A battery that charges and discharges on a fixed schedule is a battery, not an asset. Optimisation is the difference — and it's a harder problem than most people assume before they own one.
Battery storage optimisation means deciding, for every settlement period, whether to charge, discharge or hold — across every revenue stream the asset can access simultaneously, not just one. A battery run on a fixed daily schedule captures a fraction of its real value and is exposed to whatever the forecast got wrong. Done properly, it's a linear or mixed-integer programming problem: state of charge, degradation and grid limits as constraints; price, demand and revenue-stack rules as the objective; re-solved continuously as forecasts update, not locked in once a day.
Buy a battery and someone will hand you a default: charge overnight when power is cheap, discharge in the evening peak. It’s not wrong, exactly. It’s just a small slice of what the asset is actually capable of earning — and in a market where prices move by the quarter-hour, a fixed schedule set once a day is already out of date by lunchtime.
A grid-connected battery rarely earns from just one thing. Depending on the market, it can stack several revenue streams at once: energy arbitrage (buy low, sell high), frequency response and other ancillary services, capacity payments, and peak shaving for a site or portfolio behind the meter. A fixed schedule can chase one of these reasonably well. It can’t chase all of them, because they pull in different directions at different times — the quarter-hour that’s best for arbitrage isn’t necessarily the one the grid operator will pay for frequency support.
Optimisation is what resolves that conflict properly: a model that sees every available revenue stream and every physical constraint at once, and picks the combination that maximises value across the whole horizon — not the best move available right now.
The obvious approach — forecast tomorrow’s prices, solve for the best schedule, run it — has a structural flaw: the forecast is wrong, just not by a known amount. Commit fully to one predicted price curve and the schedule is optimal for a day that doesn’t happen. We’ve written before about what battery operators can learn from thirty-year-old hydroelectric dispatch practice: the good ones never bet the whole plan on a single number.
Degradation adds a second layer. Every cycle costs some of the battery’s remaining life, and that cost has to be priced into the same decision as the revenue — otherwise the model happily trades away years of asset life for a marginal arbitrage gain that isn’t worth it.
A battery that never sees the constraint it’s about to hit isn’t optimised. It’s just committed to a plan for a day that hasn’t happened yet.
This is energy asset optimisation in the form it usually takes for storage: a linear or mixed-integer programming model with state of charge, round-trip efficiency, degradation and grid connection limits as constraints, and the available revenue streams as the objective — built using the same applied mathematics that underpins our dispatch work elsewhere. Rather than solving once a day, the schedule re-solves on a rolling basis as prices and forecasts update, using robust and scenario-based methods so a single bad forecast doesn’t wreck the plan.
We’ve built this pattern for frequency response on distributed batteries in the UK — real-time detection and dispatch across a fleet of assets, not a single site — and it’s the same discipline behind treating dispatch as a design choice rather than an afterthought once the hardware’s installed.
Most batteries run on a schedule built for one forecast and one revenue stream. We build the optimisation that captures the rest. Let's talk about yours.