Aircraft Gearbox Vibration Failure

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

AAIB Bulletin No: 7/2005 Ref: EW/C2004/01/02 Category: 1.

INCIDENT

Aircraft Type and Registration: Jetstream 3202, G-BYRA


No & Type of Engines: 2 Garrett TPE331 turboprop engines
Year of Manufacture: 1989

Date & Time (UTC): 10 January 2004 at 1930 hrs


Location: Near Farnborough, Hampshire
Type of Flight: Public Transport (Passenger)

Persons on Board: Crew - 3 Passengers - 17


Injuries: Crew - None Passengers - N/A
Nature of Damage: Mechanical failure of the right engine propeller
reduction gearbox

Commander's Licence: Airline Transport Pilot's Licence


Commander's Age: 64 years

Commander's Flying Experience: 30,000 hours (of which 2,000 were on type)
Last 90 days - 97 hours
Last 28 days - 50 hours
Information Source: AAIB Field Investigation

Synopsis

The incident occurred during a charter flight from Southampton Airport to Manchester Airport. The
aircraft was passing FL120 in the climb when there was a loud bang from the area of the right
engine. High levels of vibration did not permit reading of the engine instruments and control of the
aircraft was difficult due to unexpectedly strong tendencies to yaw and roll. The right engine was
identified as not producing power and an emergency checklist shutdown carried out. After the right
propeller feathered a single-engine diversion to Farnborough was safely accomplished.

History of the flight

The aircraft was carrying out a charter flight to take a private party of 17 passengers from
Manchester Airport to Southampton Airport. The party was then to return to Manchester that
evening on the same aircraft.

19
In order to undertake the charter the aircraft was positioned that morning from Leeds Bradford
Airport by a flight crew of two pilots. The aircraft was not fitted with an autopilot (AP) but it had a
serviceable yaw damper; the only unserviceable item was the left engine Single Red Line Computer
(SRLC). The company procedure was for the co-pilot to carry out the pre-flight inspection whilst the
aircraft commander completed the internal cabin and cockpit checks.

Having completed their preparations, the aircraft departed Leeds Bradford and, following an
uneventful flight, arrived at Manchester. The flight crew met the cabin attendant who was to
accompany them for the next two sectors on which the passengers were to be carried and, following
a briefing, crew and passengers boarded the aircraft. Departure from Manchester was normal and an
uneventful transit to Southampton was made at FL170 in good weather arriving at 1241 hrs.

Whilst the passengers attended their function, the crew rested at the airport. Prior to the scheduled
departure time of 1845 hrs, they prepared the aircraft for the return flight to Manchester. The crew
adopted the same procedure as before with the co-pilot carrying out the pre-flight inspection, which
included checking the engine oil levels. The passengers were delayed arriving at the aircraft but as
soon as they were boarded, the aircraft's engines were started. Both engines started normally and the
aircraft was taxied for Runway 20. With the cabin secure and all checks completed, the aircraft
departed Southampton at 1905 hrs.

The takeoff was carried out with the commander as the pilot flying (PF) and the co-pilot as the pilot
non-flying (PNF). The PNF set the propeller RPM levers to 100% and the PF then set the engine
torque to 100% on both engines. The right engine achieved 100% slightly before the left but both
were matched at 100% torque. The aircraft accelerated normally with the PNF calling the airspeeds
in accordance with normal company procedures. The commander followed the Standard Instrument
Departure (SID) which required a climb to 2,000 feet with a right turn onto a track of 360º climbing
to 4,000 feet. Landing gear and flaps were retracted and power and propeller RPM set for the climb.
During the climb the aircraft passed through a layer of stratus cloud but no airframe icing was
encountered; the engine anti icing system had been selected on prior to departure. The behaviour of
the aircraft, including the rate of climb, all appeared normal. Radar control of the aircraft was passed
from Solent Radar to the London Terminal Control Centre (LTCC) and the aircraft cleared to the
requested cruising level of FL160 as they proceeded under their own navigation direct to NORRY
(a virtual waypoint near Pangbourne) In the darkness, there was no clearly defined horizon or
external references such as ground illumination or stars, and so the aircraft was being flown by sole
reference to the flight instruments.

Immediately before the incident, the PF had both his hands on the control column and his feet on the
floor. As the aircraft passed FL120, there was a loud bang from the area of the right engine.

20
Simultaneously, the red Central Annunciator Panel (CAP) warning light illuminated and the aircraft
decelerated, yawing and rolling rapidly to the right. Some passengers saw a shower of sparks from
the right engine, swept aft by the slipstream. The PF immediately placed both feet on the rudder
pedals and attempted to oppose the yaw and roll to the right with large amounts of opposite rudder
and aileron. The bang had been accompanied by a severe high frequency vibration which made
reading the engine instruments very difficult, and confirmation of the right engine problem was not
possible at that stage. With the IAS decaying rapidly, and the aircraft still yawing and rolling, in
order to maintain control the PF reduced both engine power levers to idle and lowered the aircraft's
nose to ensure a safe airspeed was maintained. The PNF was instructed to transmit a MAYDAY
distress on the operating frequency informing them of the situation and to ask for radar vectors to the
nearest airfield; this was acknowledged by London Control, who passed a heading of 080° to vector
the aircraft towards London Heathrow Airport.

