In theory,
Self discharge current of the most self-discharging cell, minus the self discharge current of the least self-discharging cell, divided by the ratio of time you can spend balancing.
An example:
Highest leakage: 30µA.
Lowest leakage: 20µA.
-> Imbalancing difference: 10µA
Stupid balancing algorithm that only works in CV state - battery fully charged once a month - CV phase average length 1 hour: balancing time duty cycle = 1/(30*24) = 0.0014
-> Balancing current needed: 10µA/0.0014 = 7 mA
In reality:
* Cell self-discharges are unspecified.*
* Balancing algorithm is unspecified, if you use an IC
* Even if you knew the algorithm, usage pattern is extremely difficult to predict
*) I made my own extensive study over 1.5 years and ~40 samples, and can say with a confidence that any proper 18650 cell leaks less than 10%/year or approx. 30µA, when stored 100% charged, at elevated (40 degC) temperature, but most much less, and the leakage is practically zero below about 50% SoC, and around 3-4%/year for room temperature even at 100%. But leave room for error.
Obviously, balancing algorithms that only work at the charge end (CV) phase - still very usual - completely fail to do any balancing whatsoever if the user never fully charges the battery, which is usually recommended for good battery life
I have dissected imbalanced packs with balancers, and balanced packs without balancers. It's very hard to guarantee balancing (which is also why, for worst edge case safety analysis, a balancer is modeled as imbalancer!)
A good algorithm can enable 100x lower balancing current to be used, by doing the work over 100 hours instead of 1 hour! Like I said in an earlier thread, I had no problems whatsoever keeping an abused (overdischarged and revived) 250Ah-ish 20kWh pack in balance with mere 40mA balancing currents, while the competitors do sell products with 1A balancing currents (and stupid balancing algorithm) for the exact same market - they require proper thermal design!
As a result, balancing current is almost always hand-waved (yes, even by very professionals), based on experience and feelings. This is an error-prone process, but because cell balancing is not a safety feature (if someone claims so, don't let them close to designing battery protection systems!), this is not a big issue.
The most important thing,
by far, is to guarantee that whatever balancing circuit you use, it won't
cause problems. It has to have very low leakage characteristics (if non-negligible, then it has to be balanced with little unit-to-unit variation) so that it doesn't
cause imbalance, or, worse, overdischarge cells in long-time storage.
If you can guarantee no ill effects, then any balancing would be a plus, and the more you have, the more capable it is in restoring energy storage of strange, corner case packs.
If you mean redistributive by "active", no, it's practically useless, but remains an academic subject. Looking only cost, the ROI is in thousands of years or infinite; complexity and system cost goes up with either no benefits at all, negative "benefits", or very minor energy or energy density gains.