Jump to Content

At about 1200 Eastern Standard Time on 30 May 2005, a Boeing Co 747-300, registered JA8184, was being pushed back from its gate at Sydney International Airport for a scheduled passenger flight to Osaka, Japan. During pushback, the ground staff heard a loud cracking noise. The pushback was stopped and an inspection by the ground crew identified a structural failure in the left wing landing gear forward trunnion fork.

Examination of the trunnion fork revealed that it had failed due to fatigue cracking that had originated on the inner surface of the trunnion fork bore. It was found that the wall thickness at the crack origin was below the minimum allowed by the design and that the inner surface of the bore did not meet the specifications of the design. These factors contributed to the formation and development of the fatigue crack, which lead to the final failure on pushback.

The trunnion fork had amassed a total of 25,095 landing cycles and had been overhauled by the operator on four occasions. During the overhaul the item was inspected for cracks and on each occasion the item was passed. The inspection procedure was general for the item and did not specifically indicate that the area where the cracking originated required particular attention. The surface finish of the inner surface of the bore may have masked indications of any cracks that may have been present.

As a result of this occurrence, the aircraft manufacturer and the aircraft operator have commenced actions to determine the extent of the problem in the remaining fleet and improvements in the inspection of items during maintenance.


Sequence of events

At about 1200 Eastern Standard Time on 30 May 2005, a Boeing Co 747-300 (747), registered JA8184, was being pushed back from its gate at Sydney International Airport for a scheduled passenger flight to Osaka, Japan. During pushback, the ground staff heard a loud cracking noise. The pushback was stopped and an inspection by the ground crew identified a structural failure in the left wing landing gear forward trunnion fork (trunnion), as shown in Figure 1 and 2. After an on-site inspection by the Australian Transport Safety Bureau (ATSB), the aircraft was moved to a hangar for maintenance and the fractured component was removed from the aircraft and sent to the ATSB for a detailed examination.

Figure 1 : Left wing landing gear

Figure 1

Figure 2: Looking up and outboard into wing landing gear well

Figure 2

Examination of fractured trunnion

The trunnion had sustained a complete through-section fracture, located approximately mid-way between the ball-end and the fork-end (Figure 3).

Figure 3 : Fracture location on trunnion


Figure 3

A general inspection of the fractured trunnion revealed a discoloured (orange/brown) region on the fracture surface (Figure 4) in the upper outboard region. The corroded nature of that region, compared with the adjacent bright fracture surfaces, indicated the presence of a pre-existing defect, and that the trunnion had been cracked for a period of time prior to the final failure during the pushback.

Figure 4 : Fracture surface

Figure 4

The fractured component was examined in a metallurgical laboratory under the supervision of the ATSB. Chemical analysis of a sample taken from the trunnion near the fracture showed that the material met the specification for AISI/SAE 4340M alloy steel. Metallographic examination confirmed a fine-grained lightly tempered martensitic microstructure, typical of the 4340M alloy in the hardened and tempered condition. The inner and outer surfaces were also observed to have been finished with a metallic type plating and painted with a surface primer and top coat.

Hardness measurements taken indicated that the material had an ultimate tensile strength of approximately 275,600 psi (1900 MPa).

Wall thickness measurements taken around the fractured trunnion circumference showed that the minimum local wall thickness of 0.137 inches (3.48 mm) corresponded with the corroded and discoloured area of cracking.

A detailed technical examination of the corroded region identified two transverse fatigue cracks originating at the inner surface of the trunnion bore (noted as C1 and C2 in Figure 5). The cracks had initiated approximately 11mm apart and had joined to form a single crack. This crack continued to grow until the final fracture occurred during the pushback.

Figure 5 : Fatigue crack development

Figure 5

Crack C2 presented well defined fatigue progression marks, including several distinct regions of fatigue and corrosion (Figure 6). The region bounded by the red dotted line was a distinct region of heavily corroded fatigue cracking and was about 4 mm long and 1.6 mm deep.