With the aircraft under control, but with the vibration still present, the pilot advanced each power
lever individually to identify the vibrating powerplant and shortly after the failure the right engine
OIL caption, right generator failure and bus tie lights illuminated. Having positively identified the
right engine as the failed engine, the PF had his diagnosis confirmed by the PNF. The PNF
completed the memory items of the "ENGINE FAILURE OR IN FLIGHT SHUTDOWN" emergency drill with
the PF confirming the correct selection of operating controls prior to them being moved. Having
placed the right power lever to idle and the left propeller RPM lever to 100%, the PNF turned and
pulled the right engine propeller feathering lever and heard the distinct sound of the propeller
feathering. He then visually confirmed that the propeller had stopped, and that the engine fuel cocks
for the right engine had closed. The PF considered that this activity led to a loss of some 2,000 feet
altitude before the aircraft levelled off. The aircraft was re-trimmed for the asymmetric condition
using rudder and aileron trim.

The secondary checklist items were commenced with the PF monitoring the PNF actions. LTCC
offered Farnborough Airport as a closer diversion which was accepted with a request from the PF to
be vectored for an 8 mile final approach.

The cabin attendant who had been serving a light meal at the time of the engine failure moved to the
flight deck. She was aware of the high level of activity being undertaken by the two pilots, and was
careful to choose an appropriate moment to talk to the PNF. She was asked to brief the passengers of
their intended diversion to Farnborough and she informed the flight crew that there had been some
misty vapour and a smell of burning at the forward part of the cabin. All the passengers had their
seat belts fastened at the time of the incident and although understandably alarmed by the incident,
they were calm. The cabin attendant informed the passengers of the situation and secured the cabin
for landing.

21
The PNF tried to complete the emergency drills but had to respond to ATC messages and make a
written note of the Farnborough weather. He then extracted the Farnborough Approach charts from
the Aerad book and set the ILS frequency and course bar for Runway 24 for both pilots. The
approach minima were checked and the PF briefed for the approach. The PNF then completed the
emergency checklist followed by the descent and approach checks. During this time the aircraft
heading increased by some 90º, which, when noticed by the crew, was corrected although the PF was
having difficulty in holding the heading. Air traffic control of the aircraft was transferred to
Farnborough Approach, who requested which direction of turn the pilot would prefer. Left turns
were requested, towards the live engine, which took the aircraft onto the ILS localiser. An accurate
ILS approach was then flown, during which the landing gear and two stages of flap were lowered.
The PF maintained a higher than normal approach speed at 150 kt, some 20 kt faster than the 130 kt
required for the weight, and he first saw the runway lights at a height of about 400 feet. After touch
down a lower than normal amount of reverse thrust was used on the live engine together with
moderate wheel braking to bring the aircraft to a stop. The aircraft was then taxied to the allocated
parking stand, attended by the airfield Rescue and Fire Fighting Service (RFFS), where the crew and
passengers disembarked.

Weather

The synoptic situation at 1900 hrs showed a warm sector covering southern England with a light or
moderate south-westerly flow over the initial part of the route from Southampton to Manchester. The
surface weather was overcast with mist and some outbreaks of drizzle from extensive, low-altitude
stratus cloud. The METAR's for Southampton and Farnborough airports covering the departure and
landing were as follows:

Southampton: EGHI 101850Z 24007KT 5000 –RADZ SCT003 BKN004 11/09 Q1010=
Farnborough: EGLF 101920Z 22012KT 8000 DZ OVC005 11/11 Q1008=

Powerplant details

The engine involved was a Garrett TPE 331-12UHR-703H, built in 1989 as a model
TPE331-12UAR-705H but modified and redesignated as a –12UHR-703H in July 1998. It was rated
at 1,100 shaft horsepower (SHP) for takeoff, with a continuous rating of 1,050 SHP. The engine had
been owned by the engine manufacturer from new, and had been used as a loan unit throughout its life.

At the time of the incident the engine had accumulated a total of 10,154 hours from new. It was
installed on the right wing of G-BYRA on 25 November 2003, having accumulated 10,046 total
hours. Previously, in December 2000, the engine had been removed from a sister aircraft of the
operator's fleet and returned to the manufacturer's overhaul facility in Germany for inspection

22
following a birdstrike. Between completion of the birdstrike inspection in January 2001 and its
installation on G-BYRA in November 2003, it is understood to have remained in storage at the
manufacturer's overhaul facility in Germany.

The engine's last overhaul was in April 2000, some 525 hours prior to the incident, at which time the
reduction gearbox high-speed pinion, bull gear, and the sun gear forward bearing were replaced with
new components. The engine logbook contained no entries of significance to the investigation. The
records showed that the engine had been subject to spectrographic oil sample analyses (SOAP
checks) at the intervals specified by the manufacturer to provide warning of impending gear failures.
The SOAP trends showed no anomalies.

The engine was driving a Dowty Rotol variable pitch propeller.

Examination of right powerplant in situ

Oil was visible externally on the upper surface of the wing in the vicinity of the engine, and ground
staff at Farnborough reported that oil could be seen dripping from the engine cowlings immediately
after the aircraft had landed.

The propeller blades were fully feathered, but the propeller itself could not be turned beyond a very
small amount, comparable to the backlash which would normally be expected in the splined coupling
and second-stage reduction gears. It was apparent that the gearbox was effectively locked by some
form of internal obstruction.

The engine cowls were intact but a large number of the cowl fasteners had loosened off, or were
missing. Removal of the cowlings revealed widespread damage to ancillary components mounted on
the engine, consistent with exposure to severe vibration. A detailed description of this damage is at
Appendix A.

The engine/gearbox and propeller were removed from the wing in preparation for detailed strip
examination, with no further damage being noted. In particular, the rubber engine mount blocks
exhibited no signs of damage.

