Engine information
The aircraft was powered by four Honeywell International Inc (Honeywell) LF507-1F (LF507) dual-spool, high-bypass turbofan engines. For maintenance and serviceability, the LF507 engine is divided into four modules: the fan assembly, gas producer, combustor turbine and accessory gearbox. This occurrence related to the combustor section of the combustor turbine module (CTM) (Figure 2).
The CTM consists of the combustor and the low-pressure turbine that are retained within the combustor housing (housing). The housing has four welded bosses located at the two, four, eight and ten o’clock positions (looking forward from the rear of the engine). A spigot pin is screwed into each welded boss to locate and retain the combustion liner. During normal operation, the fuel/air mixture is ignited in the combustion liner and supplies energy to the low-pressure turbine.
Figure 2: Left-side cut-out view of an exemplar LF507 engine showing the locations of the combustor housing, combustion liner and combustor turbine module
Source: Honeywell, modified by the ATSB
VH-NJI (NJI) was introduced into Australian service in 2012 with engine serial number LF07405, CTM serial number 93K004 and the occurrence combustor housing (serial number 363) fitted. The engine and these components were subsequently operated in the No. 2 engine position on NJI until the occurrence in April 2014. In the intervening period, the combustor housing accumulated 1,319 engine cycles in service since its last heavy maintenance inspection.
Engine examination
Examination of the engine found that the combustion liner welded boss (welded boss) at the two o’clock position had fractured and separated from the housing (Figure 3).
Figure 3: Exemplar LF507 engine showing the combustor housing and combustion liner highlighting a typical welded boss and combustion liner locating pin
Source: Honeywell, modified by the ATSB
A portion of the combustion liner, a separate component that normally contains the combusting fuel/air mixture gases, fractured adjacent to the 2 o’clock welded boss and sustained significant fire damage (Figure 4). The breached combustor housing and combustion liner created a radial escape path for the high-pressure combusting fuel, which quickly burnt through the engine cowling support structure and engine cowling in that location.
A review of the Engine Condition Trend Monitoring[3] data for the engine was conducted by the ATSB and Cobham Aviation Services (Australia) (Cobham) several days after the occurrence. The review confirmed that there were no air leakage indications or parameter shifts prior to the occurrence that may have alerted maintenance staff of a pending defect or failure.
The engine was removed from the aircraft and dispatched to an engine overhaul facility in the United Kingdom (UK) for technical examination. This disassembly and inspection was carried out under the supervision of the UK Air Accidents Investigation Branch, with assistance from Honeywell. Components that were relevant to the investigation, including the combustor housing, combustion liner and the four combustion liner locating pins, were dispatched to the United States (US) for detailed examination.
Detailed examination of the housing, combustion liner and the four locating pins was conducted by Honeywell under the supervision of the US National Transportation Safety Board. The fractured welded boss, which was subsequently found in the engine cowling and recovered by Cobham, was initially examined by the ATSB before also being sent to Honeywell in the US for further detailed examination.
Figure 4: No. 2 engine showing the combustor housing, combustion liner (seen internal to the combustor housing) and insert image showing the locating pin secured in the fractured welded boss
Source: ATSB
Results of the detailed component examination in the US
The detailed examination in the US identified an area of the combustion liner that had thermal damage and was fractured adjacent to the two o’clock welded boss position. That damage was consistent with the effect of temperature and pressure variations between the combustor housing and the combustion liner once the welded boss fractured from the housing (Figure 4).
Honeywell conducted a detailed metallurgical examination of the combustor housing and the recovered welded boss. It was determined that the sheet metal of the combustor housing fractured as a result of fatigue cracking adjacent to the boss weld line (Figure 5). In addition, weld repairs were identified on cross-sectional samples taken through the joints surrounding the recovered welded boss.
Honeywell advised that the fatigue fracture was consistent with low cycle fatigue and did not appear to be associated with the weld repairs.
A non-penetrating crack, a crack that did not penetrate through the full thickness of the housing sheet metal, had propagated over a period of about 2,680 load cycles. Each load cycle on the affected components was considered by Honeywell to be associated with pressure variations within the engine. Pressure variations can be caused by engine power level changes, compressor stalls, combustion rumble and temperature variations that lead to expansion and contraction of the housing. Numerous load cycles can occur during normal flight.
Engine maintenance is based on engine hours, cycles[4] or calendar days, depending on the type of component. Engine maintenance is not based on load cycles as each flight can produce a significant variation due to conditions at that time.
