NTSB releases preliminary report on JAL 787 incident

The factual findings are here. 48 PDF pages. The NTSB preamble to the Interim Report:

The National Transportation Safety Board (NTSB) notes that the information discussed in this interim factual report is based on initial findings from the investigation of this incident. Because the investigation is continuing , no conclusions or recommendations are being made at this time. Readers are encouraged to access the public docket for this incident (DCA13IA037) for further details about the information presented in this report.

In addition, readers are advised that the information presented in this report could change if new evidence becomes available.

The Seattle Times reports the NTSB plans public hearings on lithium-ion battery safety.

Reuters reports that the NTSB has not found the root cause of the Japan Air Lines incident.

EADS CEO Tom Enders weighs in.


28 Comments on “NTSB releases preliminary report on JAL 787 incident

  1. – “I think the certification authorities, whether it’s the FAA or any other, are probably a little bit nervous about these new planes now coming in, about the materials and the systems and the processes,” said EADS chief executive Tom Enders.

    – “This is why we have refrained from… any schadenfreude about the problems in the 787, because we have had similar problems in the past,” Enders said in a breakfast meeting with reporters in New York.

    – “If industry runs into trouble, particularly as certification is concerned, that affects not just one manufacturer, but others as well,” Enders said.

    – Enders said one lesson from the Dreamliner’s problems is that there may be a benefit to upgrading existing aircraft rather than embarking on entirely new designs.

    – “It’s not completely risk-free” to modify existing designs, Enders said. “But it’s of course much less risky than developing new aircraft and it comes for the fraction of the cost.”

    Sounds like an A330neo in in the works!


  2. 1.9 Additional Information
    To understand the potential impact of smoke generation resulting from a battery failure in the cabin and cockpit environment, the NTSB documented the environmental control system on the incident airplane. This system provides pressurized and heated or cooled air to the 787 passenger cabin and E/E bays and enables smoke removal.
    The aft E/E bay contains two smoke detectors. When smoke is detected, the avionics cooling function is designed to exhaust smoke overboard through fans in the cooling ducts and changing supply valve positions (and the use of differential pressure if the airplane is in flight). During this incident, the supply valves (which are electrically driven) lost electrical power after the APU shut down because the APU was the only source of electrical power being used at the time. As a result, smoke generated by the APU battery could not be effectively redirected outside the cabin and aft E/E bay.

    and a bit earlier

    Overall compliance with applicable 787 main and APU lithium-ion battery safety requirements was shown through formal analyses and tests. In addition to the Boeing analysis and tests noted previously, analyses and tests were performed by Thales and GS Yuasa, which were reviewed by Boeing project engineers, Boeing safety reliability and maintainability engineers, and Boeing authorized representatives. Formal analyses included a battery functional hazard assessment, fault tree analysis, and failure mode and effects analysis as well as a battery and battery charger system safety assessment. Battery testing consisted of full-performance, environmental qualification, and destructive tests. The destructive tests included external short circuit (low and moderate impedance shorts at battery terminals), overcharge (charge battery at 36 volts for 25 hours), high-temperature storage (185º F for 18 hours), and overdischarge (discharge battery to zero volts) tests. Boeing indicated that the tests found no evidence of cell-to-cell propagation failure or fire.

    RE fault tree analysis –

    Looks like someone was still hanging from the tree !!

    So if the apu battery goes out on the ground- the apu stops- and there is no power to the vent valves and fans

    And we put the LI battery in an aluminum box- so if things go wrong- we cannot use water

    and put the monitoring circuits and memory in the box ?

    And make it so one has to remove kick shield to get to quit release of battery /

    Somebody must have done a cost benefit analysis and decided that since no problems should happen on the ground- we need not be concerned. ?

    Well thats what an MBA is for !!