In summary, examination of the powerplant in situ suggested that a major failure of the input stage to
the reduction gearbox had occurred, resulting in a severe vibration which lasted for a period sufficient
to cause widespread secondary damage including disruption of fuel, oil, and electrical systems.

23
Engine and gearbox configuration

The Honeywell (Garrett) TPE 331 series powerplants comprise a range of single-shaft turbine
engines which drive the propeller via an integral reduction gearbox, the casing of which also
incorporates the engine intake duct. The gas generator comprises a two stage centrifugal compressor
supplying air to a reverse (forward) flow annular combustor, which surrounds the three-stage axial
turbine section. The combustion gasses are turned back through 180º at the inlet to the turbine
section, and after passing through the turbine, discharge rearward into a conventional exhaust duct.
The engine can be installed either erect (with the intake below the engine axis) or inverted (intake
above). On the Jetstream, the engine is mounted inverted.

The –12 engine, as installed on G-BYRA, was introduced in 1988 as a development of the –11,
providing an increased power margin for hot and high conditions. Although there are significant
variations in the detailed design of components across the full range of the TPE 331 format, the –11
and 12 variants are, for all practical purposes, identical.

Figure 1 is a cut-away view of the TPE 331 series engine and gearbox of the type installed in
G-BYRA. The 41,750 RPM rotational speed of the gas generator is reduced to 1,591 RPM at the
propeller in two stages. The first stage reduction is achieved via a simple pair of straight-cut gears,
comprising a high-speed (input) pinion mounted on the engine shaft driving a large bull gear. The
second (epicyclic) stage reduction comprises a sun gear, formed integrally with the forward face of
the bull gear, driving a set of planet gears which in turn drive the propeller shaft directly via a splined
coupling to the planet gear carrier.

Figure 2 shows a schematic sectional view through the reduction gearbox. The input pinion
(coloured dark blue) and bull gear (coloured red) are supported within a split aluminium housing
within the gear casing known as the diaphragm; the two halves are manufactured as a matched set.
The epicyclic second stage occupies the front half of the gear case cavity, formed by the nose casing.
The ring gear of the epicyclic stage is anchored to the forward face of the diaphragm.

Negative torque sensing

Because the power absorbed by a windmilling powerplant could potentially result in very high drag
being developed, and attendant aircraft control problems, a negative torque sensing (NTS) system is
provided which automatically adjusts the propeller pitch to minimise drag in the event of power loss.
It should be noted that the NTS system is a drag reduction system not a drag elimination system; it
cannot drive the propeller into the feathered position. Feathering per se is achieved solely under the
control of the pilot, by pulling the feathering handle to manually open the feathering valve. Until the
propeller is feathered and as long as the propeller rotates, there will be some windmilling drag.

24
To accommodate the NTS system, the attachment of the epicyclic ring gear to the diaphragm
incorporates a sloppy link, which permits a very small amount of rotational movement of the ring
gear to occur in either direction, before coming up against a hard stop and reacting any torque being
developed by the gear. The slight movement of the ring gear (within the sloppy link regime) arising
from a negative torque condition moves a linkage which causes the NTS dump valve to close,
creating an accompanying rise in NTS oil pressure. The increasing NTS pressure acts on an internal
portion of the feather valve, opening a bleed path in the oil circuit to the propeller dome, and
coarsening off the blade pitch toward the feather position. When a stage is reached where the
propeller is no longer back-driving the engine and a positive torque is sensed by the torque sensing
system, the change in torque produces an opposite movement of the ring gear within the sloppy link
regime, causing the bleed path to close and allowing oil back into the propeller to reduce blade pitch.
Thereafter, whilst the engine remains unable to drive the propeller normally, the torque tends to
cycle between positive and negative as the NTS system endeavours to maintain the propeller in a
minimum drag condition.

Detailed examination of the engine

Bulk teardown inspection

The engine was strip examined under the direction of the AAIB investigator at the manufacturer's
plant in the USA.

Preliminary external inspection revealed evidence of additional external damage, over and above that
detailed at Appendix A, consistent with heavy vibration comprising:

1. Deformation of the bearing plate on the accessory case which supports the hydraulic pump
take-off shaft.
2. Partial loosening of the feathering valve housing at its attachment to the engine case.

Removal of the nose casing, with the propeller shaft in-situ, revealed that the epicyclic reduction stage
was intact. The sun, planet, and ring gears displayed no overt signs of significant damage or
deterioration, and the mechanism which provided the mechanical signal to the NTS system was
undamaged. The crushed remains of a threaded bolt shank were found inside the nose casing together
with a quantity of metallic shards, and numerous internal casing screws were also loose or missing.

When the diaphragm plate was split from the rear section of the gear case, it was evident that a major
failure of the first stage gears had occurred, see Figure 3. The debris thus exposed included: broken
pieces of gear tooth; crushed remains of various bolts and studs; fragments of the subsidiary housing
which enclosed the input gears; and a large quantity of general debris.

25
A large segment of the bull gear rim, comprising approximately one third of the gear's
circumference, had separated and burst out through the side of the subsidiary housing and had
become jammed hard up against the starter generator shaft, fracturing and crushing the hollow shaft,
see Figure 4.

Although debris restricted movement of the remaining part of the bull gear, a small range of
movement was possible: sufficient to show that it was able to turn freely in its bearings.

Amongst the debris which fell clear during disassembly of the diaphragm were:

• A third, much smaller, fragment of bull gear rim encompassing four gear teeth.

• A small segment of the bull gear web, from a region adjoining the juncture of the fractures
associated with the large and small rim segments.
• Pieces of shattered subsidiary housing, from where the large bull gear segment had broken
through.

