12S3P pack cells die for what reason ?

Hi,

Sounds like a normal situation, but I received from my company I worked for a bunch of left over battery packs and I decided to use them to built a 12S pack for a bike. The original packs were German manufactured according to our specs using Panasonic 18650PF cells. Packs were never in operation but as the product we made ended the left over packs were given to me. The packs we 2S3P and they were professionally manufactured with all necessary protection circuits etc in them. I took out the together welded parallel cell pairs and simply joined 6 of them in series to create 12S3P. Then there is a balancing circuit built next to it and all low voltage cut off circuits etc. My pack is connected to handle bar display unit where I can see constantly voltages from all cell units in the pack.
Motor controller is 350W KT, limiting current is 17Amps which is lower than what 3 x PF cells can output.
Now after 2 weeks of use and testing cells started to die, first I changed the cell 12, and now after a day of use, Cell 1 is having low capacity expecting to die aswell after few runs.
Currently I'm charging the system with charger / balancer combo built around LTC4020 and LTC6804-2. System monitors all cells constantly and shows the progress of all cells on display during the charging. Charging voltage is only 4.05 Volts per cell. Discharge limit is down to 3.3Volts which I never reach before recharging.

Is there a known failure mechanism in this kind of system configuration? I apparently never exceeded the output current nor I never over charged the cells, neither going too low and all cells were fresh never used.
Is there a reason why cell 12 died first, and now cell 1? or is it just coincidence.

Thanks.

Possibly the cells were over discharged at some point and damaged. If they were ever below 2v or so, you can get irreversible damage. Frequently the "protection" circuit can slowly drain the cells below the safe limit.

If the cells were never below 2v or otherwise abused, I can't see a reason for them to fail.

BMSs can cause problems as often as prevent them.

At this point I'd pull out the cells and load test each individually, then reassemble the good ones using a known-good BMS.

You are better off with 6p of used cells.