In a thesis manuscript, I read that a cell having IR 3 times bigger than nominal (indicated on cell datasheet) has to be rejected.
Yes, but thesis are often based on theory and not much experience. I prefer having both theoretical and practical information and then start doing my tests. What I found is this. I have quite a few Samsung 22F cells. The datasheet says this about IR:
Which isn't too valuable as an information... The student who wrote the thesis could suppose that 3x100mOhm (or 3x75mOhm) would be a higher limit for IR. But actually what
@cak says makes more sense. What I measured on my second hand 80-90% SOC cells is this:
In the 2100-2199mAh range I have 89+12=101 Samsung 22F cells. Most of them are in the 50-59mOhm IR range and a few are in the 60-69mOhm range.
Plus, what I noticed while processing cells is that the older ICR chemistry cells had a higher IR, between 80 and 100mOhm. The really old cells went above 100mOhm (up to a maximum of 110mOhm). And these high-IR cells seem to degrade in capacity a lot faster that the others; I measured that by doing many C-D-C cycles and letting cells rest between each cycle.
So, the datasheet says that IR for Samsung 22F new cells should be <=100mOhm. My conclusion if that 50-59mOhm is a nice value for second hand Samsung 22F cells (quite a few measure 50-51-52mOhm) and 60-69mOhm is acceptable. And, just like
@cak I won't use cells above 80mOhm in the powerwall. I will use them for non-critical applications (my bike lights, battery backup devices, my ATTiny IoT boards and so on).
Something else it that use pack of cells with roughly the same IR is cautious when building packs.
Yes, I think everybody has a rough idea of an acceptable range for both IR and capacity, when mixing cells in a pack. My range will be about 400mAh for capacity and 30mOhm for IR.