• Numerous miscellaneous fragments comprising broken and/or crushed remains of


components originating from the input stage gear casing.

The subsidiary housing was removed from the diaphragm, allowing removal of the remaining part of
the bull gear and the input (high speed) pinion. The input pinion was intact but all of the teeth had
been stripped, leaving just the remains of the tooth roots, see Figure 5. The drive shaft had sheared
from the pinion at the reduced section, which effectively formed a weak point shear neck.

Gear fracture details

Close inspection of the fractured pieces of bull gear revealed clear indications of fatigue consistent
with propagation from an origin region in the web at approximately 60% radius. Figure 6 shows the
fractured bull gear, with the principal separated fragments held in position to illustrate the fracture
paths and position of the origin region, from which two primary crack fronts propagated, labelled 'A'
and 'B' respectively.

The shorter of the two primary cracks ('A' in Figure 6) propagated radially outwards for a short
distance until it met the thicker section of the rim, where it turned and ran circumferentially for a
short distance before reverting back to a radial direction and intercepting the free edge of the wheel
at the root of a tooth. The much larger primary crack ('B') propagated radially inwards initially, in
opposition to crack 'A', then turned briefly in a circumferential direction (also away from crack 'A')
before turning back inward again to follow an oblique path towards the centre of the wheel. As it
approached the blend radius marking the transition from the web into the hub, the crack was

26
deflected again and thereafter it followed an oblique path back outwards towards the rim of the
wheel, where it intercepted the free edge at the root of a tooth. During the course of its progress,
numerous secondary cracks branched off crack 'B' to form a network of fractures in the adjoining
web. The consequence of this multiplicity of cracks was the separation and detachment of:

1. The large rim segment comprising approximately ⅓ of the wheel circumference, with
vestigial web attached.

2. The small rim segment comprising four teeth.


3. Numerous fragments of the inner web from within the region of secondary cracking,
identified in Figure 6, of which four pieces in total were eventually recovered and
positively identified. (Figure 6 shows only the largest of the recovered pieces.)

Significant heat discolouration was evident in the region of secondary cracking around the hub blend
radius, consistent with large-scale cyclic deformations of the material in these areas at a late stage in
the fracture process, prior to rim separation.

Tooth condition

The condition of the teeth on the main section of the fractured bull gear varied progressively from
completely stripped at 'X' in Figure 7 (adjacent to the large segment) through to a substantially
undamaged state at 'Y' (adjacent to the small separated rim segment). This pattern of damage,
together with the completely stripped condition of the high-speed pinion, was consistent with the
main segment of the fractured bull gear attempting to re-mesh with the high-speed pinion following
separation of the rim segments.

Metallurgical examination of the failed gear wheel

The fractured bull gear was subjected to detailed metallurgical examination at the manufacturer's
materials laboratory. This confirmed the provisional assessment of crack propagation made during
the teardown, and established that the fracture origins were on the aft face of the web approximately
3.7 inches from the rotational axis.

The failure characteristics were broadly comparable to those seen previously by the manufacturer
during post-failure investigations of fractured bull gears from TPE 331 –11 and –12 series engines,
extending over many years.

27
History of gearbox failures on TPE 331 series reduction gears

The –12 series bull gear is reportedly the most highly loaded of any gear in the Honeywell series of
propulsion engines, and the bull gear itself and/or its associated components have suffered a number
of failures since the type's introduction into service in 1983. Some of these failures resulted in the
ejection of uncontained debris, either from the gear case directly or via the air intake duct. Of the
latter, a number resulted in ejected debris being struck by the propeller and forcibly projected against
the fuselage side, in one instance resulting in penetration into the cabin. The manufacturer identified
imperfect tooth contact as being a significant causal, or contributory, factor in a majority of these
failures, giving rise to:

• Load pulses in the bull gear which excited a resonance mode leading ultimately to the
initiation of a fatigue crack in the web of the bull gear, and consequent separation of
segments of rim.
• Initiation of fatigue cracks in the roots of the teeth, which propagated into the rim and web
of the bull gear, resulting in separation of rim segments.

Factors previously identified as contributing, or potentially contributing, to contact pattern


degradation included:

• Changes to manufacturing methods, which had tended to produce an involute profile with
inadequate tip and/or root relief, leading to increased tooth loading (on an already very
highly loaded gear).

• Distortion, over time, of the housings in the diaphragm plate which supports the bull gear
and high-speed pinion bearings, resulting in displacement of the centres of rotation of the
support bearings by as much as 0.006" from their correct positions; giving rise to
associated tooth misalignment and increased wear.

• Re-dressing of gear teeth profiles during overhaul.


• Installation of gears in unmatched sets.

Remedial measures taken to date

In an effort to address the problem of bull gear system failures, the manufacturer implemented a
number of remedial measures, with variable success, which are summarised together with their
outcomes below.

28
Service Bulletin TPE331-A72-2011

A design change to the bull gear was introduced in 1997 via Service Bulletin (SB) TPE331-A72-
2011, which comprised:

1. The introduction of a Metco spray coating on the web adjacent to the rim, intended to
increase the damping inherent in the gear itself thereby reducing the potential for damaging
amplitudes to occur within the various flexural modes of the gear.

