Duration — energy divided by power — reads like a spec-sheet detail. It’s actually the single decision that determines which revenue streams your project can legally and physically touch. Same inverters, same interconnection: at one hour you’re a sprinter with a narrow job list; at eight you’re infrastructure. Here’s the honest map.

For a visual companion, explore the interactive The Duck Curve diagram on BESS.Engineer.

1–2 hours: the specialist

Short-duration systems are power-dense and cheap per MW, built for the fast markets: frequency response and regulation, short arbitrage bursts, grid services where speed is the product. The catch arrives at the capacity market door: resource adequacy wants sustained delivery across peak windows, so short systems are heavily derated — credited for only a fraction of their nameplate MW — or excluded outright, depending on the market’s rules. And their home turf betrays them: ancillary markets are shallow. Frequency products need megawatts in the hundreds-to-low-thousands per system, so the first wave of batteries saturates them and prices collapse — the repeated history of every storage market’s opening act. Short duration is a fine first chapter and a dangerous whole book.

4 hours: the standard for a reason

The industry’s center of gravity sits at four hours because several curves intersect there:

C-rate is just the inverse of the hours — and four of them is the line the capacity market draws.
4 h — the common firm-capacity thresholdfast frequency responsethe short end of the range0.5 h = 2C1 hour1 h = 1C2 hour2 h = 0.5C4 hourthe default US resource-adequacy shape4 h = 0.25C8 hourlong-duration energy shifting8 h = 0.125C1 h2 h4 h8 hduration at rated power, and the C-rate it implieshow far short of the capacity threshold the shape falls
  • The peak is about that long. Evening net-load peaks in solar-shaped markets (the duck’s neck) run a few hours — four hours of discharge covers the window that matters.
  • Capacity rules historically anchored there. The influential convention crediting four-hour systems with full (or near-full) capacity value made “4h” the bankable default — though modern effective-capacity methods (ELCC-style) now slide credit with duration and with how much storage the grid already has: the more batteries serving the peak, the longer the residual peak becomes, and the more duration full credit demands. Duration requirements are a moving target that rises with storage penetration.
  • Marginal energy is cheap to add — to a point. Going from 2h to 4h reuses the same inverters, interconnection, and land; you’re buying mostly DC blocks. The MWh economics of falling cell prices made that stretch progressively easier.

Four hours, in short, is where a merchant battery can hold the full stack: meaningful capacity credit, the fat evening arbitrage, and ancillary participation on the side.

8 hours: infrastructure, if someone pays for it

Doubling again buys access to longer peaks, deeper renewable-shifting, and the emerging long-duration procurements where planners explicitly buy 8h+ resources. The obstacles are symmetrical: every added hour must pay for itself, and hours five through eight discharge into progressively cheaper parts of the evening — declining marginal revenue meeting linear marginal cost. Merchant 8h rarely pencils today; contracted 8h — capacity contracts, LDES tenders, utility procurements in duration-hungry systems — increasingly does. Eight hours is a policy-and-contract product wearing a technology costume.

The portfolio truth

Real fleets treat duration as a portfolio variable: augmentation can stretch duration over a project’s life as cell prices fall and capacity rules tighten, and sites are increasingly designed power-rich with room to grow energy-rich. The discipline is to underwrite duration against your market’s capacity accreditation rules and saturation trajectory — not against last year’s, and never against a conference slide’s.

FAQ

Why do capacity markets derate short batteries? Because adequacy is about surviving the whole peak window: a 1h battery at a 4h peak covers a quarter of the problem, and accreditation methods increasingly measure exactly that contribution.

Is longer duration always safer? No — unpaid hours are stranded capital. Duration should match the revenue that funds it: contracted length for long builds, market spreads for the middle, and speed products only ever as seasoning.

What about 100-hour storage? A different technology conversation (iron-air, hydrogen, thermal) for a different problem — multi-day and seasonal gaps that lithium’s economics don’t reach. See the beyond-lithium landscape.

Duration strategy — including the augmentation math behind “grow into 8h” — is a core module of my Grid-Scale BESS: Complete Guide.