And how does one go about balancing capacity, given that generally each has a unique capacity?
With the caveat that I can't speak for @toms am not positive what he means by capacity balanced, and the theory behind balancing is one of the weakest parts of my understanding/knowledge, this is what I am envisioning:
Red lines equal max bandwidth of balanced pack, green lines equal reduced bandwidth of unbalanced pack (?)
In all three scenarios, mismatched cells will leave some capacity on the table, as the lowest capacity cell (or weakest cell?) would limit the entire pack. But with a top or bottom balance, SOC is at least aligned at one or the other end of the SOC range, so that all cells theoretically hit either full or empty at the same time. At the opposite end of the range, cell differences will become apparent and the smallest or weakest cell would hit the cutoff threshold first.
With an unbalanced pack I imagine it being like the left hand example. If SOC is never aligned you could have cells (limiting the bandwidth at both the top and the bottom of the SOC range.
As I said, balancing theory is one of my weakest points, by a long shot, so if anyone sees an error in my thinking, or a fundamental misunderstanding anywhere, or even just questions any part of it, please point it out.