, climb projections from standing take-off estimates indicated that OIS would pass close to, but slightly above the maximum nominal tree height had an upwind departure been conducted. Further climb performance analysis for curved flight paths are discussed in the section titled Turn and final track information below.

Initial climb gradient

At the time of the accident, flight planning legislation required pilots to plan to take-off with the aircraft weight sufficiently low to allow the aircraft to achieve a minimum climb gradient. This was stipulated in paragraph 7.1 of Civil Aviation Order 20.7.4, which stated:

In the take-off configuration with landing gear extended, an aeroplane must have the ability to achieve a climb gradient of 6% at take-off safety speed, without ground effect, and with all engines operating at take-off power.

Based on flight data, the ATSB calculated the climb gradient of OIS during the initial climb from Coombing Park to be just over 7%. This was based on the flight data for OIS between 50 ft and 200 ft above the lift-off point (shown by the red line with blue circles in Figure 7). OIS was very likely outside ground effect for this segment of the flight. Further, the climb gradient of OIS was typical of climbs performed earlier in the flight at Orange.

Initial climb rate

The ATSB evaluated the climb rate of OIS after take-off from Coombing Park compared to climbs after take-offs at Orange Airport, earlier in the flight. Climbs after take-off were evaluated from 50 ft above the surface, to reduce uncertainty associated with the transition from the ground roll to a stable climb, including ground effect. Corrections for density altitude were performed on the climbs at Orange (increasing the actual values by 3.6%) to compare with the Coombing Park initial climb which had a lower elevation. Figure 8 shows the results of this comparison.

Figure 8: Climb rate of OIS after take-off at Coombing Park ALA, and climb rate for OIS at Orange Airport corrected for density altitude at Coombing Park ALA, 4 November 2020

Chart showing comparisons between initial climbs conducted at different times during the flight

Source: ATSB

The blue diamonds in Figure 8 show the mean climb rate calculated for each climb away from the runway, indicated in ft per minute by the vertical axis, with labels on the horizontal axis indicating the local time that each climb was commenced. The height of the boxes and vertical lines around the mean illustrate uncertainty in the rate of climb, with the boxes indicating where the mean climb rate likely existed, and vertical lines (whiskers) indicating where the mean climb rate very likely existed.[11]^{, [12], [13]}

The left most data point labelled ‘YCPK 1709’ shows a mean climb rate of about 460 ft per minute during the straight component of the initial climb from Coombing Park runway 07. This rate was comparable to the initial climbs conducted in OIS at Orange at 1603, 1616, 1620 and 1637. The considerably lower initial climb rate at 1610 is consistent with an aborted landing being conducted with full flaps extended, a planned exercise as part of the CPL assessment.

Based on the AFM for OIS applied to the observed conditions at Coombing Park ALA, the best rate of climb was 580 ft per minute. The best rate of climb in the AFM is calculated based on test flights conducted under normal operating conditions, typically using a new aircraft and engine, with the aircraft flown in an optimal configuration. For this reason, it is not expected that OIS would have achieved the published climb rate in the AFM. Due to the damage to OIS, it was not possible to perform further assessments to establish why the rate of climb differed between OIS and the expected rate of climb in the AFM. However, it should be noted that the examination of the wreckage by the ATSB did not reveal any mechanical faults or defects with OIS, as described in the section titled Wreckage and accident site information.

In summary, flight data analysis conducted by the ATSB identified that the climb performance of OIS during the initial climb at Coombing Park ALA was comparable to climbs conducted at Orange Airport earlier in the flight.

Turn and final track information

Flight data indicated that immediately prior to turning left, OIS was approximately 125 ft above the ground (Figure 7), tracking toward steeper terrain with a significant number of trees. Figure 9 shows an image facing in a similar direction and at a similar height to the pilots just prior to where the left turn was initiated. The projected upwind track with distance markers shows the rising terrain corresponding to the elevation profile shown in Figure 7, with the same reference tree as shown in Figure 6. The projected upwind track provides a relative guide for analysis of climb performance covered in other section and does not necessarily indicate the intended track of OIS.

Figure 9: Image looking uphill at a similar height and position to OIS immediately prior to left turn

Image showing upwind track of OIS, approximate track during left turn and projected track if OIS continued in a straight line

Source: ATSB

Figure 10 shows a top-down image of the upwind area of runway 07 for Coombing Park ALA. This shows the entire track of OIS in red lines with blue circles, including the initial upwind track for the climb profile in shown in Figure 7. For reference, common labels such as the distance markers and reference tree are included in Figure 6, Figure 7 and Figure 9.

Figure 10: Image looking down on upwind area of Coombing Park runway 07 showing track of OIS and accident site

Down-looking image of runway 07 end of Coombing Park ALA and flight track of OIS

Source: Google Earth, with image overlay and annotations by ATSB

An indicative track of the left turn and final track of OIS is also shown in Figure 10. Flight data indicated that as OIS approached a line of trees, the track changed by about 50° to the left in less than 7 seconds, corresponding to a turn rate of more than 450° per minute. The radius of the turn was estimated to be about 185 m, with an angle of bank calculated as 22°.

Compared to other climbing turns conducted during the flight, the turn rate was almost twice as high, and the bank angle was at least 60% steeper. During the turn, OIS continued to climb, although at a reduced rate, gaining about 25 ft by the completion of the turn, with the mean groundspeed decreasing slightly.

The ATSB conducted a climb performance analysis to evaluate 3 hypothetical ‘escape’ scenarios for OIS:

• continuing the left turn,
• turning right instead of turning left, or
• ceasing the left turn half-way through and flying straight up the gully (shown in Figure 9).

This involved projecting a climb profile over the terrain elevation map based on Google Earth and the estimated height of surrounding trees. Based on this evaluation, the hypothetical flight profiles for the left and right turns both passed within about 66 ft (20 m) of the ground. This was a similar height to the trees in the area.

The flight up the gully was estimated to pass within about 72 ft (22 m) of the ground. Although this was marginally higher than the trees in the area, due to uncertainty in the data, it was not possible to determine if OIS would have been able to climb away in this scenario.

Flight data and onsite measurements indicated that following the left turn, the height of OIS is likely to have been similar to the canopy height of surrounding trees for the third- and second-last data points, although the precise height could not be determined. Beyond the tree close to the flight path (shown in Figure 11), projections indicated that if the climb rate was restored after the turn and OIS had cleared this tree, the aircraft would have passed within 66 ft (20 m) of the ground in the rising terrain beyond the dam at a similar height to trees in the area.

The last 3 flight data points indicated that the ground speed of OIS likely increased, with OIS travelling approximately 110 metres in 3.2 seconds between these 3 points. OIS likely descended by at least 25 ft, possibly by 50 ft between the last 2 flight data points. This was likely to the left of and below the canopy of the tree close to the flight path. Approximately 1 second after the last flight data point, OIS collided with the embankment on the far side of the dam at a similar height.

Wreckage and accident site information

Accident site

The accident site was located on private property about 38 km south of Orange township and 27 km south of Orange Airport (Figure 1). This property was adjacent to Coombing Park ALA, with OIS located about 600 m from the end of runway 07, and about 300 m left of the extended runway centreline (Figure 3).

Wreckage examination

Site and wreckage examination did not identify any aircraft defects or anomalies that might have contributed to the accident. Examination of the area surrounding the accident site, including the tree close to the likely flight path (labelled), did not identify any trees or terrain that might have been struck by OIS prior to the wreckage location (Figure 11).

Figure 11: Image looking down on accident site and surrounding trees viewed from the direction of travel

Image showing accident site and surrounding trees

Source: ATSB

The aircraft struck the embankment of a small dam in an upright attitude. Flight data and ground scars marks indicated that OIS was tracking in a generally northerly direction at the time of the collision with terrain. The accident site was located slightly to the left of the projected track, with propeller strike marks indicating that the nose of the aircraft was pointed to the left of this track around the time of the collision. During the collision, the aircraft rotated in a counter-clockwise direction, coming to rest facing toward the south-south‑west, about 180° to the direction of travel.

The damage to OIS was consistent with significant upward and rearward forces being transferred through the aircraft structure during the collision. The left wing and left-wing flight controls, aircraft canopy, cockpit area, nose landing gear, engine cowling, propeller and tail section of the aircraft were severely disrupted by impact forces. The counterclockwise rotation during the collision was likely induced by the left side of the aircraft (mostly from the left wing that was completely destroyed) colliding with the dam embankment prior to the right side of the aircraft.

Engine and propeller

The engine assembly was examined and found to be complete with no evidence of pre-accident engine control problems or defects. The inspection found all engine controls were connected from the cockpit controls through to the engine, with all electrical wiring and hoses connected.

Both propeller blades fractured and separated from the propeller hub during the collision. Both tips of the propeller blades were located at the accident site. No evidence of pre-accident damage was identified in the propeller blades, propeller hub or propeller hub attachment to the engine crankshaft. The propellers broke into multiple fragments, and most fragments were located at the accident site. However, one large portion of blade was located about 27 m to the east of the site, consistent with this portion breaking away and being thrown through the air during the impact sequence.

Fuel

OIS was fully fuelled immediately prior to departure from Bankstown Airport and had sufficient fuel to conduct the flight. An onsite inspection identified that both the left- and right-wing fuel tanks were compromised with no fuel being identified in these tanks. Fuel was identified in both carburettor bowls and was found to be clear of contaminants, including water.

Emergency locator transmitter

OIS was fitted with an emergency locator transmitter that activated automatically during the accident sequence. This was deactivated by the ATSB onsite. The signal from the ELT was not received by emergency services, likely due to damage sustained to the antenna during the accident.

Flight controls

Flight controls were examined to the extent permitted by impact damage. Nothing was identified in the wreckage with respect to a flight control defect or malfunction that may have contributed to the accident. Complete continuity of flight controls from the cockpit to the control surface was established for the right-wing aileron, flaps, elevator, rudder, and elevator trim. The elevator trim position was unable to be established.

Continuity of the left-wing aileron and flaps was partially established from the cockpit to the control surface attachment brackets. Push rods and bell cranks associated with the wing mounts of these control surfaces were detached from the left wing during the collision, however, no pre‑existing defects were identified in these components.

The flaps were separated from the flap control actuator during the collision. The flap control actuator included a screw jack with the function of moving and holding the flaps in a fixed position. The extension of the screw jack had a corresponding flap position in the normally functioning aircraft. The ATSB was advised by the aircraft manufacturer that the measurement on the screw jack extension corresponded with the flaps being fully retracted.

Precautionary search information

A precautionary search and landing is a procedure for conducting a safe, powered landing away from an airport or ALA with known suitable landing surface conditions. It is normally conducted for two reasons:

• A landing on an unprepared landing surface made necessary due to an abnormal or emergency situation, such as deteriorating weather, insufficient remaining daylight, fuel shortage, technical problems, developing medical conditions or any other reason determined by a pilot
• A pre-planned landing when the pilot is unfamiliar with the landing area, or its condition is unknown.

Soar operations manual

Procedures at Soar were governed by an operations manual, as required by Civil Aviation Safety Regulation Part 141, regulation 141.260. This manual was available for use by flight instructors through an online portal. Version 2.5, published on 2 April 2020, was the most current version of the operations manual at the time of the accident. This operations manual was developed from the CASA sample operations manual version 2.1 dated October 2016. All relevant text cited in this report is common between the Soar Operation’s manual and the CASA sample operations manual.

Criteria for the assessment of the precautionary search

The purpose and importance of a pilot demonstrating the ability to conduct a safe precautionary search was stated in paragraph 11.1.3 of Advisory Circular 91-02[14]:

The ability to accurately assess the prevailing environmental conditions, potential obstacles, surface conditions, dimensions and ultimate suitability of a landing area, will be enhanced by using a well‑practiced procedure to maximise the likelihood of a safe landing outcome.

Clause 3.4(f) of Schedule 5 ‘Flight test standards’ in Part 61 Manual of Standards (MOS) page 586 dated 11 December 2018 noted that a precautionary search was required as part of the test specific activities and manoeuvres in the commercial pilot licence (CPL) flight test and was therefore also to be covered in the pre-CPL flight test. Further, Soar’s operations manual stated that the CASA standard syllabus, lesson plans and planning matrices would be used for the schools training courses.

A ‘flight test report’ form (described in the section ‘Assessment plan for the CPL flight test’ on page 24) was used by the instructor to assess CPL competency during the flight. This form included item 3.4(f) labelled ‘Conduct precautionary search’ under the section ‘Test specific activities and manoeuvres. A reference to MOS A6.4 was included for this item, and this is reproduced in Figure 12. No assessment notes were recorded against this, or other items of this section of the form.

Figure 12: Extract from Schedule 2 ‘Competency Standards’ of Part 61 Manual of Standards dated 24/09/2018 page 126

Precautionary search height

The importance of the height selection for inspection runs during a precautionary search was articulated in the following paragraph 11.1.4 from Advisory Circular AC 91-02¹³:

It will be particularly important to consider appropriate heights to be able to conduct such a procedure safely, while remaining cognisant of potential engine failure considerations, especially if the requirement for a precautionary procedure was initially necessitated by an aircraft malfunction, low fuel state, or other related issue.

Further, for the inspection of the proposed landing path, page 59 the CASA Flight Instructor Manual (FIM)[15] stated:

This preliminary inspection should be sufficiently low for the surface to be inspected but not so low that it is necessary to avoid obstacles. Another point to impress on the student is that the inspection runs should be made at a constant height whilst safely avoiding upwind obstacles. If not satisfied with the surface complete at least one other inspection run at a lower height if necessary.

Further the FIM stated:

The need for a really good lookout whilst carrying out this low flying exercise cannot be over‑emphasised.

And

This exercise should be practiced only in approved areas or at approved fields and even then, all effort should be made to avoid frightening livestock and annoying people.

Inspection heights flown by OIS

During each circuit pattern the altitude of OIS varied, with flight data indicating that the pilots descended OIS during the southerly and easterly pattern legs before returning to a nominal height between 1,000 and 1,100 ft AGL. Based on the Soar guidance to instructors for use when conducting training exercises (referred to as patter notes), the inspection pass would have been the easterly facing leg of each circuit pattern (Figure 3). Each eastern facing leg was flown just over 1 km from the centreline of runway 07. Based on flight data for OIS, each inspection pass was flown at the following heights:

• Pattern 1 (Figure 3 yellow lines): OIS descended to a minimum height of 750 ft AGL, and 400 ft above the ridge on the extended centreline for runway 07 (the ridge), during the easterly leg before returning above 1,000 ft AGL.
• Pattern 2 (Figure 3 orange lines): OIS descended to 850 ft AGL (500 ft above the ridge) during the easterly leg before climbing above 1,000 ft AGL for the northern leg and descending throughout the western leg.
• Pattern 3 (Figure 3 red lines): OIS levelled out between 450 and 475 ft AGL (100 to 125 ft above and 3.3 km from the ridge), before commencing a climb just after crossing over the extended centreline of runway 07/25 (start of pattern 3 in Figure 3). OIS continued to climb through the third easterly facing leg from about 500 to a peak height of 1,200 ft AGL mid‑way along the third northern leg. The flight instructor sought permission to conduct a touch‑and‑go from the property manager of Coombing Park ALA at some time along the southerly and/or easterly legs of this pattern.

Guidance in Aircraft flight manual for OIS

Under a section labelled ‘Emergency Procedures’, the AFM for OIS included guidance for the conduct of a precautionary search, stating that the inspection height shown be flown at an altitude above 500 ft. It is expected that this was intended to refer to height instead of altitude.

Section 2A2 of Soar’s operations manual stated ‘Where the Aircraft Flight Manual conflicts with other publications the Aircraft Flight Manual has precedence.’ However, the circumstances described in the AFM related to landing on an un-prepared field and did not include guidance for the conduct of a precautionary search and landing in normal operations or training.

Operator’s guidance

The operator produced guidance for instructors to use when conducting flight training exercises, referred to as ‘Patter Notes’. Under the patter notes for the demonstration of a precautionary search and landing, a series of 3 circuits were described, each with different heights to be flown. Each circuit was to be conducted at a nominated circuit height, with a descent during final approach to progressively lower heights for each circuit. The third circuit was listed as ‘only conducted if still uncertain about field conditions. After each circuit, the guidance suggested a climb back to the nominated circuit height.

The heights were listed in the patter notes as follows:

• Circuit height (every circuit): 1000 feet AGL ([to] simulate [d an actual height of] 500 feet AGL)
• First inspection pass: 1000 feet AGL ([to] simulate [d an actual height of] 500 feet AGL)
• Second inspection pass: 700 feet AGL ([to] simulate [d an actual height of] 200 feet AGL)
• Third inspection pass: 550 feet AGL ([to] simulate [d an actual height of] 50 feet AGL)

For each circuit, 2 sets of heights were presented, a higher, simulated set for the purpose of an exercise over an unprepared field (figures shown outside brackets above), and the other set relating to the actual heights to be flown for an effective inspection (shown within brackets). For example, the simulated height of the second inspection pass was 700 ft AGL, with the actual height listed in brackets as ‘simulated 200 feet AGL’. At the completion of the third circuit the guidance stated:

Turning final go around (if at an ALA or aerodrome, conduct short field landing).