2. Shot peening of the tooth roots, to locally increase resistance to fatigue initiation in those areas.

Testing of this revised design showed that it potentially cured the resonance and related fatigue
initiation problems within the bull gear, but the manner in which the revised gear was introduced into
service threw up a range of new, but related, failures affecting the high-speed pinion. Specifically:

• Production of the revised bull gear included reworks of the original pattern gear
(P/N 3102585). Re-working of the gear teeth, a consequence of which tended to be a
further reduction in the amount of involute tip/root relief.
• Mixing of existing (used) high speed pinions with reworked/new-manufacture revised pattern
bull gears occurred, with consequent deleterious implications for tooth contact pattern.

The outcome was that whilst the revised bull gear was successful in reducing the incidence of
resonance-induced failures and rim separations specifically, the (unhardened) splines on the high
speed torque shaft, which transmitted torque from the engine to the high speed pinion, started to
wear. As a consequence of this wear, high cycle fatigue cracks initiated in the non working splines
which in turn led ultimately to failures of the high speed torque shaft.

Compared with bull gears which had been reworked from the original pattern to incorporate the
changed design, those gears manufactured as new components to the revised design (allocated
P/N 3108197) appeared relatively immune to these problems.

The issue of diaphragm distortion, and attendant alignment problems, remained unaddressed.

Service Bulletin TPE331-A72-2062

SB TPE331-A72-2062 was introduced in 1999 removing the Metco coating, on the assumption that
in-service failures would return to a pattern similar to that associated with the original design of bull
gear. The shot peen element of the revised design was retained, however.

29
The situation did indeed revert partially to the original failure pattern: all of the previously identified
initiating factors were present including gears with re-worked teeth; gears installed on a mix and
match basis; and diaphragm distortion. Moreover, without the damping provided by the Metco
coating, resonance-induced failures returned, including some rim separations. Mostly the failures
affected re-worked gears, but some new gears also failed; these were all associated with heavily
deformed diaphragms, which the manufacturer estimated caused potential increases of up to 40% in
the dynamic loading of the gear teeth.

The shot peening of the teeth, which had been retained from Service Bulletin TPE331-A72-2011,
appeared to be effective in reducing crack initiations in the tooth roots; instead, failures were tending
to originate in the web.

Diaphragm inspection & matched-pair gears

Revised interim measures were introduced in 2001 pending a complete redesign of the bull gear.
These measures comprised:

1. Inspection of the diaphragm bearing housing positions to detect and reject housings which
had suffered displacement beyond acceptable limits.

2. Introduction of matched-pair gear sets, incorporating:

• Shot-peened tooth roots.

• Improved tip relief.

• Reduction in life to 3,500 hours (previously on condition).

Introduction of SOAP checks at 100 hr intervals

Because sub-optimal tooth contact patterns had been established as a factor contributing to the
failures, and such tooth contacts invariably result in abnormal tooth wear, the manufacturer
considered that monitoring the rate of accumulation of tooth wear-product in the lubricating oil could
provide prior warning of impending failure. Accordingly, a requirement was introduced for regular
spectrographic analysis of oil samples taken when the oil filter was changed. The interval between
filter changes (and hence SOAP checks) was specified at 200 hrs when the requirement was first
introduced in 1999, later reduced to 100 hr intervals in October 2001.

30
Re-designed gear sets

A permanent solution to the problem was sought through a total re-design of the high-speed pinion
and the bull gear. The revised gear sets, which incorporate a helical tooth form, thicker web, thicker
rim, and improved cooling, was implemented by SB TPE331-A72-2114 in August 2004.

Recorded data

Data sources

The aircraft was fitted with a 30 minute Fairchild A100A Cockpit Voice Recorder (CVR) and a
25 hour, 5 parameter Honeywell UFDR Flight Data Recorder (FDR). The CVR had been left
running after the incident so did not yield any useful information. The FDR recorded altitude,
normal acceleration, indicated air speed, heading and the VHF transmission key.

Other sources of recorded information included radar data, weather data and ATC recordings from
five ATC centres including Southampton and Farnborough.

The radar provided track data from a point shortly before the failure event until shortly before touch
down. The altitudes recorded were consistent with the FDR record, and provided a common reference
allowing UTC time stamping of the FDR record. The radar track was also found to correlate well with
the FDR heading and speed parameters. The weather data correlated approximately with the difference
between radar-recorded ground speed and FDR-recorded indicated airspeed.

The quality of ATC data varied by source with some of the 60 second stamped periods taking
64 seconds to replay. The ATC data was correlated with the radar and FDR data by matching the
VHF transmission key FDR parameter with the ATC recorded transmissions.

Interpretation of the data

At 1910 hrs UTC the aircraft started climbing out of Southampton Airport in a southerly direction
before turning onto a northerly heading and starting its climb towards FL70 in accordance with ATC
directions. This climb profile was maintained in accordance with ATC instructions through various
ATC hand overs.

At 1920 hrs the aircraft was at FL120 climbing for FL140 and still heading north when the normal
acceleration trace 'spiked' down to 0.28g and became very noisy, accompanied by a heading change
to the right. These events, which were consistent with the sudden loss of thrust which was caused by
the No 2 engine gearbox failure, were accompanied by a reduction in altitude of approximately
130 feet, followed by a brief recovery; then a reducing altitude once more. Thereafter, the 'noise' in

31
the normal acceleration parameter remained for approximately two minutes before abruptly
disappearing. During that time the right hand turn continued accompanied by reducing altitude. The
turn continued for approximately one minute after the 'noise' in the normal acceleration trace ceased;
after that the flight path was consistent with ATC instructions to descend and divert to Farnborough.
During the period between the failure event and the aircraft reverting to follow ATC instructions for
the descent into Farnborough, the aircraft had turned through approximately 180º from its heading at
the time of the event.

