AAIB Bulletin: 6/2010

G-FBEH EW/C2008/08/01

SERIOUS INCIDENT

Aircraft Type and Registration: Embraer ERJ 190-200 LR (Embraer 195), G-FBEH

No & Type of Engines: 2 General Electric Co CF34-10E7 turbofan engines

Year of Manufacture: 2007

Date & Time (UTC): 1 August 2008 at 1220 hrs

Location: 40 nm NW of Wallesey, en route from Manchester to

Belfast City

Type of Flight: Commercial Air Transport (Passenger)

Persons on Board: Crew - 5 Passengers - 90

Injuries: Crew - 1 (Minor) Passengers - 4 (Minor)

Nature of Damage: No 1 air cycle machine failure

Commanders Licence: Airline Transport Pilots Licence

Commanders Age: 48 years

Commanders Flying Experience: 6,500 hours (of which 410 were on type)
Last 90 days - 147 hours
Last 28 days - 65 hours

Information Source: AAIB Field Investigation

Synopsis

The aircraft was operating a scheduled passenger ground. Two Safety Recommendations are made as a
transport flight with the No 2 air conditioning pack result of this investigation.
inoperative, as permitted by the Minimum Equipment
History of the flight
List (MEL). Whilst en route, a failure of the No 1 Air
Cycle Machine (ACM) occurred, releasing smoke and The crew reported for duty at Belfast City Airport at
fumes into the aircraft. A MAYDAY was declared 0445 hrs for a foursector day. The first sector was to
and an expeditious diversion was carried out. After London Gatwick, where the crew made a planned aircraft
donning oxygen masks the pilots had great difficulty change onto G-FBEH for the return flight to Belfast.
communicating with each other, ATC and cabin This aircraft had experienced a fault with the No 2
crew, because of technical problems with the masks. air conditioning pack on 28 July 2008. The pack had
During the emergency evacuation the right overwing remained unserviceable since then and the defect was
emergency exit door became jammed and unusable. recorded in the aircraft technical log as an Acceptable
Passengers who evacuated via the left overwing exit Deferred Defect (ADD). The flight crew confirmed from
were unaware of how to get from the wing down to the the MEL that dispatch with this defect was allowed for up

Crown copyright 2010 20

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

to 10 days; with the limitation that the maximum altitude performed, during which the co-pilot reviewed the
be restricted to FL310. After returning to Belfast they emergency descent checklist and selected the emergency
then flew the aircraft to Manchester. All three sectors code, 7700, on the transponder. Given the absence of
were without incident. any flight deck warnings or visible smoke and the limited
time available for planning the approach, the flight crew
The final sector of the day was scheduled to be from did not refer to any other emergency checklist.
Manchester to Belfast City. The aircraft took off at
1150 hrs, with the commander operating as handling Communication whilst wearing the oxygen masks
pilot. Approximately 10 minutes after takeoff, during proved very difficult due to technical problems with the
the climb to the final cruising level of FL240, both pilots masks. The co-pilot had to repeat calls to ATC to make
smelt a sulphurous burning smell, similar to that of a himself understood and communications between the
match being struck. They contacted the Senior Cabin two pilots were rendered so poor that they had to resort
Crew Member (SCCM) by interphone to ask if he could to shouting.
smell it in the cabin and asked him to check the forward
toilet, which is close to the flight deck, as they considered The SCCM had tried to contact the pilots by interphone
the smell might have been due to a passenger smoking in during the descent to inform them that the smell in the
the toilet. The SCCM and a cabin crew member from the cabin was getting worse and that the haze was now also
rear of the aircraft reported that there was no evidence visible in the front of the cabin. Although both pilots
of anyone smoking in the toilet, but they could smell could hear him, he could not hear them and the pilots
something in the cabin and a haze was visible from the activated the cabin emergency call bell. The SCCM,
rear of the cabin. When interviewed after the incident, still unable to communicate with them by interphone,
the crew commented that the smell was unfamiliar to initiated the emergency access procedure and gained
them, which heightened their concern. entry to the flight deck. The commander told the SCCM
that he intended to land as soon as possible and ordered
The smell became sufficiently strong on the flight deck him to secure the cabin. The SCCM was advised to
that the pilots decided to don their oxygen masks. expect a normal landing, but was not told that they
The aircraft was approximately midway between would be landing at Ronaldsway. The commander did
Manchester and the Isle of Man and the wind direction not make an announcement to the passengers because
of approximately 210 at about 15 kt made a straight-in of the communication problems experienced whilst
approach to Runway 26 at Ronaldsway Airport (Isle of wearing his oxygen mask and the limited time available
Man) favourable. The commander was familiar with the to prepare for the approach.
airport and, concerned that the smell might have been
due to a fire, decided to divert there. Manchester ATC transferred the aircraft to Ronaldsway
ATC who offered them either a Surveillance Radar
The co-pilot requested a descent from Manchester ATC Approach (SRA) or an NDB approach to Runway 26.
and clearance was given to descend to FL200. He then The flight crew accepted the SRA and requested that the
declared a MAYDAY and informed ATC of their decision fire services be in attendance for the landing.
to divert to Ronaldsway. An expeditious descent was

