Lyman:
They found spherical globules of Stainless Steel in the Combusted products within the identified cell, number 6. That means 2700 degrees, and most of Exponents' text refer to 1100 degrees as typical in thermal runaway.....
It is quite possible for there to be very localized heating at the ends of a high current arc without the entire surrounding area being at the same temperature.
The predicted temperature resulting from a thermal runaway is based entirely on the exothermic reactions at the anode and cathode and do not take into account either external power input or electrical heating resulting from internal or external short circuits that may occur during the process.
Is it reasonable to model the effects of two individually improbable simultaneous failure mechanisms (thermal runaway and short circuit)? Absolutely given that either one could induce the other!
Some comments were made about the inability of a single current sensor to detect a cell to case short circuit. In the Boston case at least, there is no evidence to suggest that the primary failure was a cell to case short circuit. Instead it appears to have been an internal short circuit of one cell.
In that case the + and - lead current of the series string would still be identical and the overvoltage applied to the non-shorted cells would affect cells on either side of the failed cell.
In the Narita case, the public information is not as clear as to just when in the sequence the cell to case short circuit took place.
Regarding the apparent damage to the electrodes of the main battery, I still feel that not enough attention has been publicly directed to the large amount (~10%) of volume expansion to be expected for the entire electrode stack between low voltage cutoff and high voltage cutoff charge levels. This is simply the result of chemical changes to the active material at the electrolyte surface and is in addition to any thermal expansion or contraction. The described problems leading to repeated over-discharge during ground operations has probably led to a greatly increased amount of cycling expansion and contraction compared to the predicted service life profile. Any additional complications from charge control anomalies and thermal expansion/contraction would only add to that.