An audio analysis was undertaken of those parts of the ATC tapes containing transmissions from the
aircraft, in order to check for anomalies before, during or after the vibration period. However, the
analysis was hampered by the variable quality of the recordings and the fact that the aircraft signal
would have undergone a number of processes before finally being recorded. The ATC tape captured
a configuration alert when the aircraft lined up with the Farnborough runway centre line but nothing
further of significance was found.

Analysis

The gearbox failure

In light of the extensive prior history of bull gear failures and the program of implemented and
planned activity by the manufacturer to address the issue, the focus of the technical investigation was
directed primarily towards:

1. Establishing whether this failure exhibited features which suggested that new and
previously unidentified causal factors might have come into play.
2. Studying the implications of the gear failure so far as they affected aircraft handling, and in
particular the crew's ability to retain full control of the aircraft and subsequently to execute
a safe landing.

The widespread damage within the gearbox was such that it was not possible to determine in every
detail the sequence of failure, but it was clear that most of the damage was secondary and that it had
been sustained as a direct consequence of a fatigue failure of the bull gear, which resulted in the
release of large segments of the gear rim into the gear housing and adjacent rotating components.
Detailed visual and metallurgical examination of the failed bull gear showed that its mode of failure
fell within the ambit of the modes identified in previously studied instances of bull gear failure.
There was little doubt too that the underlying causal factors associated with these previous failures
would have been applicable to G-BYRA, such as uneven tooth wear causing increased tooth loading
and the excitation of resonance modes leading to fatigue failure from origins in the web of the gear.
Whilst the full range of contributory factors in this case could not be identified, due to secondary

32
damage, these would have been likely to fall within the range of factors identified previously, and
summarised earlier in this report.

Whilst the failure of the bull gear undoubtedly fell within the known pattern of crack propagation
and rim separation, the consequences of the failure in this case was not typical insofar as it set in
train a sequence of events which compounded significantly the problems faced by the crew in their
efforts to deal with the resulting emergency. The following technical analysis will therefore focus
primarily on the implications of the failure for the continued safety of the flight.

Flight safety implications

Whilst in the case of G-BYRA no debris was ejected through the gear case or from the engine intake,
and consequently there was no direct risk to the occupants arising from potential penetrations of the
fuselage by debris, the engine was subjected to extreme levels of vibration as a consequence of the
gear failure which, indirectly, did hazard the aircraft. Specifically:

1. The level of vibration was such as to fracture, or compromise the integrity of, both fuel and
oil supply lines together with proximate electrical systems, creating a potential fire hazard.
2. Substantial numbers of stiff-nuts and studs on the gear case itself, and on associated
components mounted on the forward end of the engine casing, were vibrated off or otherwise
compromised in a manner which suggested that, had the vibration persisted and the damage
accumulated, it could have threatened the structural integrity of the gear case; with an
attendant potential for the nose housing, and propeller, to separate from the aircraft.

3. The heavy vibration associated with the failure rendered critical flight deck instruments
unreadable, adding to the difficulties faced by the flight crew in retaining control of the aircraft.

It is clear, therefore, that the vibration which followed as a direct consequence of the bull gear failure
was an important factor in terms of flight safety, and both its cause and its effects require further
analysis to determine its significance in terms of the continued safety of the flight.

Source of the vibration

If the bull gear had continued to rotate post-failure, whether driven from the engine or back-driven
from the propeller, then the resulting rotational imbalance would have generated a high-frequency
forced vibration of the whole engine installation, capable potentially of causing both the observed
secondary damage and the vibration of the instrument panel on the flight deck. No other potential
source of the vibration could be identified but for such a condition to have occurred, it would have
required the bull gear to remain clear of obstruction.

33
Based on a visual assessment alone, it was not possible to determine with confidence whether the
separated rim segment, which had burst through the side of the subsidiary housing and become
jammed against the starter generator connecting shaft, would have obstructed the remaining part of
the bull gear and thus prevented its rotation in the post-failure period.

In order to resolve this issue a simple 3-D CAD model was constructed of the detached rim segment,
together with the remains of the fractured gear and relevant parts of the adjacent gearbox casing.
After setting up the model to match the observed position of the jammed rim segment, (see Figure 8),
it was evident that the combination of its radial position, and tilted orientation relative to the
surviving part of the gear, was such that it would not have interfered. Consequently, it would have
been possible for post failure rotation of the bull gear to have taken place, and thus for it to have
been the source of the severe vibration, provided the drive line from the engine, or alternatively from
the propeller shaft, had remained intact following gear fracture.

Based on the distribution of secondary damage within the gearbox, it is very unlikely that the engine
would have been capable of driving the fractured bull gear. The stripped condition of the high-speed
pinion is consistent with its forced re-engagement with the disrupted rim of the bull gear in the
instant following separation of the detached section of the rim. This process is likely to have been
virtually instantaneous – due to the rapid acceleration of the gas generator shaft in the interval
between the shaft becoming unloaded, as the pinion disengaged momentarily from the fractured bull
gear, before attempting to re-engage again as it came into contact with the far side of the missing
section of rim. Indeed, the distribution of tooth damage on the bull gear indicates that all of the
pinion's teeth were effectively wiped off within about half a revolution of the bull gear, and there is
little doubt that the pinion coupling sheared at some point during this sequence. From that stage
onwards the engine would have been disconnected from the gearbox input and, assuming that the
fuel delivery side of the system was still operable, it would have accelerated rapidly until constrained
by the overspeed governor. Continued operation of the engine, however, would not have been
possible because of collateral damage from the gear failure, which disrupted the drive to the fuel
control unit thus rendering the gas producer inoperative. In summary, therefore, it is evident that any
post-fracture rotation of the bull gear, and the attendant vibration, must have been the result of it
being back-driven by the propeller.