 

In the event of a real prec search, short field landing, full stop landing. Do not taxi prior to walking the field.

At the conclusion of the precautionary search guidance, the patter notes stated:

Note: if conducting PSL at an ALA or aerodrome, actual heights may be used.

The ‘actual heights’ referred to are those listed inside the brackets above.

Other guidance

Guidance relating to the assessment of overshoot areas and the recommended heights that this assessment should be flown are included in various sources. However, the minimum height recommended for the conduct of an inspection run varied. Page 61 of the CASA FIM provided detail for the conduct of an air exercise by a flight instructor to demonstrate a practiced precautionary search and landing:

When in a suitable area descend to about 500FT above the ground and tell the student to assume poor weather conditions with a cloud base of about 600FT and poor visibility. Choose a suitable airstrip and demonstrate how to inspect the surface. Fly at low safe cruising speed with the optimum flap setting. Fly over the field slightly to the right of the intended landing path at about 100FT to make the first check. On this run check the surface and drift and note any high ground and obstacles in the overshoot area.

In ‘A pilot’s guide to safe flying’[16], for a precautionary landing procedure, this guide includes:

Confirm the acceptability of the landing area by carrying out an inspection run at 200 feet AGL (or well above possible obstructions), into wind, and slightly to the right of the intended landing area. Check out the following:

• …
• Obstacles, both on approach, and possible overshoot.
• …

Guidance was also produced by the New Zealand Civil Aviation Authority[17]. For the first circuit, the guidance for the focus during the inspection includes:

On this and subsequent legs, …, with particular emphasis on surrounds in the approach and climb-out areas.

Guidance for a second inspection run stated:

Assuming the chosen landing site appears suitable up to this point, a second inspection is carried out at a minimum of 200 feet AGL.

…

Established on and parallel to final, a gradual descent to a minimum of 200 feet AGL is carried out.

Descent below 200 feet AGL is not recommended, because it takes considerable concentration to fly the aeroplane level and look at the landing site surface. Also, there is a possibility of unseen obstructions, and since a climb to 500 feet AGL will be initiated on completion of this inspection, the climb is minimised. Unless committed to the landing, never descend below the highest obstacle in the go-around path.

Selection of take-off and landing direction

Influence of wind

Pilots are required to choose the most suitable direction for landing and take-off when operating at any aerodrome. Generally, the main factor for this decision relates to the wind strength and direction. On a flat runway, a take-off or landing into wind (a headwind) will reduce the distance required in each case, whereas a tailwind will increase the distance required. For OIS, performance charts indicated that a tailwind component of 1 kt would lead to increases to the landing and take-off distances between 2.5‑2.8%. This equates to about 12 metres for the ground roll and a further 12 metres to clear a 50 ft obstacle per 1 kt of tailwind.

Influence of runway slope

In a section titled ‘Inclined runways’, Geeting and Woerner (1988)[18] stated:

Where possible, plan upslope landings and downslope take-offs.

This is because conducting a landing on a runway with an uphill slope reduces the landing distance required, whereas a take-off on an uphill slope increases the required take-off distance. An extract from guidance about Coombing Park ALA in the OzRunways applications described that the ‘airfield follows ground contour, rising to the east’. OIS conducted the touch-and-go in an easterly direction at Coombing Park ALA (runway 07).

For the landing component of the touch-and-go conducted by OIS, the tailwind condition (as described in the section titled Post accident weather analysis) and slight uphill runway slope had opposing effects on the landing distance. Estimations performed by the ATSB indicated that the deceleration advantage provided by an uphill landing at Coombing Park ALA was in a comparable range to the disadvantage provided by the tailwind.

The tailwind component was probably relatively low, based on the climb gradient of OIS being comparable to earlier climbs, and the pilots likely assessing the wind strength and direction from the two serviceable windsocks at the ALA. However, due to uncertainty in the actual wind conditions, it was not possible to determine the relative impact that the tailwind and the runway slope had on the touch-and-go length. Therefore, it was not possible to determine which runway direction would have provided the shortest distance for landing at the time of the accident.

During the take‑off phase of the touch-and-go, the tailwind and uphill slope would have both combined to increase the distance for OIS to become airborne. However, the touch-and-go length was comparable to others conducted at Orange earlier in the flight as discussed earlier, and it was again not possible to determine the relative contribution that the tailwind and slope had on the touch-and-go length. Once airborne however, the tailwind component would likely have had an adverse effect on the aeroplane’s climb gradient, although this could not be quantified.

Considerations for use of runway 25

Conducting a touch-and-go on runway 25 would have required the pilots to approach over the rising terrain and trees to the eastern end of the field. This would have required a steeper angle than the approach flown for runway 07, or for the pilots to touchdown part way into the runway. Due to the same uncertainty in the wind strength noted above, it was not possible to determine the length of a touch-and-go on runway 25, and the relative influence of a downhill landing or the headwind components.

The ATSB also considered the effect of conducting a standing take-off from each runway. For take-off on an inclined runway, Geeting and Woerner (1988)¹⁷ stated:

Always take-off downhill in light wind conditions. Acceleration is greater, less runway is required, and obstacles are cleared more easily. During take-off, a 1° downslope is roughly equivalent to having 10% more runway; a 2.5° upslope is equivalent to having a 7 kt tailwind during take-off.

For the take-off of OIS on runway 07, the tailwind conditions and uphill runway slope would have increased the distance required compared to a take‑off conducted on runway 25. Additionally, the final approach gradient of OIS to runway 07 was estimated to be about 6.5%, shallower than the climb-out gradient of about 7%.

Significantly, this indicated that OIS would likely have cleared all obstacles and terrain after a standing take-off from runway 25.

Operational information

Commercial pilot licence flight test assessment plan

Section 3D1 of the operations manual titled ‘Training Plans and Syllabuses’ stated:

Soar Advanced Flight Training has elected to use standard syllabuses, lesson plans and planning matrices prepared by CASA. These syllabuses are reproduced IAW those listed in Volume 5. In designing various courses, the school may vary the lesson times or add lessons in the syllabuses as required to be consistent with the training course in use.

This is consistent with interviews with instructors who stated that the CPL syllabus followed the requirements of Part 61 of the Manual of Standards.

Section 5A2.3 of the operations manual titled ‘Assessment plan’ stated.

A student may be deemed competent to conduct a solo flight, be recommended for a flight test or issued a qualification when competency is demonstrated on at least two occasions (each occasion being on a separate flight). Pre-solo and end of course assessments have been planned on this basis. End of course assessments take into account all of the units of competency mentioned in the Part 61 MOS for the licence, rating or endorsement.

Based on previous flights, this flight would have constituted the first of two pre-CPL flight assessments for the student. All units of competency mentioned in the Part 61 MOS for the commercial pilot licence would apply. This was consistent with a partially completed ‘Flight Test Report’ form 61-1490 found onboard the aircraft. Based on this, the assessment was to include diversions, aborted landings, practiced forced landings, short-field and flapless landings, and precautionary searches. Based on flight data for OIS, these exercises appear to have been performed during the accident flight.

Pilot roles and responsibilities

For this flight (and other training flights) with the student and instructor onboard, the flight instructor was the pilot-in-command. This is confirmed in section 3B1.17 of the operations manual titled ‘Instructor-student co-ordination’:

The Pilot In Command for dual training flights is the flight instructor. The flight instructor is responsible for ensuring that there is no doubt as to who has physical control of the aircraft at any given time; hence, the hand-over and take-over procedures shall be adopted for all dual training flights.

Further, section 3B1.1 of the operations manual titled ‘Authorisation of training flights’ stipulated:  ‘prior to despatching any training flight in a School aircraft, the Pilot in Command is to ensure that they have checked the following:

• …
• changes or restrictions concerning the use of the aerodrome and training area;
• …

Therefore, for this flight, the operations manual confirmed that the flight instructor was expected to check if any restrictions existed with respect to using the Coombing Park ALA for flight training.

Pre-flight information

Purpose of flight

The purpose of the flight was for Soar to assess the competency of the student, prior to the student undertaking the CPL flight test (the pre-CPL assessment). The pre-CPL assessment was required by Soar for every student and acted as a final check to ensure that the student was ready for the actual flight test. Accordingly, the pre-CPL assessment flight was structured to include exercises for the student to demonstrate competency against the CASA CPL syllabus as directed by the flight instructor.

Flight arrangement by student

The pre-CPL assessment flight was intended to be conducted later in the week, however, on the morning of the flight, the student noted that the weather was suitable for flying and called Soar between 0830 and 0900 to ask if there were any senior instructors available. The student was informed that the instructor and an aircraft were available during the afternoon, and this was subsequently scheduled.

Just after this time at about 0900, the instructor likely started their shift as was typical for most days. The student left home at about the same time to drive to Soar at Bankstown Airport, with this trip typically taking between 40 to 50 minutes and the student likely arriving close to 1000. At 1145, the instructor conducted a dual training flight with another student involving a session of circuits at Bankstown Airport for just over 1 hour in duration.

Pre-flight preparations by student

At 1332, a flight plan was lodged for OIS using the National Aeronautical Information Processing System (NAIPS). This plan detailed a flight departing Bankstown at 1400 and tracking directly toward a set of coordinates just over 240 km to the west before returning to Bankstown Airport. These coordinates were 8 km south of the township of Greenethorpe (although lines that were drawn on flight charts were directly to the township) . The flight plan also stipulated a search and rescue time (SARTIME) for arrival at Bankstown by 1800.

Student likely unaware of actual waypoints

Prior to lodgement of the flight plan, other flight instructors at Soar reported that the student spoke with the instructor directly. Although it is not known what was discussed, it is expected that the discussion would have involved setting forward requirements for the flight to allow the student to lodge the flight plan and commence the flight. Therefore, it is not expected that the student was aware of the actual destinations or location of inflight diversions that occurred during the flight due to the exercises required by the CPL flight test. Further, it is unlikely that the student was aware that the flight would proceed to Coombing Park ALA.

Flight chart information

Pencil lines were found on the relevant world aeronautical chart (WAC) and visual navigation chart (VNC) located onboard the aircraft. Firm, straight lines between Bankstown Airport and Greenethorpe township indicated that the student had likely used Greenethorpe township for flight planning calculations and navigation. Lighter curvy pencil lines between the approximate location of the orbits and Orange Airport were also drawn on the WAC (the location of this inflight diversion is labelled in Figure 1) . These lines were likely drawn by the student inflight, consistent with being directed by the instructor to conduct an inflight diversion to Orange Airport.

A pencil line between Orange Airport through the south-west of Blayney on the WAC chart was drawn, intercepting the original planned route near the township of Burraga. This line was firmer and relatively straighter than the original diversion line, possibly indicating that this was drawn while the aircraft was on the ground at Orange, consistent with eyewitness and photographic evidence.

Coombing Park ALA was not marked on the charts used by the student, and there were no pencil markings identified on the WAC toward this region. This may indicate that the course deviation toward Coombing Park ALA after departing Orange Airport (shown in Figure 1) was a second inflight diversion exercise prior to the precautionary search and landing.

In summary, the pencil lines on the student’s flight charts were consistent with the student being unaware of the flight proceeding to Coombing Park ALA until after departure from Orange Airport. Further, there was no evidence that the student had been to Coombing Park ALA prior to this flight.

Operations to aeroplane landing areas

Company requirements

The Soar operations manual included guidance for flights to aerodromes, with section 3B1.5 titled ‘Aerodrome Suitability’ including the statement:

Except in an emergency, aeroplanes operated by Soar Advanced Flight Training will only be operated to or from aerodromes that are listed in En-Route Supplement Australia (ERSA) or Aeroplane Landings Areas (ALAs) that conform to the guidance provided in CASA publication CAAP 92-1(1).

The wording of this text was directly taken from that detailed in the CASA sample operations manual version 2.1 dated October 2016, which stated:

Except in an emergency, aeroplanes operated by [Sample Aviation Flight Training Pty Ltd] will only be operated to or from aerodromes that are listed in En-Route Supplement Australia (ERSA) or Aeroplane Landings Areas (ALAs) that conform to the guidance provided in CASA publication CAAP 92 1(1).

Further, under the section titled ‘Company register of suitable ALAs’, the operations manual included:

Form 4B13 (Aeroplane Landings Areas (ALA) Report Form) is to be used for compiling a company register of suitable ALAs of fixed wing aeroplane landing areas that are not listed in the ERSA, but have been approved by the HOO.
 

Information listed in the register is advisory in nature. The HOO should be advised if an amendment is considered necessary.

These requirements were in place in the Soar operations manual for at least one year prior to the accident and were included in the previous version of the operations manual, and possibly earlier iterations.

Form 4B13 contained fields to provide detail of the ALA, including a diagram of the landing area, and ALA characteristics including the slope, surface, elevation and obstructions. A signature box stating ‘Approved for company OPS’ was at the bottom of the form. This form could have been used to assess an ALA such as Coombing Park against the guidance of CASA publication CAAP 92-1(1), as required by section 3B1.5 of the operations manual.

At the time of the accident, Soar did not have a company register of suitable ALAs compiled.

Recommendations for aerial work and charter operations in CAAP 92-1(1)

Paragraph 8.2 of CAAP 92-1(1) stated:

A pilot should not use a landing area without taking all reasonable steps to ensure the physical characteristics and dimensions are satisfactory. For aerial work and charter operations the operator should provide evidence to the pilot on the suitability of a landing area prior to its use.

Soar Aviation’s Air Operator’s Certificate included approvals to conduct CASR Part 141 flight training. This was classified as an aerial work operation.

Scenario based training

Soar had 10 published scenarios that were available for use by their instructors for CPL training. These generalised scenarios allowed for any ‘suitable aerodrome’ to be used during the flight, with the aim to provide a more realistic training environment for students. Based on Soar’s operations manual, this included any ALA in the ERSA.

Certification standards for aerodromes

The CASA website included descriptions of 3 different categories of aerodromes under CASR Part 139. This included descriptions of where the standards were defined, and those responsible for certification of the aerodrome. For aeroplane landing areas, under the subject ‘Who is responsible for certification?’, the table noted these as an unregulated facility, with use to be:

in accordance with Aircraft Operators Certificate and/or pilot’s responsibility to determine suitability of the facility.

It was also noted that there were no defined standards, although guidance for use was included under CAAP 92-1.

Aeroplane landing area entry requirements into ERSA

All requirements for entry of an aerodrome into the ERSA related to the periodic review and provision of up to date data, covered under CASR Part 175.D as part of the Aeronautical Information Package (AIP).There were no requirements to demonstrate the suitability of ALAs listed in the ERSA, including assurance of obstacle clearance heights in the take-off and approach areas of each runway.

The CASA website included the question: ‘Are aerodrome details published in ERSA/NOTAM?’. For ALAs, this stated: ‘Airservices Australia may publish basic information for aerodromes that were previously regulated’. This was confirmed with Airservices personnel to potentially relate to some legacy aerodromes.

In response to an inquiry from the ATSB, CASA confirmed that the ERSA contained ALAs that may have been previously regulated and those ALAs that were not previously regulated. For previously regulated ALAs that did not transition to a certified standard, these were allowed to remain in the ERSA, but with operational information removed. In correspondence to the ATSB, CASA also clarified:

The exact content in the ERSA is between the aerodrome operator and Airservices Australia. However, runway distance data (LDA, TORA, TODA etc) is removed. Any flight operation/airspace information is CASA’s responsibility.

CASA also confirmed that there is no assurance of obstacle clearance for ALAs in the ERSA. CASA also noted that:

Pilot’s still have responsibilities for safe conduct of flight and minimum height rules.

The ERSA current at the time of the accident under the section ‘Aerodrome Information’ stated:

4.3 Aerodromes with limited information
 

a. Other aerodromes, also known as [Aeroplane] Landing Area (ALA) may be included in ERSA with limited information.

b. ALA are depicted in ERSA with a grey background as shown in INTRO.

c. Operators conducting regular public transport or charter operations into ALAs need to be aware of their obligations under the CASA regulations.

Prior to commencing a flight to an uncertified aerodrome, a pilot or operator must contact the Aerodrome Operator to ensure currency of aerodrome information.

Coombing Park ALA was eligible for entry into the ERSA at the time of the accident and could have been added by Airservices Australia if requested by the manager following compliance with CASR Part 175.

Discussion between instructor and manager of Coombing Park

Under the section titled ‘Company register of suitable ALAs’, the Soar operations manual also included:

The PIC must obtain permission to use the ALA when required and is responsible for determining that the area is suitable for the intended operation.