    …After additional firefighting efforts and the placement of a ventilation fan by the E/E bay door to clear smoke,21 the incident commander made the decision to remove the APU battery.22 Firefighters reported that removing the battery was difficult because a metal kick shield installed in front of the battery prevented them from accessing the battery’s quarter-turn quick disconnect knob. Also, the quick disconnect knob could not be turned because it was charred and melted. The airport security camera video showed that the battery was removed from the aft E/E bay at 1157:20,23 about 80 minutes after the initial notification of the event. The ARFF incident report showed that the event was “controlled” at 1219, about 1 hour 40 minutes after the initial notification. The National Transportation Safety Board’s (NTSB) investigation of the event began later that day.

    • “the supply valves (which are electrically driven) lost electrical power after the APU shut down because the APU was the only source of electrical power being used at the time.”

      It does seem like the guys on a-net who stated smoke evacuation was supposed to be effective only through air pressure – and therefore only in flight – were a bit off the mark. Since the system was designed (apparently) to evacuate on the ground as well, then having no venting backup from the main battery, for example, does seem like a major boo-boo!

      • But then it’s all academic anyway, as the venting will change as part of Boeing’s plan…



    1 Mar 07, 2013 Interim Factual Report 03-07-2013 48 0
    2 Mar 07, 2013 Materials Laboratory Factual Report 13-013 126 117
    3 Mar 07, 2013 System Safety and Certification Group Chairman’s Factual Report 21 0
    4 Feb 25, 2013 Airports/ARFF 16 – Factual Report of Airport Emergency Response Group Chairman 7 0
    5 Feb 12, 2013 Airports/ARFF 16 – Attachment 1: BOS Airport Diagram 2 0
    6 Feb 12, 2013 Airports/ARFF 16 – Attachment 2: BOS ARFF Fire Alarm Log 3 0
    7 Feb 12, 2013 Airports/ARFF 16 – Attachment 3: BOS ARFF Incident Report 5 0
    8 Feb 12, 2013 Airports/ARFF 16 – Attachment 4: BOS ARFF Interviews 18 0
    9 Feb 12, 2013 Airports/ARFF 16 – Attachment 5: BOS ARFF Response Route 2 0
    10 Feb 12, 2013 Airports/ARFF 16 – Attachment 6: JAL Me


    • I could download the prelim report ~12h ago. At the moment the download links error out.

    • “Dreams on dreams are flying over
      from creation to decay
      makes the traveler freeze and shiver
      arcing, sparking, burnt away.”

    • Interesting that they conclude with the possibility of the emergency lighting Li-ion battery also catching light since it’s located above the APU battery. That’s a new twist we weren’t informed of before…

      • Lighting? no.
        “Emergency power for Flight Controls”

        see ( seattle times, D.Gates, link from previous post) :
        “And sitting on a rack above the battery that burned was a smaller lithium-ion battery, also supplied by Japanese manufacturer GS Yuasa, that is used to provide emergency power for the jet’s flight controls for 10 minutes or more “when no other electrical power is available.”
        Investigators found the exterior of this battery had been “lightly scorched” by the fire below and noted its case had openings at the corners.”

        So there is at least a third Li-Ion battery around that Boeing never told about. ( and all eggs in a single basket ?)

        • Oops. Misread. So it’s more serious than I realised…

      • I’m no system engineer, but can imagine I wouldn’t preferably locate this third emergency (Li-Ion) battery on top of the big one.

      • Airbus seems to have all 4 identical batteries in the same compartment. But their setup seems to be significantly different. ( looks more like a 4 fold symmetric design.)

  4. So, having digested this, does it affect people’s views on when the 787 will start carrying passengers again?

    • The information provided has plushed out what we knew but afaics it brought no extension of scope, no new explanation or direction.
      My impression is that the NTSB does not feel like expanding their investigation out into the general electric system setup on the Dreamliner.

      Investigating the certification process is an FAA internal activity which hasn’t even shown a leg yet.

      Looks like for prediction one has to switch to the Politic Crystal Ball.

      Don’s injected information on Simulator closedown and move to Miami ( why Miami? is the current site the sole source for simulator training? )
      would show that Boeing expects no pilot preparation for deliveries will be required in the next ~12 month?

    • The majority of the financial analysts and most of the media are still in denial. In my book Q4 2013 is an optimistic scenario.