The FDR data provides further evidence in support of the contention that the severe vibration was
associated with continued rotation of the propeller. Examination of the normal acceleration trace
showed the onset of an apparently high frequency vibration, manifesting as noise on the data trace, at a
point when other recorded parameters showed changes consistent with the gear-failure event. Because
of the poor frequency response of the recording system, the true frequency and amplitude of this
vibration could not be established and so it was not possible to educe its source from the frequency

34
content of the recorded signal. However, not only does the vibration begin coincident with the gear
failure but it also ends approximately two minutes later at about the time, based on the crew's
testimony, when the propeller was feathered. Assuming that the noise on the FDR signal is indeed a
reflection of the very heavy level of high-frequency vibration evidenced by the engine damage and the
crew's account of unreadable flight deck instruments, then it was undoubtedly significant. Also, it
appears to correlate with the period between gear failure and the propeller eventually being feathered –
suggesting strongly that the vibration was associated with a windmilling right propeller. Furthermore,
the mere fact of its detection by the aircraft's normal 'g' transducer (which is mounted in the fuselage
and not sensitive to high-frequency vibrations) implies that the level of vibration at its source,
significantly remote from the transducer, must have been severe indeed.

Normally, the NTS system will modulate blade pitch to minimise the drag caused by a windmilling
propeller, but in this case the effective disconnection of the bull gear from the input pinion would
have compromised the system's ability to function as intended and as flight tested. The drive train to
the propeller governor arguably remained intact during the period that the propeller apparently
continued to rotate post gear failure, and during this period it potentially would have had the
capability to modulate propeller pitch in an effort to hold the selected propeller RPM. However, the
governor's oil supply would potentially have been compromised at an early stage, because of the
early failure of the input pinion from which the engine oil lubricating pump is driven, and the
attendant cessation of a pumped supply of oil to the oil gallery feeding the governor. Whilst the
behaviour of both the NTS and propeller governing systems under these abnormal conditions cannot
be predicted with any degree of confidence, it is evident from the FDR traces and from the crew
testimony that sufficient propeller pitch was maintained to cause the propeller to back-drive the
broken bull gear, causing the severe vibration reported by the crew. It is also possible that, until the
feathering valve was operated manually by the crew and the propeller actually achieved a feathered
state, the compromised propeller pitch control system may have caused abnormal blade-pitch
variations to occur, with attendant yawing moments being imposed on the aircraft.

Summary of findings arising out of the technical investigation

The failure of the bull gear system occurred as a result of a fatigue failure initiating in the web of the
bull gear, which propagated on two fronts leading ultimately to the separation of a large segment of
outer web and rim comprising approximately one-third of the gear's circumference. In terms of both
its overall characteristics and underlying causal factors, the failure fell within the ambit of previous
failures which have been studied in detail by the manufacturer and addressed by a series of remedial
measures, culminating in a total redesign of the bull gear and related components, implemented
under SB TPE331-A72-2114 in August 2004.

35
Following the gearbox failure and associated disconnection of the engine input, the fractured bull
gear was back-driven at speed by the windmilling propeller for approximately two minutes before
the crew was able to identify the source of the problem and feather the No 2 propeller. During this
period, the resulting rotational imbalance caused a severe vibration which resulted in extensive and
potentially hazardous secondary damage to oil, fuel, and electrical components mounted on the
engine, and to the gearbox casing. The vibration also: rendered the flight deck instruments
unreadable, seriously compromising the crew's ability to identify which powerplant was affected and
frustrated their efforts to deal with the emergency. Any oscillations in propeller pitch which may
have occurred during this period, due to abnormal operation of the NTS and/or propeller governor
systems, could have given rise to oscillatory yawing moments producing an associated dutch-roll like
motion of the kind reported by the crew. This would not only have added to their control problems,
but also made it more difficult for them to identify promptly which powerplant was malfunctioning.

Handling of the emergency

At the time of the gearbox failure, the aircraft was in the climb on a heading 360º, with the yaw
damper engaged. The loss of power from the right engine would have caused the aircraft to yaw to
the right with an associated roll to the right. The yaw damper will disengage when the rate of roll
exceeds 7º sec or at bank angles greater than 45º; consequently it is not likely to have made any
significant contribution to the crew's efforts to retain control of the aircraft.

Illumination of the oil pressure, generator fail and bus tie warning lights was indicative of the engine
having stopped and not just run down to idle yet the pilot experienced great difficulty, not only in
correcting the departures, but also in preventing himself from over-controlling as if the aircraft yaw
to the right was not constant but varying as if power was fluctuating. For reasons discussed earlier in
the technical analysis, the yawing and rolling oscillations are likely to have been caused by
fluctuations in blade pitch on the right propeller associated with either the governor's attempts to
control the RPM on the disconnected and windmilling propeller, or erratic intervention of the NTS
system. This is not a condition which would normally accompany a loss of thrust engine failure and
it would have presented significant handling difficulties, both directly and indirectly by masking the
underlying problem, thus compromising the crew's ability to take appropriate corrective action.

