Since the whole can is -, if something dramatically physical (crushing, etc in a crash) happens to the pack causing shorts between cans, the fuses can do nothing to eliminate the short (they won't even see the current flow)
That's a very good point. If a P-string is getting hot, and internal pressure building creates a breakdown internally in the weakest cell (with a manufacturing flaw?), leading to a thermal runaway, a fuse on either tip would be fine. In that case the rest of the cells in that P-string would be separated from the short.
When one cell shorts internally, it is as if that cell was replaced with a wire connecting the pos and neg busses on that P-string. Having individual cell fuses will not stop that single shorted cell from going off in a thermal runaway...its going up in heat and smoke, and there's nothing much to stop it but to take it outside and set it on concrete away from the house.
But...when that bad cells' fuse melts...it separates the other cells in that P-group from the shorted cell, to prevent the others from going into a catastrophic cascade, where severe heat causes one cell after another to go off...almost all at the same time.
There are two reasons I like the honeycomb layout, instead of a square rank-and-file. Both because doing that results in better air-circulation if the cells are spaced apart from each other about 2mm (or more). Of course you can space out the cells in a square pack, but once you draw it out life-size on paper, I am convinced that honeycomb is slightly better.
Some air-space and air-circulation between the cells allows the option of an added air-fan (I recommend two, one pulling, one pushing, centrifugal 48V DC server fans with ball-bearings...very cheap and long-lasting). But...even if an air-fan is not added, if you construct two almost identical packs (both honeycomb) but one has all the cells touching to make it compact, and the other has a 2mm air-space between each cell?...
Hard running will make the pack warm, but the cells near the edges can shed some of that heat. The center of the pack will remain the warmest part. If there is a heat-related cell-failure, it is likely to be in the middle of the pack. If all the cells are touching (to make the pack as compact as possible) it will be very bad. If they each have just a little air-space...it will be..."less bad?"
First, make the pack large enough that the P-groups are not strained. Second, chose a cell that can handle the loads you will be applying so it only gets a little warm at the worst. Third, If the center 30% of cells in a pack die an early death due to frequently running warmer than the cells at the edges, will you rebuild this two-year old pack? NO...you throw away the entire pack, so...getting the heat out of the center of the pack can double the life of a pack.
Tesla has an eight-year warranty on their battery packs, and a 2-year old Tesla-S taxi just passed 300,000 miles while the pack is still providing over 90% of its rated range when it was new. They are doing something right.
Tesla NEEDS individual fuses because...they have over 70 cells per P-group. If one cell internally shorts, then all 70 of its neighbors will instantly be dead-shorting too, through that shorted cell (acting as a solid wire during a short).
If you have a spot-welder, I now agree with amberwolf. You should spot-weld the fuse-wire onto the positive tip, and spot-weld bus-ribbon to the negative. This provides better protection in a crash scenario.
However, if you will be using a soldering system (and hundreds in the third world will be), I would still connect the fuse-wire onto the negative, to prevent even coming close to heat-damaging the internals. If you overheat the plastic separators inside the cell, they will shrivel up and expose the parts they were designed to insulate. You could be creating the very scenario that you are hoping the fuses will avoid.
My goal is not to get people to solder instead of spot-weld...my goal is to help those who are limited to soldering, by identifying the methods and materials that will make a soldered pack as safe as it possibly can be. I am open to hearing persuasive suggestions to the contrary on any point.
"Fuse-wire for individual cells, DIY Tesla style"
https://endless-sphere.com/forums/viewtopic.php?f=14&t=88039
spot-welded steel fuse-wire
If using a soldering connection, fuse-wire can be copper or aluminum (if desired)