The property manager of Coombing Park recalled that, just after the pilots commenced the third precautionary search pattern overhead the ALA (indicated by ‘Pattern 3’ in Figure 3), they received a phone call from a person identifying themselves as a flight instructor from Soar. Telephone records identified that the duration of the call was 24 seconds, commencing at 1704. The property manager, who was in the vicinity of the runway, could see and hear OIS in the circuit area and confirmed with the instructor that it was them in the circuit area.

The property manager recalled that the instructor asked, ‘Do you mind if I do a touch‑and‑go?’ The property manager reported that based on:

•  the instructor identifying themselves as a flight instructor
• OIS already being in the circuit area
• the manager assuming that OIS was going to land on runway 25
• a desire not to distract the pilots further

 the property manager granted the instructor permission to land without further discussion.

The property manager reported that normally, people called up an hour or so ahead of time, and that advice on local procedures and potential hazards was provided. The manager also reported that this was the first time they had spoken with the flight instructor and couldn't recall talking to anyone from Soar Aviation in the past.

Instructor familiarity with Coombing Park

Staff members representing Soar reported that the flight instructor may have been familiar with Coombing Park ALA. However, the precise nature of the instructor’s interactions with the ALA were not known. Further, it was reported that Soar previously held a register of ALAs that included Coombing Park. It was reported that this register had been misplaced, possibly during an office move, and was unable to be provided to the ATSB for review.

The ATSB examined data from flights previously conducted by the flight instructor to evaluate any previous instances of operating in the circuit area of Coombing Park ALA, and any other ALAs. This analysis used data stored in the flight instructor’s OzRunways account, and publicly available location data for ERSA and non-ERSA ALAs, including Coombing Park. The flight instructor’s data was limited to times when the instructor was using the OzRunway mobile phone application and the phone was in range of a telecommunications tower.

Limitations in the accuracy and fidelity of the available data prevented identification of whether a landing occurred at any of the locations. Flights to ALAs within range of Bankstown Airport and flown past within 5.5 km (indicating the 3 nm circuit area) by the instructor were of particular interest (Figure 13).

Figure 13: Aeroplane landing areas within range of Bankstown Airport and within 5.5 km of flights conducted by the instructor between 2016 and 4 November 2020

Image showing relative location of ALAs flown past by the instructor compared to Bankstown Airport and Coombing Park ALA

Source: ATSB

The analysis revealed that the instructor passed at least 59 ALAs at least once from a total of 138 flight data files between 2016 and the time of the accident. Forty-three of these ALAs were located in New South Wales, 10 in Queensland and 6 in Victoria. Approximately half of these ALAs (30) were not included in the ERSA, with the remaining 29 ALAs being listed. The flight data revealed that for most flights there was no evidence of the aircraft joining the circuit pattern at these ALAs, however, these locations were sometimes used as waypoints, characterised by a change in direction of the aircraft track. The most common locations passed were on the eastern side of the Great Dividing Range and between Newcastle to the north and Goulburn to the south.

Coombing Park was passed once by the instructor on 21 September 2018, with 2 heading changes occurring as the aircraft passed overhead (Figure 14). However, there was no evidence to suggest that any circuit or landing (including a precautionary search) was conducted at that time. No other evidence was identified for flights by the instructor to Coombing Park.

Figure 14: Flight by the instructor within 5.5 km of Coombing Park on 21 September 2018

Image showing top down map of Coombing Park ALA and Carcoar township. Red line depicts the flight track.

Source: Google Earth, annotated by the ATSB

There were 2 occasions identified where precautionary searches likely occurred at ALAs (Mangrove Mountain ALA and The Oaks ALA), with a landing possibly occurring at The Oaks. The flight data also indicated that a precautionary search was likely conducted at Mangrove Mountain ALA, a non-ERSA ALA on 3 November 2018.

In this case, the flight track did not pass along, or directly approach the runway, indicating that a landing did not occur. One flight, on 14 January 2017 indicated that the Oaks ALA (which was included in the ERSA) was likely used for a precautionary search and landing. In summary, the instructor’s previous flights stored by OzRunways did not provide evidence of any previously conducted landings at non-ERSA ALAs.

Civil Aviation Advisory Publication 92-1(1) Guidelines for aeroplane landing areas

As detailed previously, the Soar operations manual only allowed operations to non-ERSA ALAs when they had been assessed to conform to the recommendations in Civil Aviation Advisory Publication (CAAP) 92-1(1) - Guidelines for aeroplane landing areas (the CAAP), and this had been approved by Soar’s head of operations. Such an evaluation of Coombing Park ALA had not been performed by Soar.

CAAP 92 overview

At the time of the accident, the introduction of the CAAP, released in July 1992, noted:

This publication sets out methods that may be used and which experience has shown should, in the majority of cases, ensure compliance with the Regulations. However, before using the information in this publication the user should always read the Civil Aviation Regulations listed…’

The CAAP then stated:

These guidelines set out factors that may be used to determine the suitability of a place for the landing and taking-off of aeroplanes. Experience has shown that, in most cases, application of these guidelines will enable a take-off or landing to be completed safely, provided that the pilot in command: (a) has sound piloting skills; and (b) displays sound airmanship.

CAAP 92 assessment of Coombing Park

The ATSB assessed Coombing Park ALA against the obstacle clearance recommendations of the CAAP. This recommended the ‘approach and take-off areas to be clear of objects above a 5% slope for day...operations’. This was an area extending 900 m beyond the end of the runway and is illustrated in Figure 15. The accident flight path was within this area prior to the final left turn (Figure 10).

Figure 15: Summary of eligibility requirements for ‘other’ physical characteristics of a CAAP 92-1(1) aeroplane landing areas

Source: Civil Aviation Safety Authority, CAAP 92-1(1) ‘Figure 2A – Single engined [sic] and Centre-Line Thrust Aeroplanes not exceeding 2000 kg MTOW (day operations)’

Table 1 shows Coombing Park ALA compared to the minimum recommended requirements of the CAAP for the approach and take-off area slope. The 2 left-most columns show the required CAAP value and relevant characteristic, with the 4 right-most columns relating to measurements conducted for Coombing Park ALA. Three sets of measurements were conducted for Coombing Park ALA based on the actual runway length, and characteristics if the runway length was shorter. The purpose of the theoretical measurements was to examine if the minimum recommended obstacle clearance requirements of the CAAP could be met at Coombing Park if runway 07 was shortened, increasing the distance between the runway 25 threshold and the rising terrain and obstacles to the East of the field.

Runway length requirements (CAAP 92(1)-1 Clause 5.1) and theoretical shortened runway calculations were based on 2 scenarios. The first scenario labelled ‘Theoretical shortened runway– AT01 Nil wind TODR [take‑of distance required]’ simulated a scenario where the runway was shortened to match the take-off distance required for VH-OIS on the day of the accident in nil wind. This represents a baseline scenario.

This calculation was based on VH-OIS performance charts for the estimated density altitude on the day of the accident, with an increase of 25% length for a short dry grass surface at Coombing Park. This data was considered ‘factored’ for the purpose of CAAP 92(1)-1 clause 5.1 that recommended an additional 15% be added to the length when using ‘unfactored’ data. The other scenario labelled ‘Theoretical shortened runway– worst case standing take-off’ is the result of a 5% slope applied from the location of the longest estimated standing take-off (836 m), as described in the section titled Evaluation of touch‑and‑go length and shown in Figure 7.

Table 1: Partial comparison of Coombing Park ALA to recommended minimum physical characteristics of landing areas according to CAAP 92-1(1) section 5 for Aquilla AT01 day operations

The assessment against CAAP 92-1(1) recommendations revealed that Coombing Park ALA likely did not meet obstacle clearance recommendations for VH-OIS at the time of the accident. Approach and take-off area slopes are illustrated in Figure 16 relative to the terrain profile (and upwind climb profile of OIS for reference), as also shown in Figure 7. These slopes are shown by the yellow, blue and orange lines, and have labels prefixed with ‘CAAP 92 surface...’ in Figure 16. Each of the 3 obstacle clearance slopes terminated 900 horizontal metres from the (simulated and actual) runway ends, as per the recommendations in CAAP 92-1(1). For reference, the horizontal axis shows the distance from the start of runway 07, corresponding to figures shown earlier. This shows that the CAAP 92-1(1) obstacle clearance surface originating from the end of runway 07 penetrated the rising terrain about 150 m from the runway end (orange line), as also noted in Table 1.

Figure 16: CAAP 92(1)-1 Runway approach and take-off obstacle clearance slope from various theoretical runway end positions projected onto the terrain profile of Coombing Park ALA compared to actual climb performance of VH-OIS

Figure showing plot of height vs distance travelled of OIS during the initial climb from Coombing Park ALA runway 07 compared to terrain and projections for the CAAP 92 obstacle clearance slope

Source: ATSB

The approach and take-off slope originating from the theoretical ‘nil wind’ shortened runway distance (shown in yellow) did not intersect with the terrain within the 900 m horizontal limit. However, this may have intersected with some of the trees in this area, and the projection of the shortened runway slope intersected the rising terrain approximately 250 m beyond this point.

The approach and take-off slope from the estimated worst-case standing take-off for VH-OIS is also shown in Figure 16 by the blue line. This also shows that the approach and take-off slope may have passed the CAAP 92 recommendation for a 5% obstacle clearance within 900 m, although this intersected with the terrain almost immediately after this point. However, it is likely that trees in this area would have penetrated this slope. Note that the CAAP 92 slopes from both theoretical shortened runways shown in Figure 16 correspond to theoretical take-offs shown in Figure 7 of the same colour.

Related occurrences

The ATSB’s national aviation occurrence database was reviewed to identify accidents and serious incidents involving collisions with terrain at ALAs that were reported to the ATSB in the 30 years between 1991 and 2020. These records were refined to occurrences where the characteristics of the runway and surrounding terrain may have influenced the outcome. In total 9 occurrences were identified meeting these criteria over this period.

Three of the occurrences identified related to ALAs listed in the ERSA and were therefore approved for use by Soar’s operations manual. One of the 3 ALAs were located at a similar or shorter distance than Coombing Park ALA was to Soar’s base of operation at Bankstown Airport (within range of Soar). A summary of these occurrences is below.

• ATSB investigation AO-2018-025 Runway excursion and collision with terrain – Van’s RV-6A, VH-OAJ, Somersby ALA, New South Wales on 18 March 2018. This investigation found that features surrounding the runway, including undulating terrain and a small watercourse immediately at the end and trees at the edge, increased the likelihood and severity of occupant injury in the case of a runway excursion. And further, that CAAP 92-1(1) did not have guidance for the inclusion of a safe runway overrun area.
• ATSB occurrence 200903966 Collision with terrain involving a Piper Aircraft Corp PA-31, Bungle ALA (Bellburn), Western Australia on 06 June 2009. During final approach to runway 10, the aircraft's right wing tip struck a tree branch, causing minor damage to the leading edge. This was a commercial charter flight with one pilot and 9 passengers onboard. A review of the approach area to runway 10 at Bungle ALA did not reveal any significant undulation in terrain, however there appeared to be slight height variation near the runway threshold. This ALA was in the ERSA and the occurrence is also notable because it relates to a charter operations to an ALA in the ERSA that was not assured as meeting CAAP 92-1(1) obstacle clearance recommendations.
• ATSB Occurrence 199700351 Collision with terrain involving an unknown ultralight aeroplane, Coominya ALA, Queensland on 9 February 1997. The pilot was attempting an uphill take-off. Soon after becoming airborne, the pilot reported becoming concerned about the closeness of the trees off the end of the runway. The pilot commenced a 180-degree turn, although the aircraft struck trees before the turn was completed. Subsequently, the aircraft struck the ground nose first and the pilot received serious injuries during the impact.

A further 2 occurrences were identified by the ATSB that were within range of Soar but were not listed in the ERSA. Therefore, these aerodromes were not approved for use for Soar operations without further risk assessments being conducted. These occurrences appear to be related to either obstacles and or terrain in the take-off and landing areas. These are of note because all aerodromes were eligible for entry in the ERSA (subject to a risk assessment), making them approved for use by the Soar operations manual.

• ATSB investigation AO-2018-013 Collision with terrain involving Cessna 206, VH-WZX, Apollo Bay ALA, Victoria on 31 January 2018. There was a pilot and 5 passengers on board. The ATSB found that a go-around commenced late during the landing and the pilot did not immediately follow the go-around procedure. These factors, combined with the heavy aircraft weight and rising terrain, reduced obstacle clearance and the aircraft struck the airfield boundary fence.
• ATSB Occurrence 201901041 Collision with trees involving Piper Aircraft Corp PA-38, VH‑CNT, Currandooly ALA, New South Wales on 23 February 2019. During the initial climb, the aircraft was unable to outclimb rising terrain and stuck trees. The aircraft subsequently collided with terrain resulting in substantial damage to the aircraft and minor injuries to the pilot. The occurrence is notable because the sequence of events and terrain appear similar to the accident scenario for VH-OIS. The aerodrome was within range of OIS from Soar’s base of operations at Bankstown Airport.

Safety analysis

Introduction

While conducting an assessment in preparation for a commercial pilot’s licence flight test, an Aquila AT01 registered VH-OIS (OIS), conducted a precautionary search at Coombing Park Aeroplane Landing Area (ALA), about 27km south of Orange Airport, New South Wales. Flight data indicated that after three precautionary search patterns were flown, a touch‑and‑go was conducted, with the aircraft climbing toward rising terrain after take-off. As the aircraft approached an area of steeper terrain, a climbing left turn was conducted toward a slightly lower and less wooded area. The aircraft continued to fly in a controlled manner, likely accelerating and descending slightly toward an open area of rising terrain beyond a small dam. After flying over the dam, the aircraft collided with an embankment, bordering the dam and the open area.

Site and wreckage examination, and aircraft performance assessments did not identify any defects, anomalies or sudden performance loss that might have contributed to the accident. Additionally, no evidence was found to suggest any medical or fatigue related issues that would have affected the pilots’ performance on the day of the flight. Therefore, this analysis will focus on the operational factors that led to an experienced flight instructor and a commercial student pilot to conduct a touch‑and‑go toward rising terrain that the aircraft was unable to outclimb.

Development of the accident

Pre-flight planning

Due to the nature of the flight, in preparation for the commercial pilot licence flight test, only the instructor is likely to have been aware of Coombing Park ALA being a waypoint for the flight prior to departure, although it is not known if a landing was originally planned by the instructor. Coombing Park ALA was not listed in the ERSA and therefore, use of the ALA for landing was not approved by the company operations manual without further assessment, which had not been performed. This requirement was likely known to the instructor as it had been in the operations manual for at least one year. For this accident, if this procedure was followed, the aircraft would not have landed.

It was reported by staff at Soar that a previous register existed which contained Coombing Park ALA as an approved aerodrome. However, this register could not be provided to the ATSB. Further, if a risk assessment was performed by Soar for Coombing Park ALA previously, it is unlikely that it would have been identified as suitable for conducting a landing. This is due to the rising terrain to the east not meeting the recommendations of CAAP 92-1(1). It is also considered likely that the flight instructor would have reached the same conclusion if this hazard was assessed prior to flight.

It was also reported that the flight instructor was likely familiar with the ALA. This is consistent with one previous flight identified by the ATSB, where Coombing Park ALA was likely used as a waypoint by the instructor. However, there were no flights identified where the instructor conducted a precautionary search or landing at Coombing Park ALA. Therefore, there was no evidence that suggests the instructor had an opportunity to effectively assess the rising terrain hazard for conducting a take-off, touch-and-go or landing at Coombing Park ALA in previous flights.

The flight instructor contacted the manager of Coombing Park ALA to seek permission to conduct a touch-and-go while OIS was in the circuit area after 2 complete precautionary search patterns. This phone call only lasted 24 seconds. This was the only telephone call that the manager of Coombing Park ALA received from the instructor that day and could not recall receiving a phone call from any instructor from Soar in the past. The manager of the ALA reported that they usually provided a more in depth briefing for people calling ahead of time. This likely represents a missed opportunity to gain a knowledge of present conditions, operating procedures and hazards to be aware of.

It should be noted that until the point where the decision was made to conduct a touch-and-go, the flight had proceeded as expected. This included conducting the precautionary search patterns. Although it could not be confirmed, it is likely that the instructor considered the risk of a touch-and‑go to be low. This was likely based on Coombing Park ALA having a sufficient runway length for a landing (not a touch-and-go). It was not established if the instructor had intended on flying to Coombing Park ALA before the flight. However, the timing of the phone call to seek permission to conduct a touch-and-go after completion of two precautionary search patterns, may indicate that the decision to conduct a touch-and-go by the instructor was a last-minute decision. This was possibly under a motivation to provide a realistic experience for the student.

