Emirates A332 at Karachi on Oct 4th 2014, protective breathing equipment catches fire

Last Update: July 21, 2016 / 15:22:41 GMT/Zulu time

Bookmark this article

Photo of Emirates Airlines A6-EAQ, Airbus A330-200 — Emirates Airlines A6-EAQ, Airbus A330-200 (Photo credit: ╚ DD╔ / Flickr / License: CC by-sa)

Incident Facts

Date of incident
Oct 4, 2014

Classification
Accident

Airline
Emirates Airlines

Flight number
EK-609

Departure
Karachi, Pakistan

Destination
Dubai, United Arab Emirates

Aircraft Registration
A6-EAQ

Aircraft Type
Airbus A330-200

ICAO Type Designator
A332

An Emirates Airbus A330-200, registration A6-EAQ performing flight EK-609 from Karachi (Pakistan) to Dubai (United Arab Emirates) with 68 passengers and 14 crew, was being pushed back from the park position when smoke developed in the cabin prompting flight attendants to don their protective breathing equipment (PBE). However, upon activation of the PBE by one flight attendant this unit self ignited and caught fire causing injuries to the flight attendant. An evacuation via slides was initiated, emergency services responded and put out the fire. The flight attendant was taken to a hospital.

The occurrence aircraft remained on the ground in Karachi until Oct 7th before resuming service.

On Feb 24th 2015 Germany's BFU reported in their October bulletin, that Germany's BFU is supporting the investigating authority to examine the PBE.

On Jul 21st 2016 the United Arab Emirates' GCAA released their final report concluding the probable cause of the occurrence were:

The Air Accident Investigation Sector determines that the causes of the dense mist entering the cabin, and the subsequent PBE fire onboard the EK609 Airbus A330 were:

- The failure of a yellow hydraulic system rudder servo hose that allowed leaking hydraulic fluid to enter the APU where the fluid was heated and atomized and was then fed into the cabin airconditioning system. The cause of the hydraulic hose failure was not determined.

- It is probable that manufacturing defects in the PBE candle caused a vigorous chemical reaction in the candle which resulted in abnormal ignition when the cabin crewmember, who had donned the equipment, pulled the activation lanyard.

Contributing Factor to the Incident

As the flight crew were unable to identify the source of the mist/smoke, they decided to leave the APU running in case it became necessary to shutdown both engines, but they did not close the APU bleed as required by the SMOKE/FUMES/AVNCS SMOKE checklist.

The result of this decision was that hydraulic fluid mist continued to enter the cabin. This decision was taken without having positively identified the sources of the smoke/mist.

The GCAA reported that the aircraft had been pushed back and had reached its push back position, engine #1 was running and engine #2 in the start sequence, the crew was about to request taxi clearance when the crew received a chime that was repeated after 3 seconds, the first officer reported the yellow hydraulic fluid quantity was indicating low. The captain advised to continue the engine start sequence, the quantity low indication would sort itself out. The first officer subsequently advised that the yellow hydraulic pump had gone off and the yellow hydraulic pressure was fluctuating. Soon after the yellow hydraulic pressure fell to zero. About 72 seconds after the first chime the crew observed a "terrible" smoke and the first officer recommended to return to the stand. The flight crew donned their oxygen masks, confirmed the communication was working, the captain checked with the gound engineer whether they could see any smoke outside with a negative reply, then queried with cabin crew who confirmed the presence of smoke in the cabin with passengers becoming "agitated", a lavatory smoke detector triggered causing another chime sound, the commander advised the ground crew they had yellow hydraulic system low quantity and low pressure and a lavatory smoke detector had triggered and the captain requested a return to the stand. The ground crew requested the tug, which had already departed the aircraft, to return to pull the aircraft back to the stand. The flight crew queried the tug several times, the cabin crew advised the situation had become "very, very bad", the captain instructed to shut down both engines but keep the APU running for the return to the stand, and subsequently to shut down the bleed air supply through the APU. The smoke subsequently cleared in the cockpit, the lavatory smoke indication ceased, cabin crew however advised the situation was "really, really bad" and they would need to evacuate. The captain queried how many seat rows the flight attendant could see with reply 4 rows and whether she would consider an evacuation necessary which she affirmed. The captain therefore decided to disembark the passengers. ATC was informed the aircraft would be evacuated, the captain inquired with ground crew whether stairs would be immediately available.