Figure 6 : Fatigue progression and corrosion marks

Figure 6

Close examination of the origins in the inner wall of the trunnion found that crack C1 had initiated from multiple closely spaced origins at the root of a machining groove, giving the appearance of a longer crack following the machining groove (or mark). Crack C2 had also originated at the root of a machining groove from multiple closely spaced origins, but over a much smaller distance before aligning with the principal stress plan1, resulting in the apparent difference in the planes of the cracks as shown in Figure 7. The plane of the initial cracks in both C1 and C2 were approximately parallel and were aligned with the machining grooves in the inner surface.

Figure 7 : Plane of crack origins

Figure 7

The inner and outer surfaces did not have a consistent surface roughness. Well defined machining marks were observed in the large diameter bore and to a lesser extent on the small bore. However, the outer surface and the taper region in the bore were relatively smooth without defined machining marks. The well defined machining marks on the large and small diameter bore blended into the smoother surface of the taper region (that is, there was no abrupt change in surface roughness).

Examination of the surface microstructure in the region of the cracks revealed that the smooth regions (outer surface and taper section) had a thin layer of deformed material typical of a cold working process such as shot peening2. The area of surface deformation in the taper region ran out just before the radius (a few millimetres from the cracks). Figure 8 shows the differences in the surface roughness at a microscopic level (scale is 25µm, or 0.025 mm).

Figure 8 : Smooth surface (upper); surface with distinct machining marks (lower)

Figure 8

The sections taken from adjacent to the primary cracks for micrographic examination contained multiple independent fatigue cracks of various sizes. One example is shown in Figure 9. Each of these cracks originated in the root of the machining groove and were associated with shallow intergranular penetrations, which also existed in the roots of the machining grooves (Figure 8).

Figure 9 : Secondary fatigue crack indicated by arrow

Figure 9

Component manufacture

The landing gear trunnion was manufactured to the aircraft manufacturer's specifications by an approved external supplier. Both the supplier and the aircraft manufacturer informed the ATSB that the trunnion was manufactured at some time prior to 1975; however the original manufacture documentation (including the manufacture plan and conformance records) was destroyed in 1994. The trunnion specifications were supplied by the aircraft manufacturer. Those documents included construction drawings and process specifications.

In the trunnion specifications, the aircraft manufacturer specified the use of 4340M steel, heat treated to an ultimate tensile strength of 275,000 to 300,000 psi. Therefore, the material used in the manufacture of the failed component met the steel alloy and strength requirements of the design.

The minimum allowable wall thickness specified3 for the trunnion at the location of the failure was 0.180 inches (4.57 mm). Therefore, the minimum wall thickness measured at the crack of 0.137 inches (3.48 mm) was 0.043 inches (1.09 mm) thinner than the design allowed.

Component maintenance

The maintenance documents supplied by the aircraft operator indicated that the trunnion had been fitted to five aircraft and had amassed a total of 25,095 landing cycles during its service life. The records also showed that the trunnion had been overhauled by the operator's component repair workshop on four occasions (Table 1). The landing gear assembly had an overhaul interval of 8 years or 12,000 cycles, whichever occurred first. Therefore, the landing gear was not due for overhaul for another 4 years, or 9,509 cycles.

Table 1 : Overhaul history

Date Total cycles at overhaul


October 1979



November 1987



March 1992



October 2001


The overhaul records showed that on each occasion, the trunnion had undergone repair work. The documents indicated that the repairs were limited to the lugs and were within the repair limits permissible by the aircraft manufacturer. There was no record of any repair work carried out in the bore of the trunnion or on the outer surface in the region of the bore diameter transition.

Comparison of the overhaul instructions maintained by the operator with those supplied by the aircraft manufacturer confirmed that the operator's workshop maintained the correct instructions for the overhaul. The overhaul instructions for the component did not require a dimensional check for wall thickness or a specific check for surface finish (roughness) in the bore.

As one of the first processes in the overhaul, the protective finishes (including metallic plating) were removed from the surface of the trunnion. These finishes are removed using a chemical process, some of which, including water for cleaning, can have a corrosive effect on the trunnion material.