Crown copyright 2010 21

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

The cabin crew stated that the smell came and went approximately midway up the cabin, over the wings. On
during the flight. The SCCM reported that whilst on hearing the order to evacuate, the cabin crew opened their
the approach to Ronaldsway the smell intensified again, allocated doors, the escape slides inflating automatically.
becoming stronger than before and smoke was now The SCCM initially prevented passengers using Door1
visible in the cabin. He advised the commander, who Left (D1L) as the slide had not fully inflated by the
considered that he would probably conduct an evacuation time the first passenger arrived there. Once it was fully
on landing. He did not communicate his intent to the inflated, the SCCM had to push himself past the flow of
SCCM or ATC as he thought that to tell them anything at passengers to reach Door 1 Right (D1R) to open it. He
this late stage of the flight might cause confusion should commented that had he opened this door first, given the
he decide not to order an evacuation. layout of the cabin, he would not have been able to push
past passengers to get to D1L (Figure1).
The pilots continued with the SRA and became visual
with the runway at an altitude of about 700 ft. The Passengers commented that they found the slides very
commander completed a visual approach and landing on steep and were surprised by the speed at which they slid
Runway 26 and brought the aircraft to a halt at a runway down them. The slides also ended without any round-out
intersection, turning it into wind as he did so. He then at the bottom, causing passengers to slide straight onto
ordered the cabin crew over the Passenger Address (PA) the ground at speed. This, and attempts by passengers to
system to stand by, and a few seconds later, gave the slow themselves on the slides, were the principal causes
order to evacuate. of injury reported. The cabin crew became aware of
the problems and tried to reduce injuries by instructing
Aircraft evacuation
passengers to sit down as they got onto the slide and
The aircraft was equipped with six emergency exits: four by controlling the flow of passengers down the slides.
doors fitted with inflatable slides, two at either end of Particular attention was paid to the older and more infirm
the cabin, and two Type III emergency exits located passengers.

D1R

D1L
Figure 1
Forward cabin layout, showing forward exits (Doors 1 Left and Right)

Crown copyright 2010 22

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

When the order to evacuate was made,

passengers were able to open the left overwing UP
Ceiling
exit door and evacuate onto the wing. Attempts edge panel
to open the right overwing exit door proved
unsuccessful, as the forward upper part of FWD

the door trim had become jammed under the

ceiling edge trim panel, preventing the exit Door trim
from being opened (Figure 2). panel

Passengers evacuating via the left overwing

exit reported that once out on the wing, there Figure 2
was confusion as to how they should get off Right overwing emergency exit showing door trim partially
jammed (circled) under ceiling edge panel
the wing down to the ground. A 61 cm-wide
walkway was demarcated at the wing root in black paint, deployed. Some passengers had been queuing to use the
with arrows pointing towards the trailing edge (Figure 3). overwing exit when they were called to the rear of the
This was not noticed by some passengers; one passenger aircraft by the cabin crew to use the rear exits, once they
thought that the markings denoted an engineers walkway, were clear of other passengers. This included a passenger
rather than an escape route. The overriding comment seated one seat away from the left overwing exit, who
from passengers who evacuated onto the wing was that stood in the aisle and assisted passengers evacuating via
it was not obvious to them that they were meant to climb that exit.
off the wing via the trailing edge. Although the wing
flaps were lowered in accordance with the emergency
evacuation checklist, there remained a considerable drop
to the ground of about 1.7 metres.