Overall, it is likely that if the suitability of conducting a touch-and-go at Coombing Park ALA was assessed prior to flight, the instructor would have identified Coombing Park ALA as unsuitable for landing. Therefore, it is expected to be likely that there was limited planning conducted prior to the flight with respect to assessing the suitability of Coombing Park ALA for a landing, take-off or touch‑and‑go. It is also possible that the decision to conduct a touch-and-go was an impromptu deviation from the original plan. Further, the lack of planning, or deviation from the original plan, removed a defence, placing more pressure on assessments in flight (which were also ineffective). These factors reduced assurance that the aircraft could outclimb the rising terrain, which subsequently led to the accident occurring.

Conduct of precautionary search

The southerly and easterly legs of each precautionary search pattern flown at Coombing Park ALA likely provided the best opportunity for the pilots to identify the rising terrain hazard to the east.  

The easterly inspection legs running parallel to the runway in the direction of landing were the inspection runs, For the duration of the third easterly leg, OIS was in a climb from 500 to 800 ft AGL, about 150 to 450 ft above the ridge and over 1 km to the right of the runway centreline. During this leg, the elevated nose position of OIS would have likely made assessing the rising terrain more difficult. Additionally, the changing perspective of the terrain may have been more apparent if the aircraft was in a straight and level configuration. In the two earlier patterns, the easterly legs were flown at more than 750 ft AGL, more than 400 ft above the ridgeline. Further, in all legs, OIS was more than 1 km to the right of the runway centreline, reducing the opportunity for the pilots to assess the overshoot area (and the runway surface).

At the start of the third pattern while tracking on the southerly leg (the third southern leg), OIS descended to a minimum height of about 450 ft AGL. Although this leg was only about 100 ft above the ridgeline on the extended centreline for runway 07 (the ridge), it was 3.3 km away from OIS.

Southerly legs were not considered inspection runs in the Soar ‘patter notes’ (these were the eastern legs), with these inspection runs noted to be flown at the lowest heights during the precautionary search patterns. Despite this, the third southern leg was flown at the lowest height of any legs at Coombing Park ALA during the precautionary search, prior to the final approach for the touch-and-go. This was at least 250 ft higher than recommended by the Soar ‘patter notes’ when conducting an actual precautionary search inspection run. This guidance was generally aligned with procedures for other organisations.

It is expected that the perspective offered by the ridge being slightly below the aircraft and more than 3 km away on the third eastern leg, reduced the likelihood that the pilots would identify this as unsuitable for take-off from runway 07. Further, the (easterly) inspection runs were flown at heights above the ground and offset positions from the runway centreline that likely reduced the pilots’ ability to detect the significance of the ridge at the end of runway 07 for take-off.

Therefore, it is likely that the heights and positions of OIS during the precautionary search patterns flown at Coombing Park ALA likely limited the opportunities for the pilots to detect the rising terrain hazard at the end of the runway, contributing to the decision to conduct a touch-and-go.

Take-off direction

It is likely that the ideal direction for landing and take-off were in opposite runway directions at Coombing Park ALA at the time of the accident. For landing, the light tailwind on runway 07 was countered by an uphill slope. Therefore, it is possible that either runway 07 or runway 25 would have allowed the shortest landing distance. However, there were fewer obstacles on approach to runway 07, likely making the approach less demanding on the pilots compared to runway 25.

The headwind component and downhill slope on runway 25 indicated that using that runway for a standing take-off would likely have allowed OIS to take-off in a shorter distance along the runway compared to runway 07. Further, due to no significant obstacles beyond the end of runway 25, it is very likely that OIS would have been able to safely climb away after lift-off. Finally, a standing take-off on runway 25 would have allowed OIS to start the take-off roll from the beginning of the runway, allowing the entire length to be used.

Therefore, a landing on runway 07, followed by a standing take-off on runway 25 would have likely prevented the accident.

The ATSB conducted further analysis (described in the next section) for a standing take-off compared to a touch-and-go on runway 07. This indicated that OIS may have been able to clear the terrain if a standing take-off was conducted on runway 07.

Touch-and-go

All touch-and-goes conducted by the student at Orange Airport likely used more runway than the standing take-off at Orange Airport (based on the distance from the runway threshold to the end of the take-off (where the aircraft climbed to 50 ft)). The touch-and-go conducted at Coombing Park ALA was likely typical of those conducted at Orange, with the approach reported by an eye‑witness as appearing normal. Based on flight data and calculations from the aircraft flight manual for OIS, the estimated standing take-off length at Coombing Park was estimated to be between 630 and 836 m, shorter than the actual touch‑and‑go length of 929 m. Therefore, it is very likely that a standing take-off would have led to OIS becoming airborne earlier along runway 07, reducing the likelihood of a collision with terrain.

There are more factors involved with the conduct of a touch-and-go compared to a standing take-off. This is supported by available literature and flight data for OIS that indicated a higher variability and longer length rolls from every touch-and-go compared to a standing take-off. This supports that there is likely less consistency with the length of a touch‑and‑go compared to a standing take-off.

ATSB analysis identified that the reduced runway length required by conducting a standing take-off at Coombing Park ALA was estimated to increase the height of OIS between a worst-case of 22 and a best-case of 70 ft at each point in the initial climb. This would have increased the margin between OIS and the terrain, in particular the trees. Adding the worst-case standing take-off height gain (22 ft) to the projections for the touch-and-go for different initial climb routes shows that all climb routes would have likely cleared the terrain by a greater height than the trees.

For a worst-case standing take-off, these estimations predicted these initial climb options pass the terrain by about 67 ft if tracking on the runway centreline up to 92 ft if a flight up the gully was conducted. The options to turn right, continuing the left turn, or climbing away from overhead the dam were all predicted to the terrain by a minimum of 88 ft. However, based on these estimations, and the maximum nominal height of trees being about 66 ft, it should be noted that any escape flight path would have still been at a very low height.

Therefore, the additional height offered by the worst-case standing take-off would have likely provided more options to the pilots, likely allowing the pilots to climb away from the terrain.

Forced landing

During the initial climb, the aircraft was likely performing normally. A projection of the flight path for OIS on the original track during the initial climb showed the aircraft likely would have collided with trees if flight was continued in that direction. It is likely that the pilots realised that the aircraft would be unable to outclimb the terrain and elected to turn left toward terrain that was less steep. Flight data analysis indicated that it was unlikely the pilots could have continued the left turn or conducted a right turn without colliding with trees. The aircraft rolled level flying toward the small dam, with flight data indicating that the aircraft was likely accelerating and descending just prior to crossing the surface of the dam. This indicates that the aircraft was likely in a controlled state at this point, about 1 second prior to impact. Based on this data, possibly after realising they would be unable to out climb the terrain, it is likely that the pilots were attempting to gain airspeed to affect a transition for an uphill forced landing. There was also likely minimal time for the pilots reconfigure the aircraft for a forced landing. During this attempted forced landing, the aircraft collided with the embankment of the dam.

Operator’s procedures

Soar's operations manual allowed training flights to be conducted to any aerodrome deemed suitable as part of a scenario-based training program. All aerodromes listed in the ERSA were considered as suitable for flight training operations. Therefore, Soar procedures did not require further assessment of ALAs listed in the ERSA. Further, for entry and ongoing listing in the ERSA, ALAs did not require any assessment of aerodrome suitability for aircraft operations.

In contrast, to be considered as suitable for flight training operations, ALAs not included in the ERSA were required to be assessed and found to meet all recommendations in CAAP 92-1(1). Therefore, Soar’s operations manual was inconsistent in the treatment of operational hazards between operations to ALAs listed in the ERSA and non-ERSA ALAs. This means that there were fewer controls in place to assure that obstacle clearance surfaces were suitable for use by Soar at ALAs listed in the ERSA.

Based on Soar operating in the aerial work category, the guidance in CAAP 92(1)-1 was suggesting that Soar would be expected to provide evidence to pilots on the suitability of landing areas. Although Coombing Park ALA was not in the ERSA, it was eligible to be without any further assurance that it was suitable for use by Soar. This is important because this ALA could have been approved for Company operations and would have resulted in the accident occurring at an ALA considered as suitable for company operations.

In summary, the CASA sample operations manual wording adopted directly by Soar Aviation provided approval for operation to ALAs in the ERSA without providing evidence and assurance that these were suitable for use by company aircraft.

CAAP 92 Recommendations

The Soar operations manual required that all ALAs not in the ERSA be assessed against the recommendations of CAAP 92-1(1) prior to use. Although this assessment was not performed for Coombing Park ALA, it could have been.

Section 5.5 of CAAP 92-1(1) included the recommendation that the terrain be clear of objects above a slope of 5% up to a distance of 900 m, however, there was no requirement for obstacle clearance beyond this distance. Coombing Park ALA had terrain that became steeper away from the runway. These circumstances may lead to situations where an aircraft is unable to safely climb away from the runway, despite the ALA meeting CAAP 92-1(1) obstacle clearance recommendations. There is also no assurance that other ALAs do not exist with these characteristics.

Coombing Park ALA did not meet the CAAP 92-1(1) recommendations for obstacle clearance beyond runway 07. However, if runway 07 was shortened, increasing the distance by displacing the runway 25 threshold, but still remaining at a suitable length for use by the accident aircraft, it is possible that the ALA would have met these recommendations. With the displaced runway 25 threshold, the ALA would likely have met these recommendations if several trees were removed from the take-off and approach area at the end of runway 07.

ATSB analysis showed that if the accident aircraft had conducted a standing take-off, it probably would have cleared the trees and terrain during the upwind climb by a very narrow margin. This situation would have been worse in an aircraft meeting minimum climb gradient requirements.

In summary, the CAAP 92-1(1) guidance did not assure that an aircraft would be able to safely outclimb rising terrain located more than 900 m from the end of a runway. This would have applied at Coombing Park ALA if the runway ended near the point of lift off.

Findings

From the evidence available, the following findings are made with respect to the collision with terrain involving Aquila AT01, VH-OIS, Coombing Park ALA, 27 km south of Orange Airport, New South Wales, on 4 November 2020.

Contributing factors

• It is likely that pre-flight planning was not performed to ascertain the take-off area was clear of obstacles at Coombing Park ALA as required by the operations manual. Consequently, this reduced assurance that the pilots would identify and mitigate the risk of rising terrain that the aircraft was unable to outclimb at the end of runway 07.
• The pilots were conducting the precautionary search at heights and positions that would have likely made assessing the hazard of rising terrain from the air less effective. This likely contributed to the pilots’ decision to conduct a touch-and-go landing and take-off toward rising terrain that exceeded the climb performance of the aircraft and required a low-level turn to avoid rising terrain.
• The take-off was conducted on an uphill slope with a probable tailwind and toward rising terrain beyond the runway end. A standing take‑off conducted in the more favourable reciprocal direction would likely have cleared all obstacles and terrain.
• The conduct of the touch-and-go, rather than a standing take-off, reduced assurance that the aircraft would commence the climb from a position along the runway that could clear the terrain during the initial climb after take-off.
• The combination of loss in climb performance during the turn, trees and rising terrain in the new direction likely led to the pilots conducting an uphill forced landing. It is likely that the aircraft had insufficient performance for the uphill forced landing, leading to the aircraft colliding with the embankment of a small dam.

Other factors that increased risk

• The CASA sample operations manual used by the operator that allowed any aerodrome in the Enroute Supplement Australia to be used for flight training did not assure that these aerodromes were suitable for use. (Safety Issue)
• Recommendations in CASA guidance CAAP 92-1(1) requiring obstacle clearance out to 900 m may lead to circumstances where ALAs meet these requirements however, aircraft are required to manoeuvre below a safe height or be unable to outclimb rising terrain after take-off more than 900 m past the runway end. (Safety issue)

Safety issues and actions

Sample operations manual

Safety issue number: AO-2020-059-SI-01

Safety issue description: The CASA sample operations manual used by the operator that allowed any aerodrome in the Enroute Supplement Australia to be used for flight training did not assure that these aerodromes were suitable for use.

CAAP 92 Guidance

Safety issue number: AO-2020-059-SI-02

Safety issue description: Recommendations in CASA guidance CAAP 92-1(1) requiring obstacle clearance out to 900 m may lead to circumstances where ALAs meet these requirements however, aircraft are required to manoeuvre below a safe height or be unable to outclimb rising terrain after take-off more than 900 m past the runway end.

Safety action not associated with an identified safety issue

Additional safety action addressing safety hazards at Coombing Park ALA

The property manager of Coombing Park ALA advised that the airstrip has been lengthened, extending the threshold of runway 07 by 300 metres to the west. Additionally, the threshold of runway 25 was displaced by 150 metres to the west, increasing the distance between this threshold and the rising terrain to the east of the field. Five large trees were also removed on the rising terrain to allow an escape route away from the rising terrain.

Glossary

ADS-B              Automatic Dependent Surveillance Broadcast

AIP                   Aviation information publication

ALERFA           Alert phase

ATC                 Air traffic control

CASA               Civil Aviation Safety Authority

CASR               Civil Aviation Safety Regulations

CCTV               Closed-circuit television

ERSA               En route supplement Australia

IAS                   Indicated airspeed

ICAO                International Civil Aviation Organization

INCERFA         Uncertainty phase

Runway end      End of runway in use.

Threshold         Beginning of usable portion of runway, i.e. downwind end.

Touch-and-go   Practice landing in which the aeroplane is permitted to touch runway briefly; in many cases flaps are moved to take-off setting while weight is on wheels.

Sources and submissions

Sources of information

The sources of information during the investigation included:

• accident witnesses
• Airservices Australia
• the Bureau of Meteorology
• the Civil Aviation Safety Authority
• the manager of Coombing Park ALA
• Soar Aviation Aircraft Holding Pty Ltd
• recorded data transmitted from the aircraft
• recorded data from the instructor’s OzRunway’s account
• photographs and videos taken of the aircraft on the day of the accident

References

Australian Transport Safety Bureau (2008). AR-2007-053 Analysis, Causality and Proof in Safety Investigations, Canberra, Australia.

Burnside, J. (2004). Airmanship: Touch and Goes? Aviation Safety Magazine, 20 September 2004. https://www.aviationsafetymagazine.com/

Geeting, D. & Woerner, S. (1988). Mountain flying (First edition), TAB Books Inc., Blue Ridge Summit, PA.

Vendeth, S (2003). A Pilot’s Guide to Safe Flying, A Manual for General Aviation Pilots (First Edition), Mt Eliza, Victoria, Australia

Wright, C. (2006). CFI to CFI rite of passage: The touchy subject of touch and goes. Flight Training Magazine, May 2006. https://www.aopa.org/

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• the Bureau of Meteorology
• the Civil Aviation Safety Authority
• the manager of Coombing Park ALA
• Soar Aviation Aircraft Holding Pty Ltd.

Submissions were received from:

• the Bureau of Meteorology
• the Civil Aviation Safety Authority
• the manager of Coombing Park ALA
• Soar Aviation Aircraft Holding Pty Ltd.

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Appendices

Appendix – The influence of the touch‑and‑go landing on the accident involving VH-OIS, 4 November 2020

This appendix documents a short study conducted by the ATSB to investigate the potential contribution that conducting a touch-and-go may have had on the accident involving OIS on 4 November 2020 at Coombing Park ALA. The key objective was to show the differences that conducting a touch-and-go in comparison to a standing take-off may have had on the ability of OIS to outclimb the rising terrain in the upwind area of runway 07.

Methodology

Aireon flight data was used to estimate the distance between the runway threshold and the aircraft reaching a height of 50 ft. The runway threshold (at the beginning of the runway) was used as a relative reference marking the start of the take-off run or touch‑and‑go. This was to allow comparison between standing take-off runs and touch‑and‑goes. As such, this indicates the overall average performance of the aircraft and pilot flying combination. As this was a pre-CPL flight test assessment, the pilot flying is expected to be the student pilot in every case.

The flight data point indicating the climb to 50 ft (and hence take-off) was identified by firstly reading backward in time from an established initial climb to the local minima in the data. The 50 ft point was then found by reading the first data point 50 ft above the local minima after the time of the local minima. Although there were undulations along the runways at Orange Airport and Coombing Park ALA, this was not expected to have affected the results in the locations where the initial climbs started. Further, this is expected to be within the tolerances of the flight data rounded to the nearest 25 ft. Distances were estimated as great circles using the haversine formula[19] between the latitude and longitude of the runway threshold and the flight data point as OIS passed through 50 ft. The results of this analysis are shown in Figure 17.

Results

Estimation of take-off distance

The horizontal axis of Figure 17 shows the distance between the runway threshold, at the beginning of the take-off run, and climb to 50 ft above ground level. Values associated with Orange Airport are prefixed with ‘YORG’ and are colour coded as orange for runway 29, and yellow for runway 11. The touch‑and‑go at Coombing Park is indicated in purple and prefixed ‘YCPK’. Bars indicating touch‑and‑goes include ‘T/G’ in the label, with the standing take off at Orange Airport labelled ‘YORG T/O RWY 11 0537UTC’.