The overhaul procedure for the landing gear components required that the component undergo a magnetic particle inspection (MPI) to detect any defects, including cracks, that may have developed during service. The manufacturer did not provide specific instructions on how to perform the MPI on this particular component, but provided a general process that the operator was to use as the basis for a component specific process. Neither the overhaul procedure nor the MPI process directed the MPI technician's attention to the radius in the bore diameter transition as a possible location for cracking. The MPI process used was capable of highlighting cracks of less than 0.5mm.

The overhaul records provided by the operator indicated that an MPI of the component was carried out at each overhaul. The MPI procedure defined by the operator was in accordance with the aircraft manufacturer's recommended procedure. It used the equipment and materials recommended by the manufacturer and specified sufficient examinations to highlight any cracks in the component. The overhaul records indicated that the item had been found satisfactory on each occasion, suggesting that no cracks had been detected.

The component maintenance manual for the repair of high-strength steel landing gear parts directed the operator to obtain advice from the manufacturer if cracks were detected during the MPI. The manufacturer did not have a record of any request for advice relating to the failed component.

The crack was in a location that was not readily viewable during a normal visual ground check. There was no requirement to perform a detailed inspection for cracks in the body of the trunnion between overhauls.

  1. The principal stress plane is a plane that the stress in the part acts perpendicular to. In this case, the principal stress plane was not aligned with the machining marks.
  2. Shot peening is a process where small 'shot' beads are fired against the surface of a component producing a residual compressive surface stress and a thin layer of deformed material. This process has been demonstrated to increase the fatigue life of high-strength steel components.
  3. In the manufacture drawings for the component.

The left wing landing gear forward trunnion sustained a complete through-section fracture during the pushback at Sydney International Airport as a result of fatigue cracks in the bore of the trunnion. The fatigue cracks originated at an internal bore diameter transition and developed until they intersected to form a single crack.

The development and growth of the fatigue cracks was attributed to three principal factors:

  • the wall thickness of the trunnion was below the minimum required by the manufacture specifications
  • the surface had machining marks in the surface at the radius
  • the inner surface of the bore had been inadequately shot peened.

The effect of the reduction in wall thickness was to increase the working stress in the component. This increase in working stress reduced the number of cycles required to produce and develop fatigue damage.

The radius at the transition in the trunnion bore diameter is a natural stress concentration point when the item is smooth, but the presence of the machining marks on the surface of this radius provided further stress concentration. This stress concentration further reduced the number of cycles required to produce and develop fatigue damage.

The lack of adequate shot peening likely had a two-fold detrimental effect on the trunnion fatigue life. Firstly, as the smooth regions in the bore showed, effective shot peening obliterated the machining marks. Those marks remained in the unpeened areas and thus presented an additional stress concentration. Secondly, the absence of adequate shot peening denied the component the fatigue life improving qualities that shot peening brings.

Because there were no entries in the maintenance documents regarding repairs in the internal bore and the blending of the shot peened and non-shot peened areas, it is likely that the trunnion wall thickness was below the minimum design limit and was inadequately shot peened during original manufacture.

The presence of multiple secondary fatigue cracks in the component, also emanating from the root of machining marks, further verified that the failure was not due to a single material defect. As such, it would be likely to occur in other trunnions, which do not have the machining marks obliterated by the shot peening process.

The varying nature of the corrosion within the fatigue cracks and the demarcation between the various regions suggested that the cracks had existed during several overhaul cycles of the component. During overhaul, the component was subjected to chemicals that had a corrosive effect on the material, but would not be readily flushed away from a tight crack. Therefore, it is likely that the crack was present in the component at the last overhaul. The fracture surface indicates that the crack was approximately 4mm long and 1.6mm deep at the last overhaul in 2001.