Two male passengers who evacuated via the overwing

exit were able to jump down from the rear of the wing
and assist other passengers to the ground. This included
a mother carrying a baby. They believed that had they not
been able to offer such assistance, it is likely that some
of the passengers might have received serious injuries in
attempting to climb off the wing. Passengers believed
that the situation would have been worse had it either
been raining or dark at the time of the evacuation.

Figure 4 illustrates the exits used by the passengers, Figure 3

correlated by seat position. It shows that no passengers Overwing exit evacuation route markings
used D1R, despite this door being open with the slide (left wing shown, view towards wing trailing edge)

Crown copyright 2010 23

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

D1R SLIDE
D1L SLIDE

LEFT OWE RIGHT OWE

Unknown whether right

or left rear slide used

Unoccupied seats
(one of these seats was
occupied by a passenger,
but unknown which one
or which exit the
occupant used)

D2L SLIDE D2R SLIDE

Evacuation Routes Used

Figure 4
Evacuation routes used by passengers, correlated by seat position

The cabin crew estimated that all the passengers had to the rear, collecting their high visibility vests and a
exited the aircraft within one minute, following which megaphone and evacuating via Door 2 Left (D2L).
the two cabin crew from the rear of the cabin checked that
no one was still on board. They reported to the SCCM The pilots attempted to communicate with ATC and the
that the cabin and toilets were clear before returning attendant fire services by radio, but this proved difficult

Crown copyright 2010 24

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

because of the continuing technical problems with their were sourced from a number of microphones including
oxygen masks. They eventually removed the masks and each flight crew members (headset) boom microphone,
opened the window to speak to the fire services directly. both flight crew oxygen masks, the PA system, and the
On completing the emergency evacuation checklist the Cockpit Area Microphone (CAM).
pilots entered the cabin, by which time only the SCCM
Voice and flight data
was present. The latter had been concerned that the crew
had not emerged earlier and, with no peephole to see into Each recorder was successfully downloaded. The data
the flight deck, had resorted to banging on the door to show that while climbing through FL156, the co-pilot
attract their attention. The commander conducted a final identified a burning smell similar to that of a lit match.
search of the cabin and both pilots and the SCCM then Around three minutes later, the commander said to the
evacuated via D1L. co-pilot, oxygen on mate, oxygen on. The DVDR
then automatically switched to record crew speech from
Once outside, one of the cabin crew used the megaphone the microphones in the oxygen masks.
to assemble the passengers on an area of grass at the side
of the runway. They also assisted passengers who were The co-pilot declared a MAYDAY and requested a further
distressed or injured. descent to FL100. This request was not acknowledged
initially by ATC, and only fragmented speech was
Pre-flight emergency briefing
audible on the recording from the co-pilots oxygen
Prior to departure, passengers seated next to the mask microphone.
overwing exits were briefed by the cabin crew on how to
operate the exit. There were also instructions attached At around the time the oxygen masks were donned,
to the seatback in front of these passengers, included in the FDR data show an unusual drop in the Pack 1
which is the depiction of an arrow apparently guiding flow rate and compressor outlet temperature. Prior to
passengers towards the trailing edge of the wing. Safety this, the flow rate was variable about a mean value of
cards, provided for all passengers, included diagrams around 70pounds per minute (lb/min) initially, rising to
depicting passengers climbing off the trailing edge of 75 lb/min with peaks of 90 to 91 lb/min. (Other data
the wing onto the ground. provided by the operator for the same aircraft with both
packs operating showed that the pack outlet temperatures
Following this incident the operator revised its briefing and flow rates were generally lower than under
to passengers seated next to the overwing exits to make singlepack operation. The mean dual-pack flow rates
them aware that the arrows on the wing indicate direction were generally around 50 lb/min, with transients seldom
of evacuation, ie aft over trailing edge of the wing. exceeding 75lb/min during dual-pack operation).