5 minutes 58 seconds after the first detection of smoke a flight attendant called the cockpit reporting a fire near the L3 door coming from a PBE. The PBE had been donned by the related flight attendants but caught fire as the FA activated the PBE, the FA immediately removed the PBE and threw it onto the floor. 4 Halon fire extinguishers were emptied onto the PBE without success "as the fire was self-sustaining due to its continuous production of oxygen". Three seconds after receiving the fire report the captain instructed an emergency evacuation via slides. The crew informed ATC about a fire on board and requested emergency services to be deployed. 8 minutes after the first detection of smoke all doors except L3 were opened with the slides deploying and all persons on board exited the aircraft via the slides resulting in minor injuries to a number of passengers.

Attending fire services entered the aircraft climbing the R1 slide (after removing their shoes) and extinguished the fire at door L3. There was no heat damage to the aircraft except for the FA seat, carpet and floor panels at door L3. An external check of the aircraft revealed leakage of hydraulic fluid from the vertical fin which had flowed around the fuselage and entered the APU inlet.

The GCAA stated: "The Operator’s maintenance crew found that the yellow hydraulic system pressure hose that supplies hydraulic pressure to the rudder yellow actuator, located in the vertical fin, was leaking fluid."

The GCAA reported 7 cabin crew and one passenger received minor injuries as result of the occurrence.

The failed hydraulic hose was dispatched to the aircraft manufacturer for further analysis, the GCAA summarized the findings of the report: "The report concluded that the main failure cause was, most probably, fatigue failure of the metal braiding which caused the failure of the PTFE pipe."

Tests were performed with the PBEs across the operator's fleet. The GCAA summarized the results:

A sample of 59 PBE units installed on the Operator's fleet were tested at the Operator’s facilities on 14 October 2014. The 59 units contained seven units from the same lot as the Incident affected PBE (P/N 119003-11, S/N 003-35283M). These seven units did not exhibit any anomalies.

According to records provided to the Investigation, a total of 580 PBE units (between S/N 003-34983M and 003-35563M, manufactured in June 2007) were equipped with suspect chlorate candles from the same production lot of the Incident PBE. Out of the 580 PBE units, 48 units had been delivered to the Operator.

On 19 October 2014, during tests of 14 additional units from the questionable lot, one PBE made a 'popping' sound and generated black deposits (S/N 003-35280M), which is inconsistent with its normal operation, and another PBE caught fire (S/N 003-35282M).

The three PBEs having shown anomalies, including the one having caught fire in Karachi, one having caught fire during tests at the operator's maintenance facility and the one accumulating black deposits after activation, were exposed to CT scans. The GCAA summarized:

The CT scan showed that the three selected candles displayed evidence of filter disruption and penetration. The sodium chlorate based primary layer had fully reacted, but was largely intact and undisturbed. There were signs of localized discoloration on the surface of the candle stainless steel tubes. These areas discoloration were immediately adjacent to the starter layer consistent with the effects associated with elevated heat. This could not be seen on the candle involved in the Incident because of surface damage due to the fire.

The two PBE units that had been exposed to fire (S/N 003-35283M and 003-35282M), exhibited residues of the core and filter in the oxygen path. The candles also showed evidence of localized outlet elbow burning and material loss in the vicinity of these compacted core/filter materials, or in the preceding flow path. In addition, residue from the perchlorate based starter layer of the candle, that is the region normally occupied by the starter layer, was essentially void and the starter layer residue assumed a displaced position in the direction of the candle outlet.

The GCAA analysed with respect to the hydraulic hose:

The failure of the hose may have occurred for any one, or a combination, of the following reasons:

1. The manufacturing process was not well protected from contamination;
2. The postproduction testing of the hose was not carried out correctly to comply with the design testing standards;
3. The design testing standards were not adequate to cover the spectrum of actual operational conditions, and vulnerability to other conditions, that may not have been anticipated during the development of the testing standards;
4. Handling of the spare part after production, including the storage conditions, was not appropriate; and/or
5. The in-service conditions were beyond the published limitations

The precedence of the braiding failure to the tube failure indicates that the braiding was holding its part of the internal hydraulic pressure. The scratches and the imprint marks left on the tube by the braiding wires indicate that the braiding was properly fitted to the hose assembly. Accordingly, it is believed that the hose assembly step in the manufacturing was carried out properly, but the Investigation could not determine whether or not the fabrication of each component in the hose was according to that specific component-manufacturing standard.

No records were discovered by the Investigation indicating that the hose in-service limitations, in terms of operating pressure, or burst pressure, had been exceeded. There were also no indications on the exterior of the hose that indicated that the bend radius had been exceeded. There was no indication that inappropriate physical force had been applied to the hose, or that the environment was contaminated, or that the fixing clamps were out of place. Therefore, the Investigation believes that the in-service conditions were within the published standards.