Figure 17 also shows calculated take-off distance required for each runway used at Orange Airport from the aircraft flight manual for OIS, with the label of these calculated values prefixed by ‘YORG AFM TODR’. These are based on the reported winds, density altitude and aircraft weight at Orange Airport at 1600 and 1630. Labels are all suffixed with the local time at the start of each run or the time of weather observation. Further, the open blue bars prefixed with ‘YCPK T/G est. at YORG’ show an estimated range of the Coombing Park touch‑and‑go distance if conducted at Orange Airport. The bar labelled ‘grass’ includes a correction for the dry grass runway, as required by the AFM for OIS, with the bar labelled ‘no grass’ indicating results without a surface correction. This analysis is described further below.

Figure 17: Distance from runway threshold to take-off for VH-OIS at Orange Airport and Coombing Park ALA, 4 Nov 2020

Source: ATSB

Comparison of Coombing Park ALA take-off position with previous touch‑and‑go runs

The measured distance between the runway threshold and take-off (the touch‑and‑go distance) at Coombing Park ALA was compared to the same distances for touch-and-goes conducted at Orange Airport. The length of the Coombing Park ALA touch-and-go was the third shortest of the flight. The two shortest touch-and-goes conducted at Orange Airport (1602 and 1615) had a length of just over 750 m. With the fourth shortest touch‑and‑go being the first approach conducted at Orange Airport (1556) at about 1,000 m.

Corrections for aircraft mass, density altitude and surface type were applied to the measured touch-and-go distance at Coombing Park to compare the length of the touch‑and‑go conducted at Coombing Park with those conducted at Orange Airport, as shown by the open blue bars in Figure 17. No correction was performed for wind to show potential variation from this as a factor. These corrections were performed as follows. The take-off distance required by the AFM (the AFM TODR) were calculated for Orange Airport and Coombing Park ALA based on the estimated aircraft take-off mass, pressure altitude and temperature at each location. The ratio between the AFM TODR between Orange Airport and Coombing Park ALA was calculated, and this was multiplied with the actual touch‑and‑go distance at Coombing Park, measured from OIS flight data, and is summarised by the following equation.

Where, L_{YCPK at YORG} represents the estimated length of the Coombing Park ALA touch‑and‑go if it was conducted at Orange Airport. TODR_YORG and TODR_YCPK represent the take-off distances required for Orange Airport for reported conditions at 1600, and Coombing Park at 1700 respectively. Finally, L_CPK represents the actual touch‑and‑go distance at Coombing Park measured from OIS flight data. The results of this calculation are shown by the ‘YCPK T/G est. at YORG (no grass)’ bar in Figure 17. In addition to this calculation, a correction for the grass runway was applied by multiplying TODR_YCPK by 1.25, as recommended for a dry grass runway in the AFM for OIS and shown by the ‘YCPK T/G est. at YORG (grass)’ bar in Figure 17.

The estimated equivalent length of the touch‑and‑go conducted at Coombing Park ALA to Orange Airport accounting for grass was 812 m, about 8% longer than the shortest touch‑and‑go at Orange Airport. Without accounting for grass, the Coombing Park ALA touch‑and‑go projection was about 260 m longer than the shortest touch‑and‑go at Orange Airport, and marginally longer (about 15 metres) than the first touch‑and‑go conducted at 1556. The surface at Coombing Park appeared in good condition at the time of the accident, with the surface being firm, the grass relatively short, and the conditions dry. It is expected that the grass correction factor is an upper limit for the calculation of the potential length, with the lower limit being the calculation without accounting for grass.

The touch-and-go approaches at 1609 and 1619 had lift-off points considerably further along the runway than the other runs and were consistent with flight assessment exercises being conducted. The altitude profile of the 1609 touch‑and‑go was consistent with a go-around followed by a second, later touch-down, possibly simulating an early upwind engine failure. The touch‑and‑go at 1619 was likely following a practiced glide approach onto the reciprocal runway 11. These exercises were expected elements to assess the suitability to conduct a CPL flight test.

As no wind correction was applied to the estimated projection of the Coombing Park ALA touch‑and‑go length to Orange Airport, this calculation reflects the wind speed and direction as applicable at Coombing Park ALA during the touch‑and‑go at 1708. The Bureau of Meteorology identified that there was unlikely to be a headwind component during the touch‑and‑go at Coombing Park ALA. Further, based on the reported wind conditions at 1600, it is likely that a headwind component of approximately 6 kt was present at about the time of the shortest touch‑and‑go at 1602 at Orange Airport. In addition, the touch‑and‑go at 1556 likely used more runway than the Coombing Park ALA touch‑and‑go if this was conducted at Orange Airport, despite the likelihood of a 6 kt headwind component at Orange Airport during this touch‑and‑go. Therefore, a weaker headwind component is expected to have contributed to a longer landing at Coombing Park ALA than the shortest touch‑and‑goes conducted. However, based on the uncertainty in the meteorological data, and the ground speed of OIS from flight data being close to the published short field approach speed, it was not possible to determine the precise impact of wind on the touch‑and‑go distance at Coombing Park ALA.

In summary, compared to the length of the touch‑and‑goes conducted at Orange Airport, the Coombing Park ALA touch‑and‑go was the third shortest of the flight. The touch‑and‑go at Coombing Park ALA was the third shortest touch-and-go of the flight with and without density altitude corrections. Although the precise impact of the grass surface could not be determined, it is expected that this would have had the effect of reducing the projected length toward the two shortest touch‑and‑goes. Additionally, the touch‑and‑go conducted at Orange Airport at 1556 likely used more runway than an equivalent length touch‑and‑go conducted at Coombing Park at 1708, despite the 1556 touch‑and‑go likely having a larger headwind component. Based on analysis of the flight data for OIS, it appears likely that the touch down length was within the normal variability of touch-and-goes conducted by the student. Therefore, the above suggests that the length of the touch-and-go conducted by the student at Coombing Park ALA was typical of touch-and-goes conducted at Orange Airport earlier in the flight.

Estimation of standing take-off distances

The ATSB conducted an analysis examining the potential effect from performing a standing take‑off instead of the touch‑and‑go at Coombing Park ALA under the same conditions. The purpose was to estimate the maximum distance where OIS would have lifted-off along the runway at Coombing Park ALA if a standing take-off was conducted using two independent methods. One method (the ratio method) to estimate this distance was similar to the projection of the Coombing Park ALA touch-and-go to Orange Airport described above. For this method, the ratio between the standing take-off at 1637 and the measured touch‑and‑go distance was calculated, and this result was multiplied by the length of the Coombing Park ALA touch-and-go distance (929 metres). The other method was estimated standing take-off distance based on the AFM. Results from both methods are shown in Table 2.

Table 2: Standing take-off distance estimations for OIS at Coombing Park ALA runway 07

The ratio of the shortest touch‑and‑go at 1602 at Orange Airport was calculated as 0.90, equating to the longest estimated standing take-off length of 836 m. The same was performed for the 1556 touch‑and‑go, with a ratio of 0.68, and representing an upper limit of the normal touch‑and‑goes conducted. This equated to the shortest estimated standing take-off at 630 m. Standing take-off estimates based on the take-off distance required from the AFM for OIS were between the ratio‑based estimates. These were both factored by 25% to account for the short-dry grass runway for no wind, and a 5 kt tailwind.

The headwind component of the 1556 and 1602 touch‑and‑goes on runway 29 (about 6 kt) was likely greater than the headwind component of the standing take-off on runway 11 (about 1 kt), based on the reported wind conditions in METARs at 1600 and 1630. A greater headwind component would have allowed OIS to become airborne in a shorter distance with all other conditions the same. Given these conditions, the actual ratio between the standing take-off length and minimum touch-and-go length is expected to be slightly lower, making the estimated standing take-off distances slightly shorter. This indicates that the difference between the standing take-off and touch-and-goes using the ratio method is expected to be larger, and is therefore conservative.

In summary, the standing take-off length at Coombing Park ALA was estimated to be between 630 and 836 m, notably shorter than the actual Coombing Park ALA touch‑and‑go length of 929 m.

Initial climb projections

The ATSB evaluated the potential influence of a standing take-off on the height of OIS during the initial climb. This analysis used the estimated standing take-off distances described in the section above and is illustrated in Figure 18. The vertical axis shows the height above the estimated point of lift off and the horizontal axis shows the distance from the start of runway 07 at Coombing Park ALA, and also corresponds to the markers shown in Figure 6. Terrain elevation is shown by the solid dark green area, with typical tree heights indicated by the translucent dark green (minimum 10m) and light green (maximum 20m) areas.

Figure 18: Height of initial climb profile for OIS at Coombing Park ALA, including estimations for climb profiles following standing take-off and estimated terrain profile

Figure showing plot of height vs distance travelled of OIS during the initial climb from Coombing Park ALA runway 07 compared to terrain and estimates for climb if a standing take-off was conducted.

Source: ATSB

The actual initial climb profile of OIS conducted at Coombing Park ALA at 1709 is shown in Figure 18 by the series with red lines and blue circles. The last data point of this series indicates the relative height of OIS immediately prior to the left turn. The red dotted series is the line of best fit for the 1709 initial climb and shows an estimated projection of the climb if the climb rate was maintained and OIS tracked in a straight line instead of turning left. The top right corner of Figure 18 shows the projected climb above the minimum nominal tree height but intercepting below the maximum nominal height.

Theoretical climb profiles representing each of the estimated standing take-off lengths shown in Table 2 are shown by the dashed light blue, dark blue, yellow, and orange lines. These were formed by translating the actual climb profile horizontally (indicated by the red-dotted line) to commence from each calculated standing take-off distance along runway 07.

The orange dashed line in Figure 18, labelled ‘YCPK standing take-off estimate (from YORG T/G 1556)’, shows the best-case estimate for a climb profile following a standing take-off at Coombing Park ALA, based on the estimated standing take-off length of 630 m. This estimate translated to be an estimated 70 ft higher than the climb following the actual touch-and-go at Coombing 1709 initial climb at each point in the climb. This was based on the mean difference between the intercepts from the equations of these lines shown in Figure 18. Similarly, the height of the worst‑case estimated standing take-off (836 m) was estimated to be 22 ft higher than the climb following the actual touch-and-go at Coombing Park ALA (shown by the dark blue dashed line in Figure 18). This represents the lower limit of the height difference. The climb profiles for the AFM-based standing take-off estimates listed in Table 2 are shown by the yellow and light blue dashed lines in Figure 18, between the other projections.

In summary, the vertical height difference between a standing take-off climb profile and the touch‑and-go climb profile was estimated by the ATSB. This was based on calculating the vertical difference between the actual climb profile following the touch-and-go at Coombing Park ALA, and theoretical climbs originating from the estimated location of standing take-offs. The analysis found that if a standing take-off was conducted at Coombing Park ALA, the height at each point during the initial climb would have likely been between 22 and 70 ft higher than the actual touch-and-go conducted by the student. The top right corner of Figure 18 shows all climbs following hypothetical standing take-offs as passing close to, but above the maximum nominal tree height at the local terrain peak.

Conclusion

Based on analysis of the flight data for OIS, the ATSB found that the length of the touch-and-go conducted by the student at Coombing Park ALA was typical of touch-and-goes conducted at Orange Airport earlier in the flight. The length of all touch-and-goes during the flight were notably longer than the standing take-off conducted at Orange Airport when corrected for density altitude (when the length was measured from the runway threshold to the point of take-off). These touch‑and-goes were also all longer than calculations using the aircraft flight manual for the accident aircraft (VH-OIS). When corrected for density altitude, a standing take-off length at Coombing Park ALA was estimated by the ATSB to be between 630 and 836 m for the atmospheric and weather conditions at the time of the accident, notably shorter than the actual Coombing Park ALA touch‑and‑go length of 929 m. This equated to OIS being between 22‑70 ft higher at each point during the initial climb if a standing take-off was conducted instead of a touch-and-go. In contrast to the initial climb after the actual touch-and-go projecting below the maximum tree height, climb projections from standing take-off estimates indicated that OIS as passing close to, but slightly above the maximum nominal tree height.

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2023 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

Safety summary

What happened

On the morning of 23 October 2020, an Airbus A320-232 was being operated by Jetstar Airways on a flight from Brisbane to Cairns, Queensland. As power was being applied for take-off, the crew reported feeling a vibration and hearing a ‘popping’ noise that rapidly increased in frequency and volume. At the same time, the aircraft diverged to the right of the runway centreline despite the pilot flying applying full left rudder pedal deflection. The captain immediately selected reverse thrust and brought the aircraft to a stop.

Some passengers onboard, a Brisbane air traffic tower controller, and flight crew of a following aircraft all reported momentarily seeing flames coming out the right engine. The aircraft was taxied back to the airport gate. Engineers then reported finding metallic debris in the tailpipe of the right engine.

On disassembly of the engine, it was discovered the high-pressure compressor (HPC) had sustained significant damage and a screwdriver tip was found in the combustion section.

What the ATSB found

The screwdriver tip was determined to have been in the engine for over 100 flights. The ATSB concluded the tool bit had been left in the engine after maintenance and when the engine was running, it entered the HPC leaving dents and nicks in numerous rotor blades and stator vanes. At least two of these defects initiated fatigue cracks that resulted in a blade failing during the occurrence take-off. The liberated blade caused greater damage to the HPC and the engine surged.

What has been done as a result

As a result of the occurrence Jetstar Airways issued a Safety Alert to their maintenance engineers, which highlighted the need for all tooling to be accounted for. They also conducted a risk assessment to better understand the on-going risk.

Safety messages

Tool control is an important part of maintenance processes that ensures they do not lead to foreign object damage. Small and seemingly insignificant tool components can, and have, caused significant incidents or accidents. Tool control should extend to pseudo consumable items such as screwdriver tips and drill bits.

Modern training methods include the use of high-fidelity training devices such as full motion flight simulators. Their design aims to maximise the realism of an artificial environment. However, there is a limit to their ability to replicate extreme events. Flight crew should be aware that the noise and vibration from an actual engine failure may be greater than, or different to, that experienced during simulator training and this could contribute to the effects of startle.

The investigation

The occurrence

On the morning of 23 October 2020, an Airbus A320-232 was being operated by Jetstar Airways on a flight from Brisbane to Cairns, Queensland. The first officer was the pilot flying and the captain was the pilot monitoring.^[1]

After push back, the first officer taxied the aircraft for a planned take-off on runway 01L. The crew completed the pre-take-off checks during the taxi in anticipation of a rolling start to the take-off. After being cleared for take-off on approach to the runway, the aircraft was lined up and, without stopping, the thrust levers were set for Flex.^[2] The crew reported that as both engines spooled up to the requested power setting, a vibration and ‘popping’ noise could be heard that rapidly increased in frequency and volume. At the same time, the aircraft diverged to the right of the runway centreline despite the first officer applying full left rudder pedal deflection. In response, the captain immediately selected reverse thrust and brought the aircraft to a stop. The aircraft reached a maximum ground speed of 30 knots.

Upon returning the thrust levers to idle, the vibration subsided and the popping stopped. The aircraft instrumentation then sounded two master caution alarms and the Electronic Centralised Aircraft Monitoring (ECAM) system^[3] reported two messages for Engine No. 2 (right) ‘STALL’ and ‘EGT LIMIT EXCEEDANCE’.^[4] The captain called for the cabin crew to move to stations.^[5]

The cabin crew relayed to the flight crew that passengers had seen flames coming from the right engine. Both the airport tower controller and the flight crew of a following aircraft similarly reported seeing flames coming from the rear of the right engine for a short period of time. The tower dispatched Airport Rescue and Firefighting services, who observed no apparent abnormalities with the aircraft.

The captain cancelled the flight and the aircraft was taxied back to the airport gate where both engines were shut down. Immediately after the passengers had disembarked, engineers reported finding small balls of metallic debris in the tailpipe of the right engine (Figure 1). A review of recorded data indicated the right engine had surged.

Figure 1: Debris in right engine tailpipe

Source: ATSB

Context

Engine Description

The IAE V2527-A5 engine is a variant of the V2500 family of engines. The two‑spool, axial‑flow, high‑bypass turbofan comprises low- and high-pressure stages. The low-pressure section consists of a fan stage and four low-pressure compressor (LPC) stages (labelled booster stages) connected to five low-pressure turbine (LPT) stages. The high-pressure section consists of ten high-pressure compressor (HPC) stages (numbered 3 to 12) connected to two high‑pressure turbines (HPT) (Figure 2).

Figure 2: IAE V2527-A5 engine station and stage numbering diagram

Source: IAE Training Manual.
Note: The low-pressure spool is shown in grey and the high-pressure spool in black.