The examination of the fractured welded boss identified three separate weld repairs. These were identified on cross-sectional samples taken through the joints surrounding the recovered welded boss (Figure 5). Honeywell concluded that the fatigue cracking was not directly associated with the weld. Hardness measurements taken from the housing sheet metal, boss casting, and weld indicated that heat treatment was performed on the assembly subsequent to weld repair, as specified by Honeywell.
Figure 5: Metallographic image of a cross section of the fractured welded boss. The red dots likely indicate the original weld. The white, yellow and black dots likely indicate subsequent weld repairs. The red arrows indicate a crack emanating from the forward side of the weld at the housing/casting interface. The white arrow is the fracture surface, where the boss separated from the combustor housing. The fracture initiated at the outside diameter of the welded boss
Source: Honeywell
Further examination of the combustor housing in the location of the fractured welded boss found that a section of the combustor housing sheet metal had been thinned at the inside and outside surfaces by grinding (Figure 6). This grinding was associated with a weld repair where the boss was weld fused to the sheet metal. Thicknesses of 0.035 inches–0.040 inches were measured at the CTM housing adjacent to the two o’clock welded boss.
The thickness of the combustor housing sheet metal was specified by Honeywell as nominally 0.049–0.050 inches thick. This indicated that during a repair, about 20–30 per cent of the housing thickness had been removed. In this regard, the combustor housing is a pressure vessel. Honeywell does not approve the removal of material leading to a reduction of housing wall thickness.
Removal of material, as in this case, created an area of increased local stress.
The other three welded bosses also exhibited hand finishing, indicating that material removal took place in those locations.
Figure 6: Image of the inner side of the fractured combustor housing and the locating pin protruding through the welded boss. Areas that have been ground are identified by yellow arrows
Source: Honeywell, modified by the ATSB
Repair of ignition or liner welded bosses
During normal engine operation, expansion and contraction of the combustor housing can contribute to the welds at the bosses fatiguing and cracking. Such cracking is predictable and can be monitored during scheduled maintenance.
Honeywell published a system of maintenance where cracks, under a specific criteria, could be monitored during normal operation or repaired during heavy maintenance. Repairs depended on the length of the crack and the measured growth rate. The repair consisted of grinding the cracked boss weld, chemically cleaning the component and then re-welding the area.
In respect of the repair to the No. 2 engine in NJI, the grinding of the combustor housing sheet metal significantly reduced the housing thickness. Such grinding was not part of the Honeywellapproved repair and crack repairs that extended into the sheet metal housing were not approved. Cracks extending into the combustor housing necessitated replacement of the housing.
The repair to the combustor housing of the No. 2 engine in NJI was an unapproved repair.
Blending of minor blemishes, scratches and nicks
Honeywell advised that the polishing and blending of minor blemishes, scratches and nicks was permitted. However, the use of power tools, or the reduction of housing thickness, was not permitted under the approved repair scheme.
In respect of blend repairs, the Honeywell maintenance manual, section 70-25-01, p.1 of 31 March 2006 stipulated:
- Blend repair such defects as follows:
- Repair using small diesinker type file and india or carborundum stone. Use crocus cloth (05-07, 70-80-01) or rubberised abrasive block (ST-20-ALO-88X, 06-09, 70-80-01) for final polishing.
- Blend all repairs and finish smoothly. Lines, scratches, or sharp edges that might cause concentration of stress are not permitted.
At the time of the occurrence, Honeywell had not published any combustor housing wall thickness limits in respect of the amount of material that could be removed during blending. Maintenance personnel performing blending repairs, in the absence of manufacturer’s limitations, were required to make their own judgement as to the amount of material that could be removed.
The reduction in housing wall thickness in the No. 2 engine in NJI was most likely a product of a nonstandard weld repair, as opposed to a blend of a blemish, nick or scratch. In any case, the housing at the welded boss had been ground, removing 20–30 per cent of the material thickness.
Engine maintenance schedule
LF507 engine modules can be separated and moved between engines. Engines are often swapped from engine position, aircraft-to-aircraft or to spare. Numerous components on a turbine engine have operational limits measured in hours, cycles and/or calendar days. Modules are frequently removed earlier than the manufacturer stipulates, often as a result of:
- foreign object damage
- to accommodate other required maintenance
- preventive maintenance that may be deemed more economical to perform at that time.
Prior to the occurrence, the CTM was removed from and reinstalled in a number of engines on six separate occasions, including the No. 2 engine in NJI. This included for:
- 6,000-cycle Hot Section Inspections
- non-scheduled shop visit maintenance and modifications that necessitated significant disassembly, maintenance and detailed inspections.
