The TL;DR for the quote above: Due to electrical rules, the weak cell that supposedly went to 1V would instead actually only drop to the voltage of the other cells in the brick, but likely would be providing close to no current. Then when the load is removed, to recover it would snag energy from the rest of the parallel brick and remain at the brick voltage as expected.
Bit of complexity here that's not immediately obvious.
Basic rules:
- Cells in parallel always have an equal voltage. Current can vary.
- Cells (or cell groups) in series have an equal current. Voltage can vary.
So first, every cell in the brick of 74 cells is at the same voltage. That's what being in parallel does. It doesn't matter which ones are weak or strong or whatever, they are forced to the same voltage. Your above example doesn't work because a single cell in the group can't be at 1V because its + and - are in parallel with other cells. Might seam counterintuitive, but this is how it is. The voltage at the cell sense for a brick of cells is the voltage of every cell in that brick.
That said, let's say there is a weak cell. Under load, because its in parallel with the rest, it will reach a lower voltage while outputting less current than the other cells at the same voltage. So if we load the cell group at X amps, and this causes the brick to drop from Y volts to Y - 1 volts, then we're getting X watts from that brick of 74 cells in parallel. Within that group, however, the distribution of current does not have to be constant like the voltage does. One of the cells may only be putting out 1/148th of an amp (because its weak and has half capacity of the rest of the cells) and that means the rest as a whole are putting out 147/148th of an amp, or slightly more than 1/74th of an amp per cell. The end result is the same, though, that the cell group drains each cell's SoC equally regardless of which cells are weak or strong. In parallel this doesn't generally matter as long as the connection making them parallel is low resistance (in this case, it's a solid aluminum bus plate) and the cells are of the same general type (same voltage curve).
This is generalized, but should get the idea across.
Here's some potential issues:
A problem comes in when a cell is so disproportionately matched with the group such that at some point under load or charge it attempts to output or draw more than the rating of the cell level fuse (~25A instant pop, ~20A for 15s, etc). This is where we get into the weeds a bit, but suffice it to say a weak cell will lose its "surface charge" more quickly than a strong cell under load. So let's say under a load, a weak cell quickly drops its voltage to line up with the group... then the load is released. The stronger cells will more quickly recover to a higher resting voltage because they may or may not have released their surface charge, while the weak cell may have. In this case, the weak cell wouldn't recover as quickly and charge would be shuffled through the bus plate to equalize it with the rest of the group. This inrush of current can be very high if the cell is very weak, enough to pop a cell level fuse in extreme cases.
A far less common, but more severe issue is that a cell becomes parasitic in some way. In this case, a cell is self discharging at some rate due to some defect. As it discharges, its voltage attempts to drop. But since its in parallel with other cells, it can't drop its voltage alone, so the other cells attempt to balance out with it. This effectively makes this one parasitic cell parasitic to the entire brick of 74 cells, and the voltage of the entire brick slowly drops as a result (at 1/74th the rate that it would had we been measuring just a single parasitic cell in isolation outside a brick). This becomes an issue when the parasitic load can handle high currents that are possible from many cells in parallel like this. When it gets to the point where it weakens the cell level fuse, it pops and this cell is pulled out of the brick as a result and then just self discharges by itself down to 0V, rendering it pretty inert.
Suffice it to say, there is actually a great bit of variation among the cells that are in parallel. In my testing of cells taking from completely stripped modules, they can vary in double digit percentages on capacity and still function perfectly fine as a brick. It's mostly these differences that tend to cause an imbalance under normal use and why the BMS has to balance at all. If all the cells were perfect, you wouldn't need a BMS to do balancing. Instead, what happens is during normal use some amount of energy is lost as weaker cells snag energy from stronger cells during load changes. This process is not 100% efficient, so this tiny amount of energy loss adds up to an imbalance over time. Bricks that are very well matched actually end up being out of balance by positive deltas, and bricks that are very mismatched tend to be the ones out of balance by negative deltas. With 74 cells in parallel, most bricks are pretty on par with the average capacities for its neighboring bricks in the same pack and tend towards a negative delta that is pretty even and thus not really noticeable. Then the most evenly matched groups that lose the least amount of energy end up a higher voltages over time and need to be actively balanced. There's also the scenario where a group is significantly weaker and ends up at a higher voltage due to reaching a charge peak faster, but that's another story. I'm mainly referring to the average pack here.
With all of the above as a bit of base knowledge, if you manually cut the fuse on a parasitic cell, you've stopped the drain, but now you've assured that that brick is 1/74th out of balance with the rest of the pack forever. This will manifest as being weaker under loads and normal use, and will reach charge voltages sooner than the rest of the series cell groups in the pack every single time. The BMS will only tolerate temporary gross imbalances like this. If they persist it throws a series of errors, culminating in complete shutdown for pack safety. How quickly this happens depends on actual usage (with more parked/idle time helping the BMS correct the gross imbalance each time), but the BMS can not keep up with a completely missing cell from a group for the long haul. It was never designed to. It was, however, designed to detect this issue, report it, and eventually shut things down for safety. If you use the car for a 30 mile commute once a week or something, it might last a while with a missing cell.... but daily normal use for a bit, or even one relatively long supercharger trip where the BMS has no time to tackle the imbalance sufficiently for the next charge and your "grubered" pack is now a rather large and expensive paperweight.
In theory if you could find and cut the fuse on the weakest cell in every single one of the 84 to 96 cell bricks, this "fix" would work indefinitely.... but the aforementioned common voltage in parallel makes this impossible (you can not test the individual cells to find the weak and strong), so, don't try it because you'll only make things worse (you'll remove the strongest cell in some, weakest in others, and randoms in the rest and cause even more imbalance).
If you follow through in more detail with the application of our basic rules above, some of this may seem self-correcting (higher voltage bricks will provide more power under load because of I*V coming out higher due to constant current in series, and vice verse for lower voltage bricks). However, that only works with "ideal" batteries. (Think spherical cows in a vacuum.) Since our real world batteries have varying internal resistance, vary amounts of surface charge as that resistance varies, internal resistance that varies with temperature variations between cells, etc... the self-correcting aspects of cell groups in series are almost completely negated............. that's why we have a BMS.
This is not an extensive an in depth explanation of all of this, but should make some of this make sense.
Anyway... hope this is informative. If you like this video I hope you'll like, share, and subscribe..... er, wait.