An engine surge is caused by a loss of air compression on the inlet side of the engine. This can be caused by disrupted airflow ahead of the engine, ingestion of foreign matter, excessive thrust changes or an airflow disruption internal to the engine.

Regardless of the cause, the sudden breakdown in compression results in reversal of the flow and violent expulsion of the previously compressed air out the inlet of the engine. The process is often cyclic resulting in a vibratory noise. Additionally, disruption to the airflow can allow unburnt fuel to escape the combustion chamber and ignite in the turbine section or tailpipe.

Post-flight Inspections

On wing inspection

A borescope inspection of the right engine found there was damage to HPC stages 3, 5, 6, 7, 8 and 9 (Figure 3) consistent with foreign object damage (FOD).^[6] As the damage was outside maintenance manual limits, the engine was removed from the aircraft and sent for a teardown inspection.

Figure 3: Borescope image of stage 7 HPC and stage 8 stator vanes

Source: Jetstar

Teardown inspection

Upon disassembly of the engine, the following observations and anomalies were found:

• The fan blades and LPC contained no visible damage.
• The rear fairing to the LPC bleed valve mechanism was in place but had one Torq-set screw with a damaged head. Although of an incompatible shape, testing showed that due to the damage, the head would also accept a Phillips screwdriver tip (Figure 4).
• In the HPC section, rotor blades (blades) and stators (vanes) in stages 3 and 4 contained dents and minor tears.
• Approximately half of one stage 5 blade (Figure 5) and a whole stage 6 vane (Figure 6) were missing and had failed due to cyclic fatigue. Some of the remaining stage 5 blades exhibited tip curl,^[7] dents and/or missing material.
• Four stage 7 blades and one stage 8 blade were missing and had failed due to overstress. In this section all remaining blades and vanes exhibited hard body damage resulting in severe bending and tearing to both leading and trailing edges. One stage 8 vane contained a distinctive hexagon shaped dent (Figure 7).
• Damage after stage 8 was limited to minor dents and nicks with minimal damage occurring to the final stage components.
• Bleed valve screens at stages 7 and 10 contained large amounts of metallic fragments.
• Small metallic debris and a ¼” hexagon drive screwdriver tool bit (screwdriver tip) was found between the combustion liner and engine case. The screwdriver tip was burnt/discoloured and the tip eroded due to heat and mechanical damage (Figure 7).
• Nozzle Guide Vane (NGV) cooling passages contained small metallic debris.
• Two stage 1 HPT blades contained leading edge nicks attributed to material impacts.
• Small debris was found throughout the LPT.

Figure 4: Interstage screw heads

Source: Jetstar

Figure 5: Stage 5 blade fatigue failure

Source: Jetstar, annotated by the ATSB

Figure 6: Stage 6 vane fatigue failure

Source: Jetstar, annotated by the ATSB

Witness marks on many components throughout the HPC matched the shape of the screwdriver tip found in the combustion section (Figure 7).

Figure 7: Screwdriver tip and witness marks

Source: Jetstar

A failure analysis was conducted by an engine maintenance facility. The fractured stage 5 blade showed minor leading edge mechanical damage at approximately half span, which initiated a fatigue crack that propagated through approximately two-thirds of the chord. The remaining third of the chord failed in material overstress.

The fractured stage 6 vane sustained a tear near the aerofoil root to a depth of approximately 10 per cent chord. This initiated a fatigue crack that ended 2 mm from the trailing edge with the remaining material exhibiting a material overstress signature.

Both fatigue fracture surfaces displayed fine striations consistent with a high frequency load spectrum, likely from aerodynamic vibrations. While no fatigue growth rate analysis was conducted, based on similar occurrences, it was estimated the components took tens of flight cycles to fail.

Aircraft maintenance program

The aircraft maintenance program required a fairing to be removed in the engine to lubricate the LPC Bleed Valve Mechanism every 4,000 flight hours. The task was last accomplished 112 flight cycles prior to the engine surge. This task occurred with many others as part of preparations to return the aircraft to service after being parked for approximately 4 months.

The LPC Bleed Valve Mechanism task procedures contained specific highlighted caution notes regarding the loss of any screws or other loose objects down the bleed duct. The note highlighted that lost articles would progress to the HPC and break blades and vanes. This occurs because during engine start and idle the bleed valve opens.

Engine failure during take‑off

Both flight crew commented that the event startled them due to its unexpected onset, the volume of the popping noise and particularly the levels of vibration. Additionally, the pilot flying reported that, despite the application of full left rudder, the aircraft still diverged to the right of the runway centreline. Both pilots commented that the noise and vibration was far more severe than what they had experienced during flight simulator training sessions when practicing for similar types of events.

Flight crew training includes many different failure scenarios. Being a particularly critical phase of flight, take-off is trained for extensively including engine failures that involve high-speed rejected take-offs (RTO) and post V₁ flight continuations.^[8] One crewmember commented they had never performed a low-speed RTO with engine failure.

For a high-speed engine failure, crews are trained to recognise the failure and maintain runway direction through use of the rudder, which is effective at speeds above V_MCG.^[9] The take-off is then either continued or rejected. If rejected, the thrust asymmetry is reduced by idling the remaining engine/s.

Low-speed RTOs are those which occur below V_MCG where the aircraft’s rudder is not effective enough to counter the large thrust asymmetry and thus the aircraft will diverge from the runway centreline uncontrollably if continued. This situation is countered by differential braking and promptly returning the engines to idle, thereby reducing or nullifying the thrust asymmetry.

Low-speed RTOs, while considered less dangerous because they occur at lower speeds, involve a loss of directional control. Lateral divergence from the runway centreline can be larger than for a high-speed engine failure (>V_MCG) and can cause considerable damage if the aircraft departs the side of the runway.

Related Occurrences

ATSB investigation (199800615)

On 28 February 1998, at a height of about 100 ft after take-off, the crew of a Boeing 767 reported hearing a series of loud bangs and the right engine exhaust gas temperature indicator rose rapidly into the red range. The right thrust lever was retarded to idle, resulting in the temperature indications returning to normal. ATC also observed a series of flashes from the right engine as the aircraft departed and declared a local emergency. The crew returned the aircraft to the airport and landed without further incident.

An internal borescope inspection revealed extensive damage to blades of the thirteenth compressor stage. The engine was subsequently removed for further inspection, resulting in a replaceable Philips screwdriver tip being found in the core of the engine.

The operator reported that the incident occurred on the first flight after the aircraft had undergone an 'A' maintenance check. The screwdriver tip probably entered the engine through the variable bleed valves, which are open when the engine is not operating.

Safety analysis

The damage to the HPC section and the screwdriver tip found in the cooling cavity were consistent with the screwdriver tip falling into the LPC bleed duct, passing through the interstage duct and into the HPC. The lack of damage to the LPC indicated the tip was not FOD from the runway. The damage to the rear fairing screw meant it was likely the screwdriver tip had been left in the screw or had fallen into the bleed duct during maintenance. At the next engine start-up, it then travelled through the HPC section striking various components, in some cases leaving witness marks, before being held captive adjacent to the combustion chamber.

There was evidence that fatigue cracks initiated from the impact damage in a stage 5 blade and stage 6 vane. These cracks propagated until one reached a critical size due to aerodynamic forces and vibrations, and additionally for the blade, centripetal forces. As the engine surge event occurred during the application of take-off power and acceleration of the engine internal componentry, it is more likely the blade failed first due to the increase in centripetal acceleration.

Regardless of whether the blade or vane failed first, one likely impacted the other and both then caused downstream damage, including the failure of a further five compressor blades. The resulting disruption to aerodynamic flow through the HPC allowed the engine to surge and led to the loss of engine performance.

Findings

From the evidence available, the following findings are made with respect to the engine power loss.

Contributing factors

• A screwdriver tip, left in a fairing screw or bleed duct, travelled through the high‑pressure compressor (HPC) section of the right engine before becoming lodged in the combustion section. This resulted in impact damage to HPC blades and vanes.
• Fatigue cracks initiated at damaged locations in at least one HPC blade and vane, with one crack propagating to failure. This resulted in secondary damage to the engine and surging on take-off.

Safety actions

Safety action by Jetstar Airways

As a result of this incident, the operator advised the ATSB:

• the Jetstar Aircraft Maintenance Organisation had issued a Safety Alert to maintenance engineers, which highlighted the need for all tooling to be accounted for.
• they had conducted a risk assessment to better understand the on-going risk.

Sources and submissions

Sources of information

The sources of information during the investigation included:

• the flight crew
• VH-VFF direct access recorder
• Jetstar Airways
• Christchurch Engine Centre

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• the flight crew
• Jetstar Airways

Submissions were received from:

• Jetstar Airways

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2021 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

__________

1.  Pilot Flying (PF) and Pilot Monitoring (PM): procedurally assigned roles with specifically assigned duties at specific stages of a flight. The PF does most of the flying, except in defined circumstances. The PM carries out support duties and monitors the PF’s actions and the aircraft’s flight path.
2.  Flex is an engine thrust setting used for take-off which is below the maximum rated thrust for an engine. The engine thrust requested for a given take-off is calculated based on aircraft weight, flap setting, runway length available and ambient temperature, and is used to reduce engine wear.
3.  Electronic Centralised Aircraft Monitoring (ECAM) is a system that monitors aircraft functions and relays their status to flight crew. It also produces messages detailing failures.
4.  EGT – Exhaust Gas Temperature
5.  A ‘Call to stations’ is a notification via the public address system from the flight crew to the cabin crew that they are assessing an emergency situation and that an evacuation may be necessary. On receipt of this notification, cabin crew prepare by undertaking actions such as checking conditions outside their assigned door, maintaining control of passengers in the cabin and reporting any abnormal conditions to the flight crew.
6.  Foreign Object Damage (FOD) is any article or substance, alien to an aircraft or system, which could potentially cause damage. The term FOD is used to describe both the foreign objects themselves and any damage attributed to them.
7.  Tip curl is a signature of compressor or turbine blades making contact with the engine casing.
8.  V1 is a decision speed where, thereafter, the pilot can continue the take-off safely using the remaining runway on the remaining engine/s. If a decision to abort the take-off occurs before V1 is achieved, the aircraft has enough distance to safely come to a stop on the remaining runway available. It can be said that V1 is the ‘commit to fly’ speed. V1 is calculated for every take-off as it is based on aircraft available thrust, aircraft weight, flap setting, runway length and slope, wind conditions, and airport density altitude.
9.  VMC means the minimum control speed with the critical engine inoperative and the remaining engine/s at maximum power permitted for the situation. VMC can be further separated into VMCG (ground) and VMCA (airborne) which, while both involve engine failure, are limited by different factors.

Overview of the investigation

The occurrence

On 2 October 2020, at about 1117 Eastern Standard Time,^[1] a SCT Logistics freight train 3PG1 travelling from Port Germain, South Australia to Parkes, New South Wales, began to lose brake pipe air, resulting in the application of the brakes. The train subsequently came to a stand with the leading locomotive at about 904 track km, 90 km west of Ivanhoe, New South Wales.

In response to the loss of air, the driver and driver assistant (both employed by Momentum Rail) commenced an inspection of the wagons. However, they stopped before arriving at the back of the train as the driver considered the issue to be a computer error. They returned to the lead locomotive and the driver tried to restore the brake pipe air by resetting the electronic air brake system, which was unsuccessful. The driver assistant re‑inspected the wagons, finding and closing an open brake pipe tap on the back of the last wagon (ABSY2792Y). They reported this to the driver, but for differing reasons, the end of train marker was not checked or the number of the last wagon was not confirmed.

The train’s brake pipe air was restored, allowing the brakes to release. The driver reported to train control that the loss was due to an animal strike and the trip resumed. On arrival at the Darnick yard limit board (YLB), the previously issued train order (Kaleentha loop to Trida YLB) was cancelled and a new train order from Darnick YLB to Ivanhoe YLB was issued. The crew of 3PG1 did not check whether their train was complete at this location and the train controller did not request confirmation, as required by the Australian Rail Track Corporation train order working procedure.

However, due to concerns with not having checked the end of train marker, the driver assistant conducted a roll-by inspection as the train proceeded into the Ivanhoe loop. At that time, they identified that the train consist was incomplete and the last 4 wagons were missing. The crew concluded that the missing wagons were likely left behind at the point where the train had lost brake pipe air pressure, about 2 hours prior. The section of track with the 4 parted wagons was left without protection with another train waiting at Darnick to enter the occupied section.

Investigation activities

During the investigation, the ATSB:

• interviewed the train crew and network controller
• conducted analysis of data from the locomotive’s event recorder
• reviewed recorded communications between train control and the crew
• reviewed the track access and labour agreements in place between the involved parties
• reviewed the experience, training and competencies held by the train crew
• conducted analysis of possible contributors to the parting, including a review of the wagon inspection results and maintenance history
• analysed the practical actions required, including actioning applicable rules and procedures when train crews respond to a loss of brake pipe air event
• reviewed the tasks conducted post-occurrence to recover the lost wagon.

ATSB observations

From the investigation, the ATSB determined:

• The coupler locking block and knuckle on the leading end of wagon ABSY2792Y were worn, which very likely resulted in 4 rear wagons parting, and train 3PG1 subsequently loosing brake pipe pressure.
• The crew did not identify that the end of train marker was not present on the last coupled wagon or otherwise identify that the 4 rear wagons had parted. Instead, during an inspection of the train, the driver assistant closed an open brake pipe air tap on the back of the last coupled wagon, which allowed all the wagon brakes to release. Consequently, the train resumed the journey incomplete.
• The 4 parted wagons were left occupying the track section when the protection on the section of track (Kaleentha to Darnick) was removed. The crew of 3PG1 cancelled the existing train order without conducting a further inspection.
• The driver assistant was not experienced in the role they were undertaking. This, combined with having only partially completed the relevant driver qualification, contributed to them not recognising the potential reasons for an open air tap on the back of the last wagon, which included the risk that the train had parted. This likely compromised the crew's understanding of the occurrence and led them to incorrectly attribute the problem to a known cause.
• The lock blocks on three couplers, including the coupler that initiated the parting, were not replaced at previous 'B' inspections. However, the replacement was a requirement in SCT Logistics’ maintenance provider’s (Gemco Rail) procedures. Subsequently, the component was in use past its intended service life when the parting occurred.

Reasons for the discontinuation

The ATSB gives priority to transport safety investigations that have the potential to deliver the greatest public benefit through systemic improvements to transport safety.

Given the ATSB’s constrained resources, the ATSB considered it was unlikely that further investigation would identify systemic safety issues or identify opportunities for the enhancement of transport safety. Consequently, the ATSB has discontinued this investigation.

The ATSB has briefed SCT Logistics and Momentum Rail about its observations and potential learnings. However, it considered that broader communication of this information would not be of significant benefit to other parties.

The evidence collected during this investigation remains available to be used in future investigations or safety studies. The ATSB will also monitor for any similar occurrences that may indicate a need to undertake a further safety investigation.

__________

1.  Eastern Standard Time (EST): Coordinated Universal Time (UTC) + 10 hours.

This investigation was conducted under the Transport Safety Investigation Act 2003 (Commonwealth) by the Office of Transport Safety Investigations (NSW) on behalf of the ATSB in accordance with the Collaboration Agreement. Released in accordance with section 25 of the Transport Safety Investigation Act 2003.

Safety summary

What happened

On 13 October 2020, TrainLink passenger service SN68, was operating from Moss Vale to Campbelltown, New South Wales with approximately 20 passengers on board.

The train driver stopped the train at Yerrinbool Station after being alerted to a small fire at the rear of the train. The fire was identified as coming from the vicinity of an axle bearing on the rear wheelset of the train. The passengers were evacuated onto the platform and the driver attempted to extinguish the fire using an on-board fire extinguisher.

The fire was subsequently extinguished by Fire and Rescue NSW. As a result of the fire, parts of the axle box were heat affected and sustained significant damage to the speed sensor and rubber suspension components. There were no reported injuries.

What the ATSB found

The investigation determined that the fire was the result of a collapsed axle bearing on wheel 8 on the trailing bogie on car 2811, the end of the train. The axle bearing failed when the axle end cap bolts loosened and one fractured which caused the collapse of the bearing and frictional heat to be generated. The resulting fire was fuelled by grease, oil and rubber suspension components in the immediate vicinity of the axle box.

It is likely that during the last overhaul of bogie NJA31, the locking plate tabs retaining the axle end cap bolts were not fitted correctly against the sides of the bolts. The axle bearing installation process was not sufficient to ensure the tabs on the locking plate were installed correctly during a refurbishment three months before the incident.

A wayside monitoring system at Burradoo on the Down Main line detected an elevated temperature on one bearing, but the temperature recorded was below the threshold for an alarm to be sent to network control.