Since the part was installed by the Aircraft manufacturer, and the manufacturing standards were properly implemented, the Investigation believes that handling of the hose before and after installation was proper.

The Investigation could not determine the exact cause of the braiding fatigue failure, but the Investigation believes that the hose postproduction testing could not predict the failure at this service life, or was not sufficient to cover other conditions that the hose was subjected to, within the spectrum of the testing parameters. The loss of the identification tag prevented the Investigation from determining the date of production of the hose, and made it impossible to check the documentation related to the post-production inspection.

With respect to the PBE the GCAA analysed:

The report listed factors specific to the processes by which the candle is manufactured and that could directly affect, or accrue in a manner that could influence, an increase in the reaction rate of the starter layer:

1. The purity of each substance used (either pre-existing or introduced contamination)
2. The degree of compression used in the formation of the composition, (i.e. control the application of forces during the core forming/molding process)
3. The thoroughness with which the various constituents have been mixed, (i.e. consistent mixing)
4. The shape and size of the finished piece in which the composition functions.

The Investigation determined that it is highly probable that at least one of the listed production-based defects had led to the uncontrolled reaction in the EK609 PBE S/N 003-35283M and in the PBE S/N 003-35282M that caught fire during the post-Incident test carried out in the Operator's facility.

The GCAA analysed with respect to environmental aspects:

The temporary influence of the mist generated by heating the hydraulic fluid in the airconditioning system on the health of people onboard could not be exactly determined, but the Investigation believes that the level of contaminants in the mist was below the published limitation that could cause drowsy situation. However, the mist concentration still affected the respiratory system causing difficulty in breathing, throat discomfort, and eye irritation to some occupants.

...

The exposure of the passengers onboard the Incident flight to the hydraulic mist, and of some of the passengers to the following PBE fire led them to complain of irritation. However, the exposure did not lead to ill effects to the Central Nervous System (CNS) of which known symptoms are loss of recent memory, poor concentration, increased lethargy, neuromuscular incoordination, confusion, and headaches.

With respect to the situational awareness of the crew the GCAA analysed: "The CVR indicated that the Commander’s situational awareness was reinforced by effective CRM which included continuous updates from the cabin."

The GCAA, a few paragraphs later, however analysed:

A review of the crew communications indicated that there was information flow from the cabin to the cockpit, but that the terminology used by the cabin crew to describe the situation to the flight crew and assist decision-making, was not standard terminology. The use of phrases such as “The situation is very very bad” transfers a sense of urgency and a level of emotion, but lacks information as to details of what is actually happening, as bad could be perceived differently due to personal experience. However, "visibility one meter" or "I can see four seat rows." would have transferred critical information more accurately. The Commander spent valuable time trying to perceive the actual situation in the cabin, whereas there was an assumption that as the Aircraft was on the ground near the terminal buildings, it could quickly be returned to the stand, or that an evacuation of the passengers and crew could be accomplished within seconds.

Although the terminology used by the cabin crew was not standard, the flight crew could have better interrogated the cabin crewmembers and better interpreted the communication of information from the cabin. The flight crew heard the words from the outset “Very thick smoke” stated in an agitated tone of voice. Seventy four seconds later the cabin crew advised the flight crew that: “The situation is very, very bad”, and then a further communication from the cabin stated; “Cannot breathe…” All of these statements, which convey a sense of urgency or distress, were transmitted over a period of about three minutes, but it took the Commander a further four minutes to decide to order an evacuation of the Aircraft. The Investigation believes that the time spent in deciding to order the evacuation could have shortened if more weight had been given to key words and phrases used by the cabin crewmembers.

With respect to the airport's response the GCAA analysed:

When the Commander announced the passenger disembarkation initially, and then that he intended to evacuation the Aircraft using the escape chutes, the situation was not treated with an appropriate level of urgency by ATC. Other aircraft and vehicles continued to maneuver close to the EK609 Aircraft. No effective action was taken by ATC to assist in expediting the disembarkation of passengers and crew, or to protect the disembarked passengers by stopping movements on the ramp.

Although the evacuation took place without any unsafe consequences, the Investigation believes that a hazardous situation was created when ATC allowed other aircraft in the vicinity of EK609 to continue taxiing as the Aircraft was being evacuated. ATC could have managed the emergency by observing and directing other aircraft in such a way as to facilitate a safe evacuation, and minimize any potential hazards on the ramp.