Voice and data recorders

During the descent, recorded speech from the copilots
Recorders
microphone continued to be fragmented and was
The aircraft was equipped with two identical Digital described by ATC during their communications with
Voice and Data Recorders (DVDR), each recording the aircraft as quite broken. Recorded speech from
flight and cockpit voice data. The voice recordings the commander was also fragmented, and at times could

Crown copyright 2010 25

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

be heard on the area microphone but not through his Door retention and opening (Figure 5)
oxygen mask microphone. Intercom communication
The overwing exit door is retained at its lower edge by
was also affected and the cabin crew had great difficulty
spigots which engage in recesses in the bottom edge of
understanding the flight crew. On occasions, the cockpit
the door aperture. Its top edge incorporates a locking
door had to be opened for face-to-face communication.
mechanism operated by a handle at the top of the door,
covered by a removable panel secured by Velcro strips.
The aircraft landed 20 minutes after the flight crew
Pulling the operating handle disengages the lock at the
first identified the smell. The recordings stopped when
top of the door and allows the door to pivot inwards
electrical power was lost after engine shutdown, so the
about its lower edge. The spigots remain engaged until
evacuation sequence was not recorded.
the door has pivoted inwards sufficiently for its top edge
Aircraft examination to clear the aperture, after which it is lifted clear of the

Right overwing emergency exit aperture using a fixed handle near its base to support its
weight. The door must then be thrown out of the aperture
The right overwing emergency exit door was unlatched, so that it does not cause an obstruction in the cabin to
but the forward upper corner of the door trim panel evacuating passengers. A compressible rubber bumper
was partially jammed behind the outer edge of the block limits the vertical displacement of the door during
ceilingedge panel (Figure 2), preventing the exit from the initial phase of opening.
being opened.

Ceiling
edge panel

Door trim panel

jammed behind
ceiling panel

Door trim
panel

Figure 5
Overwing emergency exit opening, showing location of jam

Crown copyright 2010 26

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

Door opening clearances Aircraft certification aspects

Although the edge of the ceiling panel was cut back The Embraer 190 and its later derivative model the
around the top edge of the overwing exit door, the Embraer 195 were both certificated by EASA, the latter
resulting clearance between the door trim and ceiling in July 2006. According to the aircraft manufacturer,
edge panel was insufficient. Measurements of the right the Embraer 195 was largely certified on the basis of its
overwing exit showed that over most of its length the similarity to the Embraer 190; this approach was adopted
clearance was just sufficient to accommodate insertion for the overwing exits. However, during Embraer 195
of a credit card, but near the forward corner of the door, development, the ceiling edge panel manufacturer
where the door trim had jammed, the clearance was only introduced changes to the configuration and dimensions
0.003 inch. of the cut-outs around the overwing exit aperture,
reducing the clearance between the ceiling panel and
Prior to this investigation, no clearance was specified at
the door trim. These changes were not notified to the
any location on or around the overwing exit door. After
aircraft manufacturer.
being alerted of this incident by the AAIB, the aircraft
manufacturer issued Service Bulletin (SB) 190-25-0092. The current aircraft certification requirements for
This required an inspection of the clearance between overwing exits primarily address the issues of capacity,
the overwing exit door trim and the ceiling edge panel, positioning, size and profile, but not that of potential
and replacement of the latter if the clearance was less jamming, except that there must be provisions
than 2mm. Additionally, a check was introduced during
aircraft production to verify a minimum clearance of to minimise the probability of jamming
2mm between the door trim panel and the ceiling edge of emergency exits resulting from fuselage
panel. deformation in a minor crash landing.

