Commercial battery storage systems (BESS) guide
When the charges it can avoid are bigger than the battery. A commercial battery energy storage system (BESS) can cut demand charges, shift energy out of peak hours and store surplus solar—but whether it pays depends on which charges the battery can avoid, when the site uses electricity, and whether the battery has the power and duration to cover the event. This page shows how to tell.
Start here
Each sign points at a different charge the battery can avoid. The evidence listed under each one is what separates a real case from a brochure number.
The bill charges for the highest kW or kVA recorded in a set window, on top of the energy used. A battery that reliably covers that interval can lower the maximum that reaches the bill.
Evidence needed: twelve months of interval data with every billed maximum, plus the tariff’s demand windows and reset rules.A time-of-use tariff charges more in set windows, often weekday evenings. A battery can charge when energy is cheap—or from solar—and serve load when it is not.
Evidence needed: the full retailer rate schedule, the network charging windows, and interval usage grouped by tariff window.Solar exported for a few cents—or curtailed by a connection limit—can be worth far more stored and used on site later.
Evidence needed: interval imports and exports, solar production and genuine curtailment records. Metered exports are not curtailment.Backup power and market income are legitimate considerations, but they rarely carry a commercial case on their own. Both are covered in the value-stream register.
The strongest commercial case
Demand charges on your electricity bill are usually the largest controllable cost a battery can reach—and the easiest to overestimate. If the phrase itself is new, start with how demand charges work on a business bill and come back.
Power · kW
Energy · kWh
Most demand tariffs bill the highest 30-minute interval inside the demand window. There is no averaging: a single hot afternoon can set the charge for the whole month.
A 100 kWh battery with a 50 kW inverter runs at full power for about two hours. If the peak event lasts three, the battery empties and the meter records a new maximum anyway.
Covering nineteen peaks and missing the twentieth can leave the billed maximum unchanged. Reliability matters more than average performance, so model a missed-peak case.
Common demand windows such as 16:00–20:00 sit after solar output falls. Solar can cut energy use all day while the billed peak survives untouched.
Most large-business tariffs bill demand in kVA, not kW. A battery discharging real power may reduce only part of the apparent-power peak, because reactive behaviour counts too. Where power factor is poor, dedicated correction equipment can be the cheaper first action—check before treating the whole kVA charge as battery-addressable.
50 kVA × 58.421 c/kVA/day × 31 days ÷ 100 = $905.53$905.53 less in the network demand component for that month. Held every month, the same reduction is worth roughly $10,100 a year across the seasonal rates.
This is gross, not net: battery cost, charging energy and losses come out of it, and one missed peak can set a new maximum. It is a network rate, not a retail quote—the retailer’s pass-through decides what the bill does. And because N19 bills kVA, confirm how much of the peak is reactive before counting the full reduction.
See the current N19 rates and time windows How to size a peak-shaving batteryEnergy arbitrage, honestly
These streams are real but thinner than demand-charge savings, and the honest number is always net of charging cost, losses and forgone export income.
Each NSW network sets its own tariff windows, and they differ by tariff. Peak windows on large-business tariffs commonly fall on weekday evenings—Endeavour Energy’s N91 runs 16:00–20:00—while a midday solar-soak window can make the early afternoon the cheapest energy on the bill. The windows and rates on the site’s assigned tariff, not a national average, set the realised spread. The NSW tariff guides list the current windows by network.
The value is the realised price spread after losses, not the headline peak rate.
On Endeavour Energy’s 2026–27 N91 network rates, moving 1,000 kWh from high-season peak (27.67 c/kWh) to the 10:00–14:00 solar-soak window (5.69 c/kWh) changes the network energy component by $219.84: 1,000 kWh × (27.67072 − 5.68656) c/kWh ÷ 100. Round-trip losses mean the battery imports more than it delivers, and the retailer’s rates—not the network’s—decide what the bill actually does.The saving is the avoided import, minus the feed-in income forgone, minus battery losses.
Exported solar is not free battery fuel: it already earns the feed-in tariff. Illustrative only—with a 35 c/kWh evening import and a 7 c/kWh feed-in, each stored kWh is worth about 28 c before losses, not 35 c.Where a connection limit forces the inverter to discard generation, stored curtailment is close to free fuel.
Curtailment must come from inverter or controller records. A smart meter cannot show energy that was never generated, and metered exports are not curtailment.Do not add these savings on top of peak shaving without a single dispatch model—one battery dispatches once through time. The stacking rules below apply.
Worth checking first
These conditions do not rule a battery out, but each one weakens the case and should be tested before equipment is discussed.
A weak battery case can still be a strong tariff-change, efficiency or power-factor project. The same interval data answers all four—see the NSW tariff guides for what the network charges actually look like.
Revenue stacking
Energy, inverter power and state of charge are shared resources. Capacity held for backup cannot also chase a wholesale event. Energy discharged for FCAS or export may be unavailable for the site’s evening peak.
A sound model dispatches the battery once through time, applies each value stream to that same dispatch, and prevents two parties claiming the same kWh.
Where subsidies fit
The next step
A useful review marks each value stream as evidenced, possible or unavailable for the site, then tests a few right-sized power and energy combinations against the assigned tariff. The inputs that make it quick:
Delivered example: All Secure, Three Kings — a 75.4 kWp rooftop array working with a 60 kWh commercial battery to extend solar use past daylight hours.
Start with the controllable signal
Research and scheme sources
General information only. Market prices, retailer contracts, tariffs and scheme rules change. A project model requires current site data and current commercial terms.