What has been done as a result

Following the occurrence Sydney Trains, which is the maintenance provider for NSW Trains, initiated an inspection of similar axle boxes in the fleet and undertook an audit of the contracted maintainer’s practices.

Sydney Trains have advised the following actions have taken place to prevent a recurrence:

• Improvements have been made to the contracted maintainer’s quality assurance processes to ensure that bolts and locking tabs are correctly installed.
• An improved process was implemented to review and retain the contracted maintainer’s certificate of completion checklists.

Safety message

Bearing failures continue to occur within the Australian rail network. This occurrence emphasises the significance of having adequate bearing installation processes and ensuring that axle bearings are correctly maintained and monitored throughout their operational life.

The Occurrence

On Tuesday 13 October 2020, TrainLink passenger service SN68, operated by NSW Trains,^[1] departed Moss Vale at 1757.^[2] The two-car Endeavour train was crewed by a driver in the front cab and a guard in the rear cab. There were approximately 20 passengers on board the train as it departed Moss Vale (Figure 1). The train was to make 11 stops before being scheduled to arrive at Campbelltown at approximately 1900.

The train stopped at Burradoo and Bowral before departing Mittagong at approximately 1809. The train crew said they had experienced no problems with the train before Yerrinbool.

At 1820, as the train was slowing to stop at Yerrinbool Station, the guard, from inside the cab at the rear of the train, heard a loud noise and noticed smoke outside the window. The guard used the train’s bell system to ask the driver to stop. The driver brought the train to a stand at Yerrinbool Station.

Figure 1: Incident location and path of SN68

Source: Geoscience Australia, annotated by OTSI

Once the train stopped at Yerrinbool Station, the guard used the trains’ public address system to ask the passengers to move to the front of the train. The guard made a second announcement shortly afterwards, requesting passengers disembark onto the platform. The driver walked along the platform to the rear of the train and observed a flame and dark smoke coming from the last wheelset of the train, wheel 8 on axle 4 under the bogie of car 2811.

The driver spoke to an Australian Rail Track Corporation (ARTC) network controller at Junee requesting permission to go down onto the track to attempt to extinguish the fire. The network controller applied signal blocking to prevent rail traffic in both directions and then gave permission for the driver to access the track. Another passenger service, SN61, was stopped by the signaller at a signal before the platform at Yerrinbool. This provided protection on the adjacent line so the driver could go onto the track. The driver went onto the track and used the on-board fire extinguisher to attempt to put out the fire. The fire continued to smoulder/burn as the heat source remained.

Fire and Rescue NSW arrived on site at approximately 1828 and ensured the fire was extinguished before departing at 1850. Train passengers were transferred to buses which replaced train services between Campbelltown and Moss Vale in both directions. There were no reported injuries as a result of this incident.

The maintenance shift manager at Eveleigh Maintenance Centre organised for the Rail Emergency Recovery Unit to arrange pony bogies (Figure 2) to be fitted under all wheels of bogie NJA31 and the train was worked back to Eveleigh over the next two nights. Under the supervision of Office of Transport Safety Investigations (OTSI) investigators, Office of the National Rail Safety Regulator representatives and Sydney Trains engineering staff a partial disassembly of the axle end cap and removal of bolts from the affected bearing was undertaken. A further strip down inspection of the axle bearing assembly was conducted at bogie maintainer United Group Limited Unipart (UGLU) at Auburn, also under the supervision of OTSI investigators and Sydney Trains engineering staff.

Figure 2: Heat affected area and pony bogie fitted under NJA31

The inset image shows the axle cover removed. Two bolts were found to be missing from the end cap and one bolt was broken. All three bolts remained within the axle housing.

Source: OTSI

The damage was contained to this localised area around the trailing axle of the rear bogie. Parts of the axle box were heat affected and the speed sensor and rubber suspension components were significantly damaged.

__________

1. NSW Trains is an operating agency of Transport for New South Wales, it is responsible for the operations of TrainLink services.
2. Times shown in 24-hour time as Australian Eastern Daylight Time (AEDT).

Context

Environment

The Bureau of Meteorology (BOM) automatic weather station at Moss Vale, recorded the temperature as 24.4 °C at 1500 on 13 October 2020. Yerrinbool is approximately 30 km north-east of Moss Vale. Weather conditions were fine and clear.

Location

Yerrinbool Station is on the Main South line in the Southern Highlands of New South Wales (Figure 3). Yerrinbool is located at 116.310 km.^[3]

Figure 3: South-bound Endeavour two-carriage set at Yerrinbool Station

The figure shows south-bound (Down) Endeavour set at Yerrinbool Station. Image of car 2811 inset

Source: railgallery.wongm.com, annotated by OTSI

Train crew

The train was crewed by a driver, operating the train in the front driver’s compartment, and a guard located in the rear drivers’ compartment. The train crew were appropriately qualified and held the required route qualifications.

Train information

Train SN68

The passenger train involved in the incident, SN68, was an Endeavour railcar two-carriage set. This diesel-powered multiple unit train was operated by TrainLInk and built by ABB Transportation in Dandenong, Victoria. There were 14 Endeavour sets in service at the time of the incident and they first entered service in March 1994. Twenty-three Xplorer cars which have the same bogie types and axle bearings were also in service at the time of the incident.

The leading car of SN68 was LE2861, with TE2811 being the trailing car. The LE carriages feature a dedicated luggage space and can seat 95 passengers. The TE carriages feature a wheelchair accessible toilet and can seat 82 passengers.

The drivers’ cab, positioned at each end of the train, is a full width driving compartment with the driver's seat offset to the left-hand side. Passengers enter and exit through power activated doors operated and controlled by the driver or the guard. Dry chemical fire extinguishers are fitted in the drivers’ compartment.

Bogies

The bogies fitted to the Endeavour and Xplorer sets are NJA and PJA bogies. The bearing collapse occurred on an NJA bogie, the rear trailing bogie on car 2811. The NJA bogie is the trailer bogie (Figure 4), as opposed to the PJA powered bogie. Built for the State Rail Authority in 1994, the bogie involved in this incident was designated as NJA31. This bogie was installed on    car 2811 on 7 July 2020 and had travelled approximately 66,087 km since installation.

Figure 4: NJA31 bogie
 

Source: OTSI

Axle bearing installation

The maintenance and installation processes for critical components holding the axle bearing were examined as part of the investigation.

Each bogie has two wheelsets which have an axle bearing at each end of the axle (four bearings per bogie). The axle bearings were Timken SP130 type bearings. Securing the axle end cap were three different brands of bolts with the head markings showing: NLGS, JDF and HEC (Figures 5 and Figure 6). The axle end cap bolts were hexagonal head, metric 16 mm diameter (M16), 40 mm length (fully threaded), 2 mm pitch, and class 8.8. There are two types of end caps used, a standard or combined end cap suitable for mounting a phonic wheel. The phonic wheel is used as part of the train’s on-board system for recording speed and detecting wheel slide during braking, this end cap was the phonic wheel type.

Figure 5: Three types of axle end cap bolts removed from hub following incident

Source: OTSI

Figure 6: Side view of three axle end cap bolts, including broken NGLS bolt

Source: OTSI

The locking plate was a Timken brand K-422091 R.S 120-130. Stamped on the plate was the axle number (073275) and installation date (07 20 – July 2020) (Figure 7).

Figure 7: Locking plate

Source: OTSI

The requirements for installing the SP130 bearings onto the axles is documented in a Sydney Trains Standard Instruction.^[4] This instruction provided maintenance workers with the details for installing the end cap, torque requirements for bolts and installing the locking tabs. There was a wheelset certificate of completion used by the maintenance workers to show that bogie NJA31 was refurbished. This work was completed on 1 July 2020. There are check boxes on a form that the maintenance workers complete when each task is done. These were all checked as completed and the installation tolerances for the bearing were correct at the time of installation.

The axle bearing installation process commences with the bearing being pressed onto the axle journal before being retained on the journal by an end cap. The end cap is secured by three M16 bolts with a new locking plate positioned underneath the bolt heads. The bolts are screwed in threaded holes and using a calibrated torque wrench are finally torqued to 80 – 100 Nm.

The bolts are torqued in sequence until there is no further movement of the bolt resulting from the application of the specified torque. It may be necessary to further tighten the bolt to align the head with the locking plate tabs. The tabs on the locking plate are then bent up to engage with the sides of the bolt head to form a tight fit. After the process is complete a witness mark is applied to the bolt head and adjacent area to show the original position of the bolts (Figure 8). Movement can be visibly determined during any subsequent inspections.

Figure 8: Correct bolt and locking tab installation bolt

Inset image shows a close of the locking tabs correctly bent up and in contact with the head of the bolt as well as the witness mark (white paint pen).

Source: OTSI

Involved parties

The Australian Rail Track Corporation (ARTC) is the rail infrastructure manager that manages the Main South Line from Macarthur on the outskirts of Sydney to Melbourne, including the wayside monitoring device. 

NSW TrainLink provides rail services in NSW and also operates some interstate services to Victoria and Brisbane.

Sydney Trains is responsible for maintenance activities on the trains operated by Sydney Trains and NSW TrainLink.

United Group Limited Unipart (UGLU) are contracted maintenance providers to Sydney Trains and serviced the bogies of the Endeavour and Xplorer fleet. UGLU is a joint venture between United Group Limited and Unipart.

Track and infrastructure information

The section of track at Yerrinbool was standard gauge (1435 mm). It consisted of an Up Main line and a Down Main line. At the time of the incident SN68 was travelling towards Sydney on the Up Main line.

ARTC maintains operational control for this area from Network Control Centre South at Junee.

Wayside detectors

The ARTC, operates and maintains the wayside monitoring systems in the vicinity where this incident occurred. There are different types of wayside devices including but not limited to detectors for hot bearings, wheel impact loads, acoustic wheel monitoring, and dragging equipment (Figure 10).

At Burradoo, 138.000 km, on the Down Main line, an operational wayside device detected an elevated bearing temperature as SN68 as it passed, in the Down direction, on the way to Moss Vale. This occurred at 1703 when the elevated temperature was recorded on car 2811 (wheel 8 on axle 4), the location of the subsequent collapsed bearing. The temperature was recorded as 84°C, below the threshold to trigger an alarm. The temperature of the other bearings on the bogie were recorded as 73°C, 65°C and 69°C.

The ARTC has two threshold categories for hot bearing detector alarms.^[5]  When a threshold temperature is exceeded an alarm is sent to the network controller who must ensure the category of alarm condition is understood by the driver and is responded to according to the required action outlined below (Figure 9).

 Figure 9: Wayside Device Alarm Categories

Source: ARTC

The other wayside monitoring system between Campbelltown and Moss Vale was  located at Menangle on the Up Main at 67.25 km. It is likely that the hot bearing would have been detected here, 50 km past Yerrinbool.

Sydney Trains also checked previous passes by car 2811 over wayside monitoring systems and did not identify any other warm bearing temperature for this wheelset for the period 5 July 2020 to the incident date on 13 October 2020.

Figure 10: ARTC Network wayside detectors

Source: ARTC, annotated by OTSI

Related occurrence

On 1 December 2020, less than two months after the incident at Yerrinbool, a related incident occurred when a hot bearing on an Xplorer train was detected by a Sydney Trains wayside detector on the Up Main at Wyee, NSW. TrainLink passenger service, NP24, travelling from Armidale to Sydney was stopped at Wyee after the driver was notified of a hot bearing. After inspection, the train proceeded at a low speed (under 25 km/h) to Wyong where the passengers were disembarked. There was no injury or damage.

The Xplorer was subsequently examined at Eveleigh Maintenance Centre where it was found that two axle end cap bolts on car 2508 (wheel 8) had moved from their original position. It was also found that the locking plate tabs were incorrectly bent up. The last refurbishment of the wheelset was 22 months before the incident on 1 March 2019.

__________

3. The kilometre distance is measured from Platform 1, Central Station, Sydney, New South Wales.
4. Sydney Trains Standard Instruction, Installation of Metric SP type Package Unit Bearings, TRS1516.00, Vol 3, Section G, 8. Applying the end cap.
5. ARTC, Response & Management of Wayside Monitoring Device Alarms – Wayside Device Alarm Categories.

Safety analysis

Mechanism of bearing collapse

The investigation determined that the fire in the wheel area of SN68 at Yerrinbool was caused by a collapsed bearing on wheel 8 on the trailing bogie on car 2811. This led to frictional heat to be generated and a fire started. The fire was fuelled by grease, oil, and rubber suspension components in the immediate vicinity of the axle box. The fire did not spread to other parts of the train.

The likely precipitating events to the bearing collapse was the loosening of two of the three axle end cap bolts which placed higher loading on the remaining bolt which then started to fracture. During every cycle of the bearing there are micro movements within the system. The locking plate tabs, if positioned correctly against the face of the bolts, assist in retaining the bolts in position. The two bolts, inadequately restrained by the locking plate, continued to loosen and fell out and the remaining bolt then fractured completely (Figure 11). With the end cap unsecured the outer cone of the bearing was able to move on the axle. Wayside data indicated that once the bolt fractured it took approximately 15-20 minutes for the failure to be identified.

Figure 11: Axle end with protruding failed bolt

Insert image showing the inside collapsed axle box and failed bolt protruding.

Source: OTSI

An initial inspection of the axle bearing and components was conducted following the incident at Eveleigh Maintenance Centre. All three bolts and the locking plate were recovered and, although damaged, were able to be examined. There were three different brands of bolts used and the failed bolt was identified as a NLGS bolt. A further inspection at UGLU Auburn Maintenance Centre was conducted when the bogie was disassembled. Present at these inspections were representatives from OTSI, Sydney Trains and UGLU. Measurements were taken and recorded for critical items, such as bearing end float, end cap installation torque and axle diameters on the other three undamaged bearings. These measurements were all were consistent and within specification.

Torque values were also measured on the other three undamaged bearing axle end cap bolts. Four out of the nine torqued bolts remaining on the bogie exceeded the 80‐100 Nm torque range and two bolts were recorded at 180 Nm and 200 Nm. The differing torque amounts was attributed to bolts being tightened further to achieve the alignment with the locking tabs. This torque was the breaking torque, not the original application torque.

An independent metallurgical examination was also conducted on the bolts and the locking plate, as well as a sample of other similar bolts. The scope of this analysis included: examination and fractography, hardness testing on all bolts, microscopy analysis of the grain structure, tensile testing and compliance to applicable standards. A report was produced following this examination.^[6]

The metallurgical examination showed that the bolts had no material or surface defects and complied with the requirements of the Australian Standard.^[7] The use of differing bolt brands was initially identified as a concern but was later ruled out as an issue as all brands complied with the requirements of the standard. The metallurgical examination also showed the locking plate to be without issue. Other results from the metallurgical examination were:

- Damage was most severe on the first two bolts which loosened and fell out. These bolts were damaged after being tumbled around inside the case.

- The third bolt had the least damage as it had fractured after the first two bolt had fallen out.

- The third bolt had partially unscrewed about 4-5 mm before failure.

- The third bolt failure occurred progressively over a period of approximately one to two hours.

- The locking plates were made of low hardness steel which provided less resistance to loosening if the stresses in the system were sufficient to unscrew the bolts.

- The locking plate showed heavy damage and bent arms at the hole where the third bolt fractured.

The metallurgical report stated: ‘A potential factor in the failures was insufficient tensioning of the bolts on installation, however, annular impressions around the locking plate holes were mostly similar, which is a rough indication that torqueing of the bolts had been similar. This observation is very subjective, and the conclusion of similar torqueing may be erroneous.’ The investigation could not verify what torque level was applied to the axle end cap bolts that came loose.

Bearing installation process

An inspection of the processes undertaken at UGLU Auburn Maintenance Centre demonstrated that standard processes were in place, marked components were used, and calibrated torque wrenches available to be used by appropriately qualified personnel. Despite this, the physical evidence shows that it is likely that during the last refurbishment of the wheelset, on 1 July 2020, just over three months before the incident, the locking tabs were not sufficiently bent up against the face of the bolts.

There was completed documentation showing the processes had been checked off by a qualified maintenance person. The relevant checkboxes for all items were completed, this included:

- checking the bolt torque (new bolts used)

- tabs locked flat on the screw head face

- torque marking of screws.

However, at the time of the refurbishment there was no additional quality assurance check once this process was completed.

The subsequent Sydney Trains investigation report into this incident at Wyee on 1 December 2020 showed strong similarities with the incident at Yerrinbool. The detection of the locking plate tabs not being bent up supports the same findings for Yerrinbool.

__________

6. MetallTech, Metallurgical investigation Endeavour bolt failure – axle box bolt, Report M2020/09-1, 23 November 2020.
7. Australian Standard AS 4291.1 / ISO 898.1 Mechanical properties of fasteners made of carbon steel and alloy steel.

Findings

From the evidence available, the following findings are made with respect to the collapsed bearing on passenger train SN68 at Yerrinbool, New South Wales on 13 October 2020.

