Originally Posted by Machaca
Here are some details and images of the 787 chargers and batteries:
Originally Posted by rr_ndb
Could you comment something on the graph?
I happen to know the man who took the data, made the graph, and presented it to more than one conference--ballpark a decade ago. I discussed it with him extensively a few hours ago.
The graph subject is heat released during a thermal runaway event, with the actual vertical axis being rate of temperature rise observed in a specialized adiabatic test fixture. In the tests, thermal runaway is obtained by very slowly heating a disconnected battery in the test fixture, it taking a couple of days to reach the beginning of appreciable heat release from the battery. This first phase uses externally supplied heat--but when self-heating is detected the test assembly switches to adiabatic mode (i.e. heat is neither added to nor allowed to escape the battery under test). From that point the test progresses up the temperature scale extremely rapidly, as is hinted by the self-heating rate scale.
Obviously different cathode compositions are being compared here. Comparing the curves by essentially integrating the area under them, one can see that the total energy released per unit volume by the lithium cobalt cathode (as used in the Yuasa batteries of interest here) is quite substantially the highest in the comparison.
However, this has relevance to heat released--and thus possibly collateral damage--
after the battery has already expired, which in all cases occurs tens of degrees cooler than the beginning of the thermal runaway shown. So this graph has nothing to do with why there was failure, though it does have something to do with the aftermath.
The comparison curves were all measured on 18650 cells--a size and form factor often used inside those black brick laptop batteries we have all seen.
One other point he mentioned: in their automotive application, Tesla uses these very small cells in very large numbers--but they are located in a sort of honeycomb structure. One must either design hoping never to have failure (in reality, extremely improbable), or design in a way so that failure causes acceptable harm. It appears Tesla decided that could not count on their lithium cells never failing, so took good care (it involves both physical and electrical considerations) to assure that the likely failure mode of a single cell would not cascade to adjacent cells. The cells they use are small enough that the energy release from a single one should not endanger crew or vehicle.
Looking at the Thales/Boeing design and the FAA special considerations, it appears that the approach was to assure cell failure would never occur--as it seems self-evident that no serious measures to avert propagation to adjacent cells were employed, nor were serious measures to contain damage to nearby systems.
Even if the specific cause of these events is determined and mitigated, I wonder if all concerned will remain convinced that all other possible causes of cell failure are sufficiently unlikely to make this approach prudent.
I'm not a pilot, and my design and reliability careers were not in aviation, but this all reminds me of the pair of risk mitigation approaches found in, I believe, every modern turbofan engine regarding bits of engine flying away and causing harm when they hit something important. For some parts, such as fan blades, the assumption is that they will indeed fly off sometimes, and an adequate shield is required to protect the rest of the airplane (and a spare engine to keep it aloft). Call this Tesla-like. For other parts, such shielding is deemed "impossible" (which really means too heavy and otherwise expensive), so the approach is to assure it will "never" happen. Call this Thales/Boeing-like. It is awkward when "never" happens anyway, as happened with the IPT disc on a Rolls-Royce Trent 900 on Qantas flight 32, and has happened on these two 787 episodes.