      Time to pack one or two ‘golden parachutes’ and go looking for a ‘saviour’ who can reconstruct Boeing, or shall we say muck out Boeing?

  5. Reply to Normand #1 – It is consistent with continued speculation that A will drop the A358.

    • Yes, you are right Chris. And the cancellation of the -8 would open the door to the A330neo. That would be in line with Enders current thinking towards lowering the risk. Considering the lack of interest for the A358, to launch it at this time would indeed be risky, if not foolish.

  6. Several odd anomalies are apparent in the NTSB’s interim report, and they raise a number of questions. To visualize the chain of problems, I constructed a timeline of events from descriptions in the text and in the tabulation of timed events shown in Table 2 of the report. The most significant item is the very high input current flow stated to have been delivered at around 40 to 45 amperes for about four seconds; this occurred about 3 seconds after a voltage drop from 32 V to 31 V was recorded.

    Presumably, this high battery input current came from the battery charger which was responding to the initial one volt drop in battery voltage that perhaps was caused by a single cell (in the series string of eight cells) that had begun to fail due to some as yet unknown cause. But if so, why wasn’t the charger immediately shut off? Wouldn’t it be helpful to know the level of any input current to the battery from the time the APU was initially turned on?

    After the initial one-second drop from 32 V to 31 V, the battery voltage continued to drop for two more seconds, reaching 29 V, then rebounded to 31 V one second later, this being about ten seconds after that sudden downward drop from 32 V first began. But by this time much cell-overcharge damage probably already had been done. Unfortunately, the report does not specify the charging current before or after those critical four seconds at 40 to 45 A until 28 seconds later, when it states that the current was no longer going into the battery, but was actually flowing out of it, discharging at rates spiking from zero to around 5 A.

    This occurred in three rapid cycles within roughly a single second about 36 seconds after the initial downward voltage drop. Also, while the current was rapidly spiking up and down those three times, the voltage was also spiking three times from zero to 28 V, with the 28 volt peaks causing a 5 A peak output current. Upon the final drop to zero volts, the APU controller went offline and the APU shut down. But about 22 seconds after the initial drop in voltage, there was a cockpit-alert report that the main battery was discharging, and 53 seconds later, there was a report that “the main battery power switch was off.” I presume these reports may not be precicely real-time.

    The interim report indicates that the wire used to ground the battery case showed evidence of having carried a high current while the battery was failing. Is there is no provision for shutting down the battery charger and disconnecting the battery output in the emergency event of a ground fault? If the main battery switch is “off”, how can the main battery be discharging? Is it possible there is some miswiring or an unknown “sneak” circuit? Why doesn’t the report provide details of the actions of the battery charger which apparently kept inputting a high charging current even as the battery voltage kept decreasing?

  7. High unchecked inrush current: could that be an effect of the patented charging approach and a ~4s ( or longer) cycle time of programmatic plausibility testing ?

    From an undetermined time onwards the 2 PCB / 4 functional units sensor/supervisory circuitry inclusive all range of internal connectors were inundated with hot foamy and definitely conductive electrolyte.
    Never designed for and thus expectable erratic behaviour/indication.

    The busbars are also “naked”. It is anyones guess where any current flowed.

    Would be nice to know when the first cell blew.

    A significant amount of current must have arced from cell 5 body to battery enclosure.

    • Figure 9 shows the outside of the battery case opposite cells 5,6,7 and 8. Figure 11 shows the remains of those same 4 cells after their side of the case (side 4) was peeled down. The pressure relief openings of cells 6 and 7 are completely blown open, and that of cell 5 appears to be partly open. Cell 8 did not blow open and maybe was the only cell to survive..

      From the (Fig. 9) appearance of the battery case outside those 3 cells, it seems they were acting like 3 plasma blow torches directed against the aluminum case, particularly with a hot fire from cell 7 blasting out against the case. Wouldn’t such hot conductive gases cause current to flow from what remained of those 3 batteries into the grounded aluminum battery case? Was either cell 6 or 7 the first to blow?

