How many lithium iron phosphate (LiFePO4) can safely be connected in parallel, in order to achieve higher power output (and capacity)? Wired directly together, without components such as resistors or power transistors limiting current flowing between parallel cells.
Precautions taken would include ensuring they're brand new cells from the same manufacturer lot, at about the same state of charge, and letting them rest for a day to equalize before charging or discharging the pack.
This chemistry is supposed to be much safer than other lithium chemistries, as it doesn't catch fire even when punctured. Is it also safe to add more cells in parallel? The most I've seen discussed for other chemistries is around 3 cells in parallel. Would it be possible to get to 10, 20, 30 cells in parallel with this chemistry? The completed pack would drive a DC motor.
If it's not safe to wire that many directly together, would it be safe if current limiting components or fuse/polyswitch were added between parallel cells? Or is it better to architect a large pack in strings of at most 3P x (however many) in series?
Note: NOT asking about other lithium chemistries such as lithium polymer. Specifically LiFePO4 in form factor of cylindrical 18650 or 26650. Advice about tradeoffs in using much larger cell sizes rather than parallel small cells welcome.
When cells in parallel are new and well-matched, they balance well. As the cells age, their internal resistances get progressively less matched. Somewhat self-correcting as a cell with lower internal resistance will charge & discharge at higher current than others - thus getting aged faster and catching up to its peers.
However if one cell fails to open circuit or very low internal resistance, current will now be flowing mostly through that one cell. Say the pack normally gets charged at 1C rate, now that cell is getting 3C rate. If there are 20 cells in parallel, now it's 20C rate. The more cells in parallel, the more current can flow to the most unbalanced cell. It seems dangerous not to limit this current.
@SpehroPefhany recommends self-protected cells. Are there methods which add protection circuitry around unprotected cells? Are lithium iron phosphate cells considered safe enough (no fire/explosion) not to require it, with cell failure simply leading to low pack performance or requiring pack replacement? How are dead/shorted cells normally handled in these large pack designs?