View Single Post
Old 27th Jan 2013, 04:00
  #182 (permalink)  
Join Date: Aug 2011
Location: hong kong
Age: 60
Posts: 93

Thank you for the great report. It analyzes pretty much all single cell failure scenarios.

Here is what I am using, covering multi cells:

Did you see this approach used before? How many amps to be shunted? Voltage control loading a charging (or charged cell) through by pass?
Common practice with larger (Pb) batteries and also Li batteries. Not encountered it with NiCd due to negative cell voltage temperature coefficient.

See above application note by TI.

The ANA main batt. geometry mismatch was likely due what kind of abuse? During charging cycles?
Picture is post event, can't tell pre-event, but there are some clues.

The event:
Labeling cells left to right, top to bottom.

  1. Cell #3 failed first, in a manner as outlined earlier, it shows "burned out" plates and also greater warping (thermal distress) around it
  2. cells #2, #6 and #8 show substantial signs of thermal runaway
  3. Cells #1, #5 and #7 show variations between plates and lesser thermal distress and over pressure. **
  4. Cells #1, #5 and #4 closest to #3 show less damage than #2, #6 and #8 **
  5. Conclude cell #3 has not thermally triggered thermal runaway of many/any?? other cells. **
  6. Conclude cells except #3 failed by electrically induced thermal runaway, triggered by over voltage after failure (short) of #3 and remaining powered by the bus
  7. Conclusion: Battery was kept at near 32V after #3 failed.
** But, it could be opposite, less bulging due to valves opening earlier.

Again, there are two independent failures, 1. and 7. Both have to be fixed.
7. seems to be the greater concern though, as it failed seven (7) very likely functional cells and multiplied the violence of the event.

Where is the source of information that the the battery is charged by the bus. Ive heard of a diode.

The direct connection of a Li Ion stack of cells to a bus seems very dangerous. A much safer approach would be:
Battery is connected to and will be charged via the bus. There is no 60A charger on those PCB's. There is also no 150A+ diode handling APU starter. Thus, the bus voltage must vary with battery voltage.

And, bus voltage must listen to charger, how that works together with bus tying I don't know. On failure (of #3) the charger should disconnect the battery from bus.

How the battery is protected from bus is the golden question. It seems it was not. If it was not, pre-event cell deterioration may also be caused by it.

With these thin wires seems just impossible to control cell voltages during the charging. A likely scenario (during charging) JAL BOS fire.
No, that's just fine. The bypass current would cover only differences in leakage and possibly capacity and be up to only 1-5 % of charging current. Less than 3A. 10s of mA typical. Once leakage, or imbalance goes beyond a few %, cell has to be replaced.

@ Turin
Thanks for info. One main and one APU battery failed. If not caused by bus, it is must be the charger continuing to power the battery.

@ Turin
Any info on bus tying. Can APU bus connect to main bus?

Edit:I keep wondering about why the four cells on the left show a pattern different from the four cells on the right. It seems like that plates have slightly moved apart, or have changed their properties somehow.

Why only these four, and could this change be pre-event, and be related to the failure of #3?

Can this pattern be observed on other in-service batteries?

Wiring errors would be a very convenient explanation of 1.
Without being aware of bypass facility, I discounted wiring errors earlier. Now, wiring errors could lead to the wrong cells being bypassed and to over voltage on the others.

Wiring errors could explain the pattern (if it is deterioration) in the cells on the left and failure of #3. Did they swap left and right monitoring or bypass wiring?

Or, is there another problem managing half the cells?

But, it still would leave 7. to explain and fix. My analysis of 7. could be wrong like everything else. While I doubt it, and stick to electrically induced thermal runaway, all other cells could also have been destroyed by thermally induced thermal runaway.

Last edited by saptzae; 27th Jan 2013 at 06:25. Reason: Typo, Add @Turin
saptzae is offline