The efficacy of SB 190-25-0092 was subsequently Source of the smoke and fumes
assessed by the AAIB, with a representative from the Background
aircraft manufacturer in attendance. This assessment
was made on another aircraft from the operators fleet At the time of this incident, only the No 1 air conditioning
on which the SB had just been implemented, with the pack was operative. The No 2 pack had been declared
rubber bumper at the top of the door correctly adjusted. unserviceable after an investigation by the operator into
It was found that the specified 2 mm clearance was the cause of a separate smoke in the cabin event that had
insufficient to prevent the door liner from becoming occurred four days previously. It was established that
jammed behind the ceiling edge panel if the door was the No 2 ACM rotor had seized. Examination of the
lifted during the initial stages of opening, or if it was No 1 pack ACM following this incident revealed that its
opened energetically, such as might be the case in an rotor had also seized. It was later confirmed that both
actual emergency. It was concluded that whilst the SB ACMs had suffered Stage 2 turbine blade failures. The
reduced the probability of a jam, the potential for a jam resultant imbalance had resulted in contact between the
had not been eliminated. turbine blade tips and the ACM casings, producing hot,
finely divided, metallic particles that were released into

Crown copyright 2010 27

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

the cabin air system, creating the reported symptoms of During single-pack operation, the nominal turbine speed
smoke and fumes inside the aircraft. is predicted to range from 42,500 RPM (25,000ft climb,

ACM failure investigation standard conditions) to a maximum of 51,100 RPM

(sea level climb, hot conditions), with an absolute
Both ACMs were returned to the manufacturer for
maximum, taking into account sensor tolerances, of
disassembly and preliminary examination; the failed
52,100 RPM.
Stage 2 turbine wheels were then returned to the
AAIB for independent metallurgical investigation. The
The manufacturer stated that an analysis of ACM
manufacturer established that both units had suffered
removals suggested no relation between ACM failure
turbine blade fatigue failures close to the blade root in a
(of any type) and single-pack operation. Following this
location of high stresses associated with a known failure
incident the aircraft manufacturer conducted a reliability
mode caused by turbine blade resonance.
analysis of the ACM, concluding that a reduction in
the current singlepack MEL operating period limit of
The independent metallurgical examination confirmed
10days was not warranted.
this finding. No evidence of any fatigue initiating
features was found near the crack origins.
A modification to reduce the probability of Stage 2
Previous ACM turbine failures turbine blade resonance, introducing a new Stage 2
turbine nozzle design with an increased vane count to
Previous failures of the Stage 2 turbine have occurred
move the blade pass frequency outside the critical range,
and were attributed by the ACM manufacturer to
was being developed when this incident occurred.
fatigue failure caused by blade resonance resulting
from an overspeed condition. Of those turbine failures Crew oxygen masks
investigated, 40% of the cases were found to have been
Overview
caused by component or control system failures that
could cause an overspeed. In the remaining 60% of The crew oxygen masks are equipped with selector
cases, no reason for an overspeed, or any other cause of valves which give the option of mixed (air/oxygen),
the fatigue failure, was found. 100% (oxygen) and force-feed (purge) modes of
supply.
Metallurgical examination by the manufacturer of a
turbine failure which occurred in 2005, after 1,279hrs The microphone system installed in the masks
and 868 cycles, established that one blade had separated incorporates a cut-out device that electrically isolates
from the wheel as a result of a fatigue crack, and a further the microphone during the inhalation phase of breathing,
two blades exhibited partial fatigue cracks. This mode of and reconnects it again during exhalation. This is to
failure was very similar to that of the failed turbine from prevent the wind-rush sound caused by the in-flow of
the No 2 ACM on G-FBEH. The positions of the crack air/oxygen across the microphone.
origins corresponded with a known location of high
stresses induced by full-blade third-mode resonance, The cut-out device comprises a small plastic balance
which the manufacturer stated occurs at 51,574 RPM beam supported on trunnion bearings in the manner of
+/3% (50,072 RPM to 53,121 RPM). a seesaw, carrying a magnet that moves in proximity

Crown copyright 2010 28

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

to a reed switch mounted beside it. The balance beam Oxygen mask examination and tests
is positioned in the gas path and is biased towards
The captains and co-pilots masks from G-FBEH, and
the microphone live position by residual attraction
the defective captains mask from the other aircraft
between the magnet and an adjacent screw head. An
were tested and strip-examined at the manufacturers
asymmetry in the area presented to the gas flow on either
facility in the United States, under AAIB supervision.
side of the pivot creates a net force on the beam, tending
The captains mask from G-FBEH was found to be
to tilt it towards the cut-out position in opposition to
non-functional and could not be tested. The copilots
magnetic bias-force. At inflow velocities below a
mask passed all of the test criteria. The other captains
certain threshold, ie during exhalation, the magnetic bias
mask operated intermittently, displaying the same
moves the beam back to its original position, restoring
characteristics as seen during the on-aircraft checks.
microphone function.