Contributing factors

• The fire that occurred on SN68 was the result of excessive heat generated from a collapsed bearing on the trailing bogie of locomotive TE2811.
• The bearing collapse was the result of two axle end cap bolts becoming loose and the remaining bolt fracturing due to the increased load. Once the three bolts were no longer holding the end cap in position the bearing rapidly collapsed.
• It is likely that during the last NJA31 bogie overhaul the tabs on the locking plate were not installed correctly. This enabled the two axle end cap bolts to loosen.
• The axle bearing installation process was not sufficient to ensure the tabs on the locking plate were installed correctly. (Safety Issue)

Other (key) finding

• The brake and bearing temperature alarm detected an elevated temperature on an axle bearing at Burradoo as SN68 passed over it approximately 60 minutes before the fire was noticed at Yerrinbool. The temperature recorded was below the threshold for an alarm to be sent to network control.

Safety issues and actions

Axle bearing installation

Safety issue number: RO-2020-017-SI-01

Safety issue description: The axle bearing installation process was not sufficient to ensure the tabs on the locking plate were installed correctly (Safety Issue).

Sources and submissions

Sources of information

The sources of information during the investigation included the:

• Australian Rail Track Corporation
• NSW Trains
• Sydney Trains
• United Group Limited Unipart.

References

Australian Standard AS 4291.1 / ISO 898.1 Mechanical properties of fasteners made of carbon steel and alloy steel.

MetallTech (2020), Endeavour Bolt Failure – Axlebox Bolt – Metallurgical Investigation, 23 November 2020

Rail Industry Safety and Standards Board (2021), Glossary of Terms.

Sydney Trains (2020), Systemic Safety Investigation Report Collapsed bearing on car 2811 at Yerrinbool on 13 October 2020.

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• Australian Rail Track Corporation
• NSW Trains
• Office of the National Rail Safety Regulator
• Sydney Trains
• Transport for NSW

Submissions were received from:

Office of the National Rail Safety Regulator

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations & publishing information

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2022 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

Safety summary

What happened

A privately operated Fairchild Industries SA226-T aircraft, registered VH-LDQ, landed at Gunnedah Airport, New South Wales, on 19 August 2020, where it was then parked overnight. The following day, at about 1230 Eastern Standard Time, the pilot commenced a take-off run on runway 29. As the aircraft accelerated, the pilot saw that there were two holes excavated from the runway pavement. The pilot attempted to avoid the holes, but the aircraft’s left main landing gear struck them. The aircraft sustained damage to the left main landing gear assembly, which resulted in it collapsing, and the left propeller striking the ground. The aircraft veered off the runway and came to rest outside the flight strip.

What the ATSB found

The ATSB found that during pre-flight planning, the pilot had not checked for relevant NOTAMs, including one stating that Gunnedah Airport was closed due to works in progress. Prior to entering the runway and commencing the take-off run, the pilot did not see evidence of the runway works.

A white cross had been placed at the main windsock, visible to aircraft arriving overhead, however, when the pilot had landed the previous afternoon the aerodrome was still open. Further, there was no works safety officer on site or any ground-visible unserviceability markings on the runway, as required by the Civil Aviation Safety Regulations Part 139 Manual of Standards (MOS) for Aerodromes. Aerodrome works staff were not aware of the updated MOS requirements that had come into effect 7 days earlier and had interpreted the superseded MOS to not require unserviceability markings if the whole aerodrome was closed.

What has been done as a result

The Gunnedah Airport operator has added folding signs to all entry gates to the airfield, which will display any total unserviceability or restricted operations signals that are currently being displayed  in the signals area adjacent to the primary windsock.

Additionally, the airport operator updated their contact details with CASA to ensure timely receipt of updates to requirements.

Safety message

An essential component of pre-flight planning is to check all NOTAMs relevant to the planned flight. This includes NOTAMs regarding all aviation facilities that a pilot plans to use.

To ensure receipt of correspondence that may affect safety of aircraft operations, aerodrome operators should ensure CASA is provided up to date contact details, particularly following changes to staff.

The investigation

The occurrence

A privately operated Fairchild Industries SA226-T aircraft, registered VH-LDQ departed Bankstown aerodrome, New South Wales (NSW) on 19 August 2020 at 0720 Eastern Standard Time (EST)^[1] for a private flight to Mudgee Airport, NSW.

The aircraft arrived in Mudgee at about 0900, before departing for Dubbo City Regional Airport, NSW, at about 1130. During this time, at 1047, a NOTAM^[2] was published, closing Gunnedah airport, NSW, on the following day, 20 August, between 0700 and 1500 for emergency runway works.

The pilot departed Dubbo at about 1515 and arrived at Gunnedah Airport, at about 1550. As the aircraft was taxied to the parking bay, the pilot noticed a rough spot on the asphalt near the end of the runway. The aircraft was then parked for the night.

On 20 August, at about 1225, the pilot advised Brisbane Centre air traffic control that the aircraft was taxiing at Gunnedah for departure. On receiving the taxi call, the controller checked for any current notices to airmen (NOTAM) and found one current for the airport, stating that it was closed due to works in progress. The controller attempted to inform the pilot but was unable to establish contact.

About 5 minutes later, the pilot commenced the take-off run on runway 29 after taxiing by the works in progress (Figure 3) to the threshold and turning around. The pilot reported seeing patches on the pavement in the distance. As the aircraft accelerated and progressed along the runway, it became apparent to the pilot that there were two holes excavated from the runway pavement (3 m wide by 5 m long), which were about 30 cm deep (Figure 1). The pilot attempted to avoid the holes but was unable to clear them with the aircraft’s left main landing gear.

Figure 1: Pavement work in progress on runway 29

Source: Airport operator

The aircraft sustained damage to the left main landing gear assembly that resulted in it collapsing and the left propeller striking the ground. The aircraft veered off the runway and came to rest outside the flight strip (Figure 2).

Figure 2: Aircraft’s position after the attempted take-off

Source: Airport operator

Context

Gunnedah airport

Gunnedah airport is a certified and non-controlled aerodrome located in northern NSW. It has one asphalt runway and one grass runway that intersect. It is primarily used by the Gunnedah aero club and Gunnedah Flight School. There were no regular public transport services.

Runway works

On 14 August 2020, following jet operations that had been conducted with a pavement concession,^[3] the aerodrome reporting officer (ARO) at Gunnedah Airport identified some lifted pavement in the centre of runway 29. The damage was located about 215 m along the runway from the runway 29 threshold, 15 m past the taxiway intersection (Figure 3).

Figure 3: Location of holes on runway in relation to taxiway

Source: Google Earth and airport operator, annotated by the ATSB

On 19 August, the asphalt crew advised the ARO that they would be able to conduct runway repairs on the following day. The ARO reported contacting Airservices Australia by phone, who advised that the airport could be closed at short notice (within 24 hours) with a NOTAM in place if emergency works were required. Additionally, the ARO reported that they confirmed their interpretation of the Civil Aviation Safety Regulations’ Manual of Standards (MOS) for Aerodromes, that unserviceability markings would not be required if the aerodrome was closed. A NOTAM was published at 1047 stating that the aerodrome was not available due to works in progress between 0700 and 1500 EST on 20 August.

At about 1200, the works crew conducting pavement work left the aerodrome for lunch, returning to find the damaged aircraft. After the incident, the NOTAM closing the aerodrome was extended for an additional 24 hours and was cancelled at 1457 on 21 August, following completion of the works.

Aerodrome closures

The MOS required aerodrome operators to ensure that aerodrome works do not create a hazard to aircraft or cause confusion to pilots. If an aerodrome is used for scheduled air transport operations or has emergency services aircraft based at the aerodrome, a method of working plan is required for aerodrome works, unless the aerodrome is closed with 14 days’ notice, or the works are of an emergency nature.

Works are defined to be of an emergency nature if they are to repair unforeseen damage to part of the manoeuvring area, or to remove an obstacle.

Aerodrome unserviceability markings

As defined by the MOS, unserviceability markings are temporary markings used for temporary and permanent closures of runways or taxiways. They consist of white or yellow crosses of various sizes displayed on the surface of a closed or unserviceable area for airborne and taxiing aircraft. When used to mark a runway as unserviceable, the MOS requires a white cross at each end of the runway and at intervals of less than 300 m.

These unserviceability markings are not required on an unserviceable runway at an aerodrome not controlled by an air traffic control provider if all the following exist:

• the runway unserviceability is of less than 24 hours duration
• the total works period is less than 5 days
• a NOTAM has been issued
• a works safety officer is present with a vehicle equipped with a radio that allows for emergency 2-way communication with aircraft
• if the aerodrome has been closed: a total unserviceability signal is displayed in the signal area (if it exists).

Unserviceability markers

As defined by the MOS, unserviceability markers are a 50 cm tall white cone with a 25 cm wide horizontal red stripe. These markers must be placed at the entrance to, and across, any part of the movement area of an aerodrome (including runways) that are not to be used by aircraft. Additionally, at least three must be displayed across the centreline of any portion of a taxiway, apron or holding bay that is unserviceable.

Aerodrome ground signals

The MOS current at the time of the occurrence, had provision for an optional ground signal area, where temporary ground signals for airborne aircraft can be displayed. The MOS noted that the area was ‘neither compulsory nor necessary because of the requirements for radio carriage and use at certified aerodromes, and the requirement for NOTAMs to be issued for the aerodrome in the event of changes in conditions.’

The aerodrome signal area must be black, 9 m in diameter and located near the primary wind direction indicator. It must have a 1 m-wide border or at least 5 spaced white markers. The markings that could be placed in the signals area included:

• total unserviceability signal – consisting of a white cross; must be displayed when an aerodrome is closed to landing aircraft.
• restricted operations signal – consisting of a white barbell; to indicate that aircraft are only to use sealed runways, taxiways, and aprons.

At the time of the occurrence, a total unserviceability signal was displayed in the signals area at Gunnedah airport.

Update to Manual of Standards (Aerodromes)

On 13 August 2020 (seven days before the incident), the Civil Aviation Safety Regulations Part 139 (Aerodromes) Manual of Standards (2019) came into force, replacing the previous 2014 MOS. This manual sets out the standards required of certified aerodromes.

A new section was added in Chapter 8 to articulate specific circumstances where unserviceability markings are not required, given that other visual aids or conditions are met. This included the addition of the requirement for a works safety officer with a vehicle and radio. The requirements for the use of unserviceability markers was expanded from situations where it was ‘possible for aircraft to bypass the area safely’, to require marking of all unserviceable movement areas.

Additionally, requirements for the notification and planning of aerodrome works and method of working plans were clarified and updated.

Communication of changes

The Civil Aviation Safety Authority (CASA) began engaging with certified aerodrome operators regarding upcoming changes to MOS 139 in April of 2020 by email and letter, with information provided on the CASA website. CASA sent this information to a mailing list that included an email address associated with Gunnedah Airport, for a person that was no longer working for the airport operator.

The Gunnedah Airport operator’s email system did not generate an ‘undeliverable’ bounce message for email addresses that have previously existed. CASA advised the ATSB that they rely on these bounced emails, or updates from aerodrome operators to ensure that information is received by a nominated person.

Flight Planning

Civil Aviation Regulation 239, required pilots to study all available information appropriate to the intended operation, including current weather reports and forecasts for the route and aerodromes, and the condition of aerodromes to be used.

Aerodrome condition information is typically communicated via a NOTAM. NOTAM and weather information can be obtained through the National Aeronautical Information Processing System Internet Service, a phone briefing, AVFAX or by radio.

The pilot had not checked NOTAM information for Gunnedah Airport before commencing take-off from runway 29.

Safety analysis

Communication of closure

The Gunnedah Airport operator had issued a NOTAM for the emergency runway works in accordance with requirements. The NOTAM issued stated that the aerodrome would be closed due to works in progress, was in place 20 hours prior to the closure and active at the time of the occurrence. Had the pilot checked the NOTAMs prior to flight, as required, they would have been alerted to the fact that the aerodrome was not available for use.

Evidence of works

As well as the active NOTAM at the time of the incident, Gunnedah airport had a total unserviceability signal displayed adjacent to the primary wind direction indicator, to warn pilots that the aerodrome was closed. However, this signal was not visible to a pilot on the manoeuvring surface and nor is it designed to be.

There were no other unserviceability markings or markers in use on this occasion. The Part 139 Manual of Standards (MOS) in force at the time required both, however the use of unserviceability markers on the runway would not have been required had a works safety officer been present. The aerodrome reporting officer was unaware of the introduction of the updated MOS and believed they had adhered to the correct standards. In addition, they interpreted the guidance received from Airservices as confirmation of their understanding of the requirements.

The use of unserviceability markers on the runway or a presence of the works safety officer together with the use of unserviceability markers would also have provided additional risk controls against the attempted take off from the closed runway.

Communication of changes to MOS

CASA relies on aerodrome operators ensuring that they have a valid, monitored email address registered for the distribution of timely updates on associated legislation and requirements. The registered contact for Gunnedah Airport had left the role and new contact details had not been provided to CASA, which mean that this mechanism for receiving updates was unavailable.

Findings

From the evidence available, the following findings are made with respect to the take-off from a closed runway at Gunnedah Airport, New South Wales, involving a Fairchild Industries SA226-T aircraft, registered VH-LDQ on 20 August 2020.

Contributing factors

• The pilot had not checked NOTAMs, which advised of the aerodrome closure, and did not see evidence of works.
• There was no works safety officer at the aerodrome, or ground visible unserviceability markers or markings on the runway, as required under the Manual of Standards for Aerodromes.
• Aerodrome works staff were not aware of the new Manual of Standards requirements and interpreted the previous version to not require unserviceability markers or markings during an aerodrome closure.

Safety actions

Safety action by Gunnedah Airport operator

Gunnedah Airport operator has added folding signs to all entry gates to the airfield that will reflect any ground signals active in the signals area (Figure 4). They have also updated their contact details with CASA.

Figure 4: New signals signage at Gunnedah airport

Source: Airport operator

Sources and submissions

Sources of information

The sources of information during the investigation included:

• Civil Aviation Safety Regulations Part 139 (Aerodromes) Manual of Standards 2019 (as amended 13 August 2020)
• Civil Aviation Regulations 1988 (as amended 13 August 2020)

Submissions

Under section 26 of the Transport Safety Investigation Act 2003, the ATSB may provide a draft report, on a confidential basis, to any person whom the ATSB considers appropriate. That section allows a person receiving a draft report to make submissions to the ATSB about the draft report.

A draft of this report was provided to the following directly involved parties:

• Gunnedah Airport operator
• the Civil Aviation Safety Authority
• Airservices Australia
• the pilot

Submissions were received from:

• the Civil Aviation Safety Authority

The submissions were reviewed and, where considered appropriate, the text of the report was amended accordingly.

Purpose of safety investigations The objective of a safety investigation is to enhance transport safety. This is done through: identifying safety issues and facilitating safety action to address those issues providing information about occurrences and their associated safety factors to facilitate learning within the transport industry. It is not a function of the ATSB to apportion blame or provide a means for determining liability. At the same time, an investigation report must include factual material of sufficient weight to support the analysis and findings. At all times the ATSB endeavours to balance the use of material that could imply adverse comment with the need to properly explain what happened, and why, in a fair and unbiased manner. The ATSB does not investigate for the purpose of taking administrative, regulatory or criminal action. Terminology An explanation of terminology used in ATSB investigation reports is available here. This includes terms such as occurrence, contributing factor, other factor that increased risk, and safety issue. Publishing information  Released in accordance with section 25 of the Transport Safety Investigation Act 2003 Published by: Australian Transport Safety Bureau © Commonwealth of Australia 2021 Ownership of intellectual property rights in this publication Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this report publication is owned by the Commonwealth of Australia. Creative Commons licence With the exception of the Coat of Arms, ATSB logo, and photos and graphics in which a third party holds copyright, this publication is licensed under a Creative Commons Attribution 3.0 Australia licence. Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided that you attribute the work. The ATSB’s preference is that you attribute this publication (and any material sourced from it) using the following wording: Source: Australian Transport Safety Bureau Copyright in material obtained from other agencies, private individuals or organisations, belongs to those agencies, individuals or organisations. Where you wish to use their material, you will need to contact them directly.

__________

1. Eastern Standard Time (EST) was Coordinated Universal Time (UTC) + 10 hours.
2. NOTAM: A notice distributed by means of telecommunication containing information concerning the establishment, condition or change in any aeronautical facility, service, procedure or hazard, the timely knowledge of which is essential to personnel concerned with flight operations.
3. A pavement concession is a privilege granted to an aeroplane operator, allowing the landing, take-off, taxiing or parking of an aeroplane on a runway, taxiway or apron which would not otherwise be available. It allows an aircraft heavier than the rated load capacity of the pavement to use the pavement under specific conditions.