That maintenance activity took place in 1999, 2001, 2005, 2006, 2008, and 2009.
Heavy maintenance was completed on the CTM in 2005 by a European workshop. With assistance from Honeywell, archived records for that shop visit were recovered and examined. The records, dated January 2005, showed that the previously-installed CTM housing, serial number 395, was removed and replaced with the occurrence housing. In the absence of additional data, it is likely that this housing was an exchanged item.
The serviceability of the CTM housing is based on its condition meeting Honeywell’s service limitations. It does not have a critical life limit based on hours, cycles or calendar days. Therefore the housing’s hours, cycles and/or calendar days were not tracked, nor were they required to be. The release certificate for the occurrence CTM housing when fitted in 2005 stated that the housing was:
Inspected and repaired in accordance with the LF507-1F engine manual section 72-41-03, repairs SP R401, 08, 09, 10, 12 (repair 12 repaired in accordance with the ALF502R engine manual 724104).
According to Honeywell, repair 12 has a different application depending on the model of engine being repaired. In respect of Honeywell engines an:
- ALF502 engine repair 12 is a diagnostic plate weld repair
- LF507 engine repair 12 is a repair of cracks in the weld of the ignitor boss or a repair of cracks in the weld of the liner retention welded boss.
In this occurrence, the CTM housing was fitted to an LF507 engine and, according to the release to service document, was repaired in accordance with the ALF502 engine manual. According to Honeywell, at that time the ALF502 engine manual was the appropriate document with which to repair that model housing.
The hours, cycles, or previous work conducted on the occurrence CTM housing could not be established prior to its installation in 2005. The European maintenance organisation that conducted the heavy maintenance on the CTM in 2006, 2008 and 2009 was no longer operating. Records recovered from that period showed that visual inspections were conducted on the CTM housing during that maintenance with no repairs to the housing recorded. The last heavy maintenance inspection and disassembly of the CTM was in 2009, about 3,438 engine cycles prior to the occurrence.
In addition to the scheduled heavy maintenance inspections of the engine modules, the aircraft maintenance system called for an ‘on-wing’ general visual inspection of the engine(s), including the CTM combustor housing, every 500 flight cycles or 6 months. The intent of that inspection was to identify abnormalities associated with the engine’s exterior surfaces. The inspection allocated 2 hours to inspect the front, centre and rear sections of engine Nos. 1, 2, 3 and 4. This included the housing, fittings, plumbing and associated accessories without removal of components.
Honeywell defined a visual inspection as:
An examination of an interior or exterior area, installation or assembly to detect obvious damage, failure or irregularity. This level of inspection is made from within touching distance unless otherwise specified. A mirror may be necessary to enhance visual access to all exposed surfaces in the inspection area. This level of inspection is made under normally available lighting conditions such as daylight, hanger lighting, flashlight or droplight and may require removal or opening of access panels or doors. Stands, ladders or platforms may be required to gain proximity to the area being checked.
No special or detailed inspections were stipulated by Honeywell or the Civil Aviation Safety Authority for the combustor housing when on-wing. No cracks in the combustor housing were recorded by Cobham as a result of its two on-wing inspections of the engine during its service with Cobham. However, non-penetrating cracks that do not show gas leakage, fretting or discolouration are difficult to identify using visual inspection techniques.
Commonly-used documents such as US Federal Aviation Administration Advisory Circular AC43.13 described the equipment and techniques that, when used, may detect cracks when performing nondestructive testing through visual inspection. According to AC43.13, the key to performing a visual inspection is to direct a suitable torch beam, at a 5°–45° angle to the inspection surface, and direct the beam towards the face. Cracks are identified as a shadow or reflected light beam. Use of a 10 times magnifying glass can confirm the existence of a suspected crack. If this is assessed as inadequate, use of other non-destructive testing techniques, such as penetrant or eddy current inspection, can be performed to verify cracks.
Previous similar occurrences
A search of historic records by Honeywell did not identify the CTM housing as having a high failure rate. In that respect, one other CTM housing welded boss fracture was recorded, 10 years prior to this occurrence.
A major United Kingdom engine overhaul facility for the LF507 engine reported that about 40 per cent of the LF507 engines presented to their facility for maintenance were cracked at other than CTM housing welded boss locations. Those cracks were reported repaired in accordance with the Honeywell maintenance instructions and did not develop into a fracture.
Subsequent occurrence
During finalisation of this investigation report, on 10 March 2016 the ATSB was notified that a Swiss Global Air Lines Avro 146-RJ100 aircraft, which was powered by LF5071F engines, had sustained a No. 2 engine fire during take-off. The aircraft, registered HBIYT, was on departure from Zurich Airport, Switzerland when the take-off was rejected due to sparks observed from the engine. The Swiss Transport Safety Investigation Board (STSB) is responsible for investigating this occurrence.
