“How much storage do we need?”#

Two questions in a trench coat.

• How fast can it deliver? — power
• How long can it keep delivering? — energy

A specification that answers one and assumes the other is how a system arrives at an event under-built.

The units do the arguing#

Duration is the ratio: energy ÷ power.

A specification, deliberately absurd#

1 GW / 1 GWh — illustrative, not a product

Nominal power of 1 GW. Stored energy of 1 GWh.

Duration: one hour.

What one hour of duration buys#

At full output it is empty in an hour.

That is genuinely useful — for a frequency event, an evening peak, a generator tripping offline.

It is the right machine for the wrong problem.

The mismatch#

• Event span: 61 hours (synthetic trace)
• Storage duration: one hour (illustrative spec)

The power rating was never the binding constraint.

Four levers#

1. Power — how fast it can deliver
2. Energy — how much it holds
3. State of charge — how full it was when the event started
4. Round-trip efficiency — what you lose storing and retrieving

Which lever moves the outcome?#

Against a multi-day deficit:

• raising power — no, and not approximately
• raising energy — yes, until it becomes unaffordable
• starting fuller — yes, and it is free if you saw it coming
• better efficiency — helps at the margin

Why the instinct misleads#

A store sized for a daily peak is solving the problem a daily cycle poses.

The daily peak is a power problem.

A drought is a duration problem.

The same word covers both, and the procurement does not transfer.

What this means for your dispatch#

In Assessment 3 you will spend a finite store across a multi-day event.

Every hour you discharge early is an hour unavailable at hour 60.