      The cover of the battery box appears to have been bulged upward due to the hot (conductive?) gasses within, and currents could have flowed from the uninsulated surfaces of bus bars to the underside of the box cover or even across the cells themselves, adding to the short-circuit conditions. What a mess!

      • FWIW – I think part of the deformation of the box was caused by the firepersons in removing and/or opening the box-

        Boeing apparently never figured on cascade or domino effect and claimed the only thing to worry about was overcharging.

        And as I mentioned in another thread – The testing of charge/discharge was done with simple resistor banks – NOT an apu starter generator simulator ( or actual )

        Why didn;t this show up before ? well in my armchair ex- spurt – tease – during test flights, they probably did not do a series of takeoff – fly for x hours, land, one hour turnaround and repeat, and repeat and repeat . . .

        In addition the variable charge discharge cyles/voltages noted **MAY** repeat *MAY** be due to the starting cycle of the apu such that at xx rpm, the starter generator reverts or also becomes a generator/charger- before APU reaches stable rpm.

        For example- MY RV old Koheler 110 VAC and small dc generator uses a motor-generator to start from 12 volts. BUT the 110 v AC does NOT come on, line till several ( 10 ? ) seconds AFTER reaching stable speed. AT which time AC comes on, and probably DC charging comes on.

        leading to the question – does the 787 APU starter generators have a xx second delay before DC/AC power comes on line AFTER stable RPM ?

        IF not- WHY NOT ??

        • Very good points, Don. An inductive load such as a motor can be quite different from a purely resistive load, such as an old-fashioned incandescent light bulb. And a dc motor can become a dc generator feeding a charging current back into the starter-battery if the engine it is trying to start suddenly revs up while the starter-motor is still mechanically engaged and still connected to the battery.

          But if this were a problem, wouldn’t a high-current diode in the line between battery and motor allow current to flow in only one direction, i.e. from battery to starter-motor? And isn’t the output from the APU ac from a separate ac generator that feeds the lithium-battery’s charger?

          Your question about the number of takeoffs and landings during flight testing is also valid. In the case of the Boston battery fire, it may have started when the 787 landed and was moving to its parking slot. I have wondered if a hard landing or vibrations from the engine thrust-reversing could have induced a problem for the materials or interconnections within a cell. Were the batteries tested adequately for shock and vibration survivability?


          • well sure- there are several ways to prevent ‘ backflow’ from motor generators that have been around for years- but were they used ? While there is a lot of info out regarding overall system – there is little public info out about the nitty gritty details. And it is the type of thing that would NOT be discovered by separate SIL labs, and as has been mentioned, NO indication of a systems test to be sure all parts play nice together in the sandbox.

            NO iron bird ever mentioned or AFIK built. Iron birds ARE expensive- but not as expensive as this screwup

            Even so, it is still somewhat surprising the problem did not show up in flight tests.

            While its obvious that something was missed, and the analysis was far off base- it sure makes me wonder how BA can now say ‘ we got a terminating- final- fix ”

            Was a sneak circuit analysis ever done ? the small burnt ground wire from the battery case seems to indicate NOT.

            so what else is out there ???

            Power point rangers still seem to be clueless.

      • The starter/generator on the 787 is a synchronous (3phase?) AC machine. Exitation is via a variable frequence drive unit.
        No idea if that unit works directly of the battery voltage or
        if there is an intermediary converter doing 32V DC to +-250V DC and from there a 4 quadrant VFD doing the variable AC for starting and in generator mode converting AC back to the HV DC bus.

        • OK, I stand corrected — I assumed the 32 V dc from the battery was used directly to start the APU engine using a dc motor, but I see that the APU starter motor runs from ac power synthesized from the 32 V dc battery, and this machine apparently then becomes a generator after the engine startup is completed. Still learning but having forgotten the details, I now copied this from the NTSB report:
          “The SPU converts DC battery power to AC power for starting the APU and provides excitation power for the APU during startup. The SPU includes a connector plug, power output port, and three input power terminals”


Leave a Reply

Your email address will not be published. Required fields are marked *