The STSB is responsible for the release of the final investigation report into the occurrence involving HB-IYT. Any enquiries in respect of the ongoing STSB investigation, or release of their investigation report should, in the first instance, be directed to the:
Swiss Transportation Safety Investigation Board
Aviation division
Aéropôle 1
CH-1530 Payerne
Email: info@sust.admin.ch
__________
In-flight engine fire
At a high engine power setting during the climb, the welded boss at the two o’clock position of the No. 2 Honeywell International Inc (Honeywell) LF507-1F engine fractured and separated from the combustor turbine module (CTM) combustor housing (housing). This led to the fracture of the combustion lining and allowed high-pressure combusting fuel and gases to escape radially from the engine. The engine cowling was weakened and melted from the resulting in-flight engine fire.
The engine fire detection and suppression system was effective in alerting the crew to the situation. The crew extinguished the fire using normal operating procedures and returned the aircraft to Perth Airport for landing.
Non-approved repairs
Metallurgical examination of the fractured welded boss found that it had been weld-repaired on three separate occasions. The only recorded weld repair to the CTM housing was in 2005 although, as the CTM housing was an exchanged unit, its history could not be established. Therefore, either the CTM housing was repaired at or prior to 2005, or the housing was repaired after 2005 and the repair was not recorded. From the evidence available, the ATSB could not determine which was the case. In any event, the housing had been ground adjacent to the welded boss, reducing the housing wall thickness by 20–30 per cent. Grinding of the housing was not in accordance with Honeywell’s approved repair scheme, and the non-approved repair was not identified in any of the subsequent heavy maintenance inspections.
The reduction in the housing wall thickness increased the operational stresses at that location. This would have affected the initiation of the fatigue crack and increased its rate of propagation during normal engine operation.
Limitations in the manufacturer’s blending process
The standard practices section of the Honeywell maintenance manual referred to blending as a means of reducing induced metal stress by removing scratches or nicks in the metal. That reference did not limit the amount of material able to be removed during the blending process. According to Honeywell, it was not the intent of the process to remove metal or reduce a combustor housing’s wall thickness when blending. In the absence of any limitation, the amount of material that could be removed during the process was open to interpretation by maintenance personnel.
It was possible for the blending process to be applied during a weld repair, where the repairer believed that, despite it not being part of the repair, it was necessary to remove stress raisers. However, this was inconsistent with the grinding evident on the fractured components. That grinding was likely achieved using power tools, as opposed to the Honeywell-defined hand blending process.
Scheduled maintenance inspections
The rate of cracking around the welded boss was reported to occur in about 40 per cent of the engines introduced into one of the approved engine repair facilities. However, Honeywell was only aware of one other event where the cracking had progressed to catastrophic failure. This indicated that this occurrence, where the fatigue crack developed into a fracture, was very rare. Honeywell considered that, when repaired in accordance with the current repair scheme, the repair specifications were generally adequate. The associated visual inspections were historically effective in detecting cracks around the welded boss prior to catastrophic failure.
The reduction in material thickness and corresponding increase in local stresses may have increased the rate of crack initiation and propagation. The increased cracking rate and resulting stresses meant that a crack might initiate, and the CTM housing ultimately fail, between scheduled inspections. Alternatively, the crack may have existed but remained undetected, or not existed at the previous inspection.
In terms of Cobham Aviation Services Australia’s conduct of the most recent 500hourly on-wing engine inspection, it was reasonable that, if the crack existed at that time, it was not visuallyidentified as:
- the crack was non-penetrating, meaning that it would probably have only been identifiable when the CTM was fully-disassembled (such as at a scheduled heavy maintenance inspection)
- there were no other, more usual, indications of a crack in the CTM, such as:
- increases in the engine condition trend monitoring data parameters
- blacking around any crack edges due to combusted gas leakage or fretting.
In addition to the lack of visual clues, the maintenance personnel were probably not expecting to find a crack. Human attention is guided by two factors:
- expectancy, where an individual will look where they expect to find information
- relevance, where an individual will look to information sources relevant to the important tasks and goals they need to carry out.
The key factor is expectancy. It is well demonstrated that people are more likely to detect targets when expected, and less likely to detect targets that are not expected (Wickens and McCarley, 2008). This occurs even when the targets are salient, potentially important and in an area to which the person is looking (Chabris and Simon, 2010).