When demonstrated by someone who routinely

On-aircraft checks
performed the production acceptance tests, the
Checks of the crew oxygen mask microphones on microphone on a serviceable mask produced clear
G-FBEH suggested that the captains microphone was speech with the oxygen flow setting in all modes.
defective, but it could be made to operate by lightly However, when tried by people less familiar with mask
tapping the face of the microphone casing. operation, the audio output in the purge flow mode was
garbled. With practice, once accustomed to speaking
Similar checks of the crew oxygen masks were performed
against the (significant) positive gas pressure in this
on another of the operators aircraft. The microphone
mode, good clarity of speech was achieved. The
on the captains mask, like that on G-FBEH, was also
tendency to produce garbled output when set to purge
initially inoperative, but became live after the mouthpiece
was evidently a feature of the system that required
was tapped sharply. A consistent pattern of malfunction
practice to overcome. The operator of G-FBEH was
was observed: during inhalation, the cut-out system
advised of this finding.
(correctly) isolated the microphone and, thereafter, it
remained isolated during the exhalation phase. Tapping Strip-examination
the mouthpiece then restored microphone function until Strip-examination of the microphone and cut-out
the cut-out mechanism isolated it again during the next assembly from the captains mask from G-FBEH
inhalation phase. revealed that the magnet was fouling slightly against
the side of the cut-out switch body, causing the balance
During these checks it was noted that with the oxygen
beam to become stuck in the cut-out position. The
supply set to purge mode, the microphone cutout
cause of the foul was the incorrect positioning of the
mechanism tended to hunt between live and cut-out
reed switch body.
modes during speech, producing a sound similar to
the garbled radio transmissions heard from the aircraft Disassembly of the captains mask from the other aircraft
during the incident. identified a spurious whisker projecting from the plastic
housing of the cut-out switch, the free end of which
contacted the underside of the flow sensor pivot. The

Crown copyright 2010 29

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

whisker acted as a ratchet, tending to inhibit movement Additional information

of the sensor vane in the direction required to reactivate Previous evacuation incident
the microphone, whilst leaving its motion in the
direction required to cut out the microphone unaffected. The AAIB investigated an incident on 1 April 2002
Consequently, the mechanism tended to stick in the cut- (EW/C2002/4/1), in which the cabin of a Fokker F28
out position, leaving the microphone open circuit. The filled with smoke. An emergency evacuation was carried
whisker or spurious material appeared to be a curl of out, during which passengers using the overwing exits
the switch casing material (Figure 6), probably created experienced similar problems getting from the wing to
either in the production of the switch itself, or during its the ground. The report stated:
assembly into the mask.
Having climbed out of the cabin, passengers
disembarking from the left overwing exit were
unsure of how to descend from the wing to the
ground. A number congregated on the wing
looking for a way down. Cabin crew eventually
noticed the confusion and urged the passengers
to get off the wing. Some passengers slid or
jumped from the wing tip and leading edge (a
drop of some 7 to 8 feet) instead of sliding off the
wing trailing edge down the extended flaps.

Of the reports three recommendations, one is relevant to

Figure 6 the incident involving G-FBEH:

Whisker of plastic material on cut-out switch

Safety Recommendation 2002-42
In the light of these findings, the mask manufacturer
The CAA and the JAA should review the
undertook a detailed review of its design and
design, contrast and conspicuity of wing surface
manufacturing processes. This resulted in an improved
markings associated with emergency exits on
physical location of the magnet at its attachment to
Public Transport aircraft, with the aim of ensuring
the flow sensor vane, the use of adhesive to prevent
that the route to be taken from wing to ground is
movement of the switch body once its position has been
marked unambiguously.
adjusted to provide the required change-point, and the
addition of quality checks to ensure that switch casings The Civil Aviation Authority accepted the
supplied to the company are free of burrs. recommendation, but no response was received from
the Joint Aviation Authority. The responsibility for
aircraft certification within Europe is now held by the
European Aviation Safety Agency (EASA).