The component had undergone the manufacturer required inspections at overhaul and no cracks were detected. The Magnetic Particle Inspection (MPI) method used to check the item for defects such as cracks is sensitive enough to detect a crack much smaller than the one suspected to have existed at the last overhaul. Possible masking of crack indications by the machining marks or a lack of expectation by the operator to find cracks in the region may have contributed to any cracks not being detected by the MPI operator.

The machining marks in the surface of the part can give non-relevant indications4 of cracks. Those spurious indications may mask true indications of cracks. If the operator was not aware that the machining marks should not be present, they would be likely to discount them and pass the component.

The aircraft manufacturer provided standard practices in relation to the inspection method used. These practices were general and were to be used by maintainers in developing their component specific procedures. Neither the overhaul procedure for the trunnion nor the general MPI process specification directed the MPI operator's attention to the radius in the bore diameter transition. Therefore, the expectation for an operator's repair shop to find cracks in that region would be low.

Because the manufacture documentation for the particular component was destroyed in 1994, the investigation could not determine how the trunnion was manufactured and released in a state that did not conform to the manufacture drawings. The overhaul and service instructions for the trunnion did not provide a mechanism by which the non-conformances could be detected.

  1. Non-relevant indications are indications that are defect-like in appearance, but are due to the local geometry and features of the component. Heavy machining marks are one such
  1. The wing landing gear trunnion did not conform to the design specifications. The component's wall thickness in the region of the failure was less than the allowable minimum and the internal bore had not been adequately shot peened.
  2. Several fatigue cracks developed in the inner bore at the bore transition region.
  3. The fatigue cracks were likely present during the last overhaul, but were not detected during the magnetic particle inspection.
  4. The fatigue cracks developed until the loads during the pushback operation exceeded the residual strength of the component, leading to failure of the trunnion.

Aircraft manufacturer

The aircraft manufacturer has informed the Australian Transport Safety Bureau that they have instigated an internal investigation and intend to publish a Service Bulletin to address the problem. The Service Bulletin will contain "on-wing" inspections and inspection and refinish actions during heavy maintenance.

Aircraft operator

The aircraft operator has informed the Australian Transport Safety Bureau that they have introduced a series of additional one-time and repetitive inspections of the landing gear trunnions on their 747 fleet. These additional inspections are:

  1. A one-time detailed visual inspection of the trunnion at the first maintenance opportunity following the failure of the component on this aircraft.
  2. Repetitive detailed visual inspections of the trunnion either before every international flight, or during the daily check for domestic flights.
  3. A one-time inspection at the first appropriate maintenance opportunity of the following:
    • Detailed visual inspection of the internal surface of the trunnion bore by borescope.
    • Eddy current inspection of the external surface of the trunnion.
    • Eddy current inspection of the internal surface of the trunnion bore.
    • Ultrasonic inspection from the external surface of the trunnion, around the area in which the fracture originated, to measure the wall thickness and determine if it is less than the 0.180 inches allowable minimum.

Completion of these inspections without detection of a fault cancelled the requirement for the repetitive inspection detailed in item 2.

  1. Repetitive inspections at every 1C Check of the following:
    • Detailed visual inspection of the external surface of the trunnion.
    • Detailed visual inspection of the internal surface of the trunnion bore by borescope.
    • Eddy current inspection of the external surface of the trunnion.
    • Eddy current inspection of the internal surface of the trunnion bore.
General details
Date: 30 May 2005 Investigation status: Completed 
Time: 1200 hours EST Investigation type: Occurrence Investigation 
Location   (show map):Sydney, Aero. Occurrence type:Landing gear/indication 
State: New South Wales Occurrence class: Technical 
Release date: 01 June 2006 Occurrence category: Serious Incident 
Report status: Final Highest injury level: None 
Aircraft details
Aircraft manufacturer: The Boeing Company 
Aircraft model: 747 
Aircraft registration: JA 8184 
Serial number: 23968 
Type of operation: Air Transport High Capacity 
Damage to aircraft: Substantial 
Departure point:Sydney, NSW
Departure time:1200 hours EST
Destination:Osaka, Japan
Share this page Provide feedback on this investigation
Last update 16 February 2016