For most of the time we have been doing this, planning a distribution system meant sizing everything for the worst hour of the worst day and accepting that it sits underused the rest of the year. Peak drove the conductor, the transformer and the substation alike. That was a sensible way to work, because power had to be generated the same instant it was consumed and nobody could do anything about that.
Storage breaks that assumption, and it is worth being precise about what it changes and what it doesn't.
What storage actually does to a plan
A battery moves energy in time rather than adding generation to the system. The distinction sounds academic and it changes the question you are answering. You stop asking how much this system can deliver at five o'clock on an August afternoon, and start asking how much it can deliver across a window, and where you can put energy in the meantime.
On a feeder peaking uncomfortably close to its rating for four hours a day, that can be the difference between a reconductoring project and a battery. Sometimes the honest answer is still that you need the wire. The point is that there are now two possible answers where there used to be one, and which one is right depends on the shape of your load curve rather than its height.
Where it gets complicated
Storage is a generating source as far as your protection is concerned, and it can push current in either direction. That matters for coordination. Fault contribution from an inverter-based resource does not look like fault contribution from a rotating machine, and relay settings developed for a radial feeder with load at the end of it were not written with that in mind.
It matters as well for your interconnection studies, your voltage regulation scheme, and anywhere you have reclosers built on the assumption that power flows one way. None of that argues against doing it. All of it argues for modeling before the purchase order rather than after.
Where we usually start
Start with the load data you already have. Most utilities and larger facilities are sitting on years of interval data that nobody has ever looked at as a shape. Before anybody discusses kilowatt-hours of battery, work out how many hours a year the system is genuinely constrained and by how much. Eighty hours and a battery is obviously the answer. Two thousand hours and it obviously isn't.
Load flow and contingency analysis come next, run with the storage in place, to see what it does to voltage and to fault duty under both normal and N-1 conditions. That is the work that tells you whether the thing will do what the proposal says it will do.