Crown copyright 2010 30

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

Analysis Overwing escape route markings

Crew decision making
It is apparent from this incident that the issue of
The commanders decision to divert to the Isle of Man ambiguous overwing escape route markings that resulted
was based on his concern that there might have been a in AAIB Safety Recommendation 2002-42 still exists. It
fire on the aircraft. The sulphurous smell experienced is therefore appropriate that this matter is re-examined.
by both pilots was something that they had never As responsibility for aircraft certification now lies with
encountered on an aircraft before, but one which they the EASA, the previous Safety Recommendation is
uniquely associated with burning. Having made the therefore re-issued as follows:
decision to divert, the commander had limited time in
Safety Recommendation 2010-007
which to achieve a straight-in approach and landing.
This task was made more difficult by the communication It is recommended that the European Aviation Safety
difficulties experienced once the pilots had donned their Agency review the design, contrast and conspicuity of
oxygen masks. Consequently, the commander omitted to wing surface markings associated with emergency exits
inform the SCCM that they were diverting and it also led on Public Transport aircraft, with the aim of ensuring
to his decision not to attempt to speak to the passengers that the route to be taken from wing to ground is marked
over the PA. unambiguously.

The fluctuating intensity of the smell meant that the Overwing exit jam
commander did not decide to perform an emergency
The jamming of the right overwing exit door occurred
evacuation until late in the flight. His intentions were
because of insufficient clearance between the top edge
not communicated to the cabin crew and passengers
of the door trim and the ceiling edge panel. To prevent
and they were therefore surprised by the command to
fouling at this location, adequate clearance must be
evacuate. However, despite the unexpected nature of the
available in the initial stages of door movement until
order to evacuate, this did not delay its commencement.
the door trim panel has passed fully beyond the ceiling
Door 1R & 1L configurations & passenger flow issues panel. In the case of the right overwing exit on G-FBEH,
there was effectively no clearance, such that the exit
None of the passengers evacuated the aircraft via
immediately jammed on attempting to open it.
D1R. This, it is considered, was influenced by the
staggered layout of the front two emergency exits. In
The AAIB checks demonstrated that, whilst improving
addition, there was only one crew member situated in
the situation, the 2 mm minimum clearance specified
this part of the cabin to direct and assist passengers
in SB 190-25-0092 was insufficient to prevent the door
during the evacuation and he was standing next to
liner from fouling the ceiling edge panel if the door
D1L. Passengers would have therefore had to find
was lifted firmly as it was unlocked, or if the door was
and use D1R at their own initiative.
jerked open, as might occur in an emergency. The
2 mm clearance requirement is not entirely effective
in eliminating the possibility of a jam. The following
Safety Recommendation is therefore made:

Crown copyright 2010 31

AAIB Bulletin: 6/2010 G-FBEH EW/C2008/08/01

Safety Recommendation 2010-008 The failure of the Stage 2 turbine on the No 1 ACM
occurred after only four days out of the 10 days of
It is recommended that Embraer modify the overwing singlepack operation permitted by the MEL. This
emergency exits on Embraer195 aircraft, to eliminate suggests that the turbine speed had encroached into the
the possibility of the exit door jamming due to resonance range during this period. It is possible that other
interference between the door trim panel and the ceiling units could be similarly vulnerable during singlepack
edge panel. operation. However, the aircraft manufacturer stated
that this event was the only known case of the failure of
ACM turbine failures
an ACM Stage 2 turbine during single-pack operation
Examinations of the failed turbine wheels from on the Embraer 190/195 fleet. They also reported that
G-FBEH showed that they had failed due to fatigue the reliability of the air conditioning pack had been
cracking originating in a location of high stresses significantly improved through various modifications
associated with a known blade resonance condition. A and maintenance actions, significantly reducing the
new Stage2 turbine housing was under development to probability of Stage 2 turbine failures. Therefore no
address the problem. Safety Recommendation is considered necessary.

Crown copyright 2010 32