Battery tests and experimenter’s regress (Part II)


Battery post fire (Image source: NTSB)

The NTSB has released it’s interim report on the Boeing 787 JAL battery fire and it appears that Boeing’s initial safety assessment had concluded that the only way in which a battery fire would eventuate was through overcharging.Therefore, so the logic went, Boeing needed only to prevent overcharging and it also prevented the possibility of a cell fire.

In a classic example of experimenter’s regress when Boeing engineers subsequently planned their type test program they weren’t explicitly looking for other causal factors for a battery fire because their theory informed them that only battery overcharging might cause a fire (1). And because their test program was not looking for evidence of other causes they didn’t find them, rather unfortunately as it turns out.

In contrast to Boeing’s assessment a 2011 study into the hazards of lithium batteries sponsored by the NFPA makes it quite clear that the risk of a thermal runaway resulting in a fire increases dramatically in relation to the state of charge, independent of the cause of the runaway (2). In fact they found that at 100% charge fire was quite common as the casing of the overheating cell would exceed the auto-ignition temperature of the gases.

The NFPA study also found that the heat transfer environment of a cell in thermal runaway can play a large role in the severity of the reaction. High temperatures or insulation increases the likelihood that any given internal fault can drive a cell to thermal runaway, and also increases the energy available to heat the cell to the auto-ignition temperature.

Boeing’s 787-8 electrical power system safety assessment also included an analysis of lithium-ion battery failure modes. This analysis determined that overcharging was the only known failure mode that could result in cell venting with fire.

NTSB report, DCA13IA037 dated 7 January 2013

So contrary to the Boeing safety assessment, whether a fire occurs depends on battery and environmental states, it doesn’t however depend on the cause of the thermal runaway as Boeing assumed. What the Boeing safety engineers appear to have done is confused the observation that most thermal runaways resulting in fires occur at or near a state of full charge with an act of overcharging, and in the final analysis this may have been simply due to the engineers misinterpreting the literature (3).


1.  For example while Boeing engineers did perform a nail puncture test on one of the battery cells to determine the results, they didn’t recognise, as NFPA investigators did, that this sort of damage is probably only going to generate a low impedance short between current collectors, which was in turn unlikely to cause a thermal runaway especially if the state of cell charge is also low.  As it turns out pinch damage in concert with a full state of charge is actually much more hazardous.

2. The likelihood that a fire will occur depends on the state of charge (most significant), the thermal environment and finally the cell chemistry used.

3. They may have done so because the majority of battery thermal runaways that occur in service do so during, or directly after a charging event. The NFPA investigators theorised that this was because:

  • Lithium dendrite formation occurs during charging; thus, shorting of dendrites that cause extreme localised shorting is more likeley,
  • Charging provides more energy in the cell by raising the state of charge increasing the likelihood of a subsequent thermal runaway, and
  • Charging provides energy to any shorting point within the cell.

5 responses to Battery tests and experimenter’s regress (Part II)

    Mike Flannery 23/04/2013 at 3:28 am

    Hi, Thought you might like to see the BBC News item on the battery problem. Apparently, we’ll never be able to work out why it happened 🙂

    Mike Flannery


      Matthew Squair 23/04/2013 at 1:23 pm

      Yep, apparently. The FAA press release is quite tight lipped about it. But the Boeing press conference sheds a little more light.

      I find it very interesting that they are still addressing the possibility of a battery fire through inclusion of the battery enclosure and vent line, working on the consequence side of the hazard equation is very very unusual for aviation. So clearly they can’t convince themselves that they can eliminate the causes and therefore likelihood of a battery fire.

      Reading between the lines, “we didn’t realise there was more than one cause for a battery thermal runaway…”. 🙂


    Mike Flannery 24/04/2013 at 1:16 am

    I’m not a ‘techie’ in this area, but haven’t we been using battery combos for some time in other models without any problems? Or is the juice requirement significantly higher for the 787 than in any other vehicle thereby significantly increasing the risk as well as the output?


    Matthew Squair 24/04/2013 at 3:09 pm

    All batteries can suffer from thermal runaway effects, but lithium batteries have a much greater energy density than lead acids (for example) so there’s more energy available to drive the effect. They’re also quite sensitive to over-charging, and need to be managed carefully in the charge cycle. Finally if they do overheat the electrolyte is quite flammable.


    Mike Flannery 24/04/2013 at 8:02 pm

    Thanks for the explanation. Would I be considered a ‘leaditte’ if I thought we should continue with the old methods until we are more sure of the new? 🙂