With no external horizon, all flight was by sole reference to instruments, which were being shaken
by the severe vibration. It was not possible to read the engine instruments and in order to try and
stabilise the aircraft the pilot brought back to the idle position the two engine power levers. His plan
was to accept a loss of altitude whilst maintaining a safe IAS in order to maintain control of the
aircraft and give him and the PNF time to properly analyse the situation. However, whilst this was
an entirely appropriate action in the circumstances, and one that would have reduced the steady-state

36
component of the thrust asymmetry, it would not have eliminated any yawing oscillations associated
with blade-pitch oscillations of the right propeller. As the instruments were still not readable, but
being aware that some form of engine problem was present, the PF advanced each engine power
lever in turn and, on finding a positive power increase on the left and none on the right, confirmed
the right engine was malfunctioning. The shut down and feathering drills were then accomplished
using the abnormal checklist, and once the right propeller was feathered the vibration ceased and the
pilot was able to continue the flight using the left engine with the flight controls trimmed for the
asymmetric condition. By the time that this process was complete and the crew were in a position to
begin their recovery to Farnborough, some three minutes after the initial failure event, the aircraft
had gently turned through 180º; this was probably due to a break-down in instrument scan, due to the
concentration required to maintain wings level and to maintain the desired IAS attitude.

The increase in the approach speed of 20 kt was to ensure that in the event of a go-around and any
re-emergence of handling difficulties, an increased safety margin above VMCA was available. The
increase still permitted a safe landing to be made in the landing distance available.

Conclusion

This serious incident was the result of a major failure of the propeller gearbox which led to the
aircraft yawing and rolling to the right and left. The crew were unable easily to identify the nature of
the problem but with some difficulty they maintained control of the aircraft. Through a process of
elimination, the crew correctly identified the right powerplant as the source of the problem. When
the engine was shut down and the propeller feathered, the aircraft's handling qualities were restored
to normal single-engine flight.

Safety action

The technical analysis has shown that the gear failure on G-BYRA fitted a pattern of failure already
identified by the manufacturer as a result of its investigations of previous gear failures. The factors
contributing to the failure on G-BYRA also fell within the ambit of factors identified previously as
having contributed to these earlier failures, which have themselves been the subject of a range of
remedial measures already implemented by the engine manufacturer. The culmination of these
actions by the engine manufacturer, comprising the introduction in August 2004 (via SB TPE331-
A72-2114) of a completely redesigned bull gear assembly, renders moot any recommendations
arising out of this investigation which might otherwise have been made.

37
input
(high-speed)
pinion

bull
gear

Figure 1:
Configuration of TPE 331 engine as installed on the Jetstream

Figure 2:
Sectional view through gearbox (schematic)

38
diaphragm

separated
segment of remains of bull gear in
bull gear rim situ inside subsidiary
casing

ancillary drive
gears

burst
subsidiary
housing

Starter generator intact ancillary drive to


connecting shaft propeller governor
Figure 3:
View onto aft face of diaphragm, with damaged components in situ

39
intact ancillary drive to
burst propeller governor
subsidiary
case

Starter generator
connecting shaft

separated
segment of
bull gear rim
Figure 4:
Detail of separated segment of bull bear

Figure 5
Input (high speed) pinion, showing sheared drive neck

40
41
Figure 8:
CAD model showing relative positions of
separated rim segment and remains of bull gear

42
APPENDIX A TO EW/C2004/01/02

DETAILED DESCRIPTION OF ENGINE DAMAGE

1. Fracture and break-up of the mounting brackets supporting the igniter control box, leaving
the box loose and supported only its connecting cables.

2. Fracture of the air-bleed pipe supplying the intake de-icer, at a point adjacent to the off-take
boss on the compressor case, together with cracking of the boss itself.

3. Loosening and separation of a 'B nut' securing a main lubrication pipe to a 'T' connector on
the right side of the engine, just below the fractured air-bleed pipe.

4. Fracture of the main fuel feed pipe adjacent to its connection to the fuel flow transducer.

5. Loosening of the fuel flow transducer body in its mounting clamp, and partial migration of
the unit out of the clamp.

6. Numerous loose and missing stiff-nuts, studs, and bolts securing the gearbox nose-cone to
the diaphragm plate, and the diaphragm plate to the main gear case.

7. Several fractured studs and missing nuts on the fuel control housing, primarily at a flanged
joint between the inner and outer ends of the housing immediately aft the unit's fixture to
the accessories case, resulting in springing of the affected joint and partial expulsion of the
associated O ring seal.

8. Deformation of the firewall ring-seal diaphragm at the jet pipe connection, consistent with
violent oscillation of the entire powerplant on its rubber mounts.

9. Fracture of studs attaching the fuel-cooled oil cooler to the gearbox case.

10. Fracture of the shear neck coupling to the starter generator, and deformation of the
associated transfer shaft bearing-retainer in the accessory case.

The engine oil tank was empty. Significant quantities of metallic shards were present in the oil filter
housing, but its impending bypass tell-tale button was in its normal (flush) position.

There was no evidence of fire or elevated temperatures anywhere within the engine nacelle.

43
A visual inspection of the turbine via the jet pipe revealed that the pair of studs securing the turbine
rear bearing cover had fractured, and the cover plate was missing. The turbine itself appeared to be
intact but slight metal spatter was visible on the final-stage guide vanes. Manual rotation of the
turbine was possible, and produced comparable rotation of the compressor confirming that the main
shaft was intact between the compressor and the turbine sections; however, substantial roughness
was apparent in the bearings. Examination of the compressor face, via the inlet duct, showed that the
first stage impeller was intact but there was evidence of severe tip rubbing which, together with the
roughness of rotation, was indicative of front bearing failure. With the exception of the missing rear
bearing cover, there was no evidence of separation or non-containment of the engine core.

44

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy