Plain-English answers to the questions that matter before you talk to anyone, including us. No jargon, no vapour numbers, and where a claim has a basis, the basis is stated.
Commercial bills have two meters running. Energy is what you use across the month. Demand is how hard you pull at any one moment, and it is priced off your single worst half-hour. The network builds its poles and wires for that moment, so that is what it charges you for. Cap the peak, and the charge follows it down.
Pull out a recent invoice and find the line priced in kW or kVA rather than kWh. That is the demand charge, and on many commercial tariffs it is set by your single highest half-hour reading of the month. The frustrating part: it does not matter whether that peak lasted thirty minutes or was your only spike all month. The network sized its response to your worst moment, so the price did too.
Two things: how spiky your load is, and whether the battery has the power rating to carry the spike. A site with a flat load profile has little demand charge to recover. A site whose compressors, welders or chargers all land at once can see the demand line become the biggest saving a battery makes. This is exactly what a load assessment reads out of your interval data.
Find your peak with a load assessment →The graphic above runs across seven days because that is how demand behaves in the real world. On the live project it is modelled on, the site held a steady load around 240 kW, dipped hard through the middle of each day while solar carried it, then threw a different-sized spike each evening. Only one of those seven spikes set the month's bill. Everything else the site did all week made no difference to the demand line, which is exactly why a battery that watches every half-hour beats any amount of guesswork.
Storage scales, and so does the saving. On the same live project, one battery was modelled to hold demand near 220 kW. Two brought the cap down to about 146 kW. Three pushed it below 55 kW, taking a site that used to spike past 350 kVA down to a draw the network barely notices. In the mid option, demand charges fell from roughly $72,000 a year to under $30,000. Each extra battery buys a deeper cap and a bigger saving, so the right number is a sizing exercise: we model your actual meter data and show you what each step is worth before you commit to anything.
Not all demand charges behave the same way, and the difference changes how much a battery is worth. A monthly peak tariff resets every billing cycle: one bad afternoon costs you one month. A rolling ratchet holds your highest reading against you for up to twelve months: one bad afternoon costs you a year. A capacity charge is agreed in advance: you pay for a fixed ceiling whether you touch it or not, and exceeding it brings penalties. Batteries help with all three, but the ratchet is where they earn hardest, because a single avoided spike keeps paying for twelve months.
Picture a cold-storage site on a February afternoon. Compressors are working hard against the heat, a delivery arrives and the dock doors open, defrost cycles land at the same time, and someone switches on the second blast freezer. None of these is unusual. Together they stack into one half-hour reading far above the site's normal ceiling, and that reading becomes the demand charge for the month, or the year. The point of the story: peaks are not caused by waste, they are caused by coincidence. You cannot roster your way out of coincidence, but a battery does not need to. It just needs to be charged when the stack happens.
Mostly no. The demand charge only cares about your single worst half-hour. You can cut consumption 20% and still pay the same demand line if your peaks are untouched.
Staggering helps where you control the timing, and it is free, so do it. But most peaks come from coincidence you cannot schedule: weather, deliveries, production surges. The battery covers what the roster cannot.
It is in your interval data, which your retailer must provide. Send us twelve months of it and we will show you the exact half-hours that set your bills.
In one sentence: your demand charge is set by your worst half-hour, and a battery exists to make sure the grid never sees it.
This is the physical behaviour behind demand reduction. The system watches your load climb toward a threshold, and the moment it gets close, the battery discharges and carries the difference. The grid never sees the spike. Forecasting matters here: the battery has to be charged and ready before the risk window opens, which is why the plan is rebuilt from fresh forecasts every day.
Shaving a peak sounds like a reflex, but the work happens hours earlier. If the battery is flat when the spike arrives, nothing can be shaved, and if it holds too much back, that energy earns nothing all day. So the system forecasts tomorrow's load, solar and prices, decides when the risk window opens, and makes sure the battery walks into it charged. That plan is rebuilt from fresh forecasts every day, and AI Mode keeps tightening the threshold as it learns how your site actually behaves.
Shaving is a power problem first: the battery's kW rating decides how big a spike it can carry. The kWh rating decides how long it can hold the spike down. A short sharp peak needs muscle; a long afternoon plateau needs stamina. Sizing the system means reading which shape your site produces, which is why we model from your own half-hours rather than a rule of thumb.
How sizing feeds the payback model →Hours ahead, the forecast flags a risk window for late afternoon and the plan makes sure the battery walks into it full. As load climbs toward the threshold, the system is watching real readings, not the clock. The moment the trajectory says the threshold will be crossed, the battery starts discharging, and from the grid's point of view the site simply flattens. When the underlying load falls away again, the discharge tapers, and if the tariff allows, the battery quietly recovers charge in the cheap hours that follow. Nobody on site notices any of it, which is the entire point.
The awkward scenario is two spikes in one afternoon. Shave the first with everything and there is nothing left for the second, which then sets your bill anyway. This is where planning beats reflexes: if the forecast sees both, it rations the shave across them, holding the demand line at the best achievable level for the whole day rather than winning the first battle and losing the war. It is also why battery energy capacity matters, not just power: stamina is what survives the second peak.
Plenty of sites have tried the human version: a demand alarm, a standing instruction to shed load, a supervisor with a checklist. It fails for boring reasons. The alarm fires mid-half-hour when much of the damage is booked. The person is busy, or on leave, or sheds the wrong thing. And the loads you can safely drop in five minutes are rarely the ones causing the spike. Automated shaving reacts in seconds, needs nobody present, and touches production not at all.
No. The site draws the same power it always wanted; part of it simply comes from the battery instead of the grid. Machines cannot tell the difference.
The battery shaves what it can and the demand line settles at the best achievable level. Sizing against your actual spike history is exactly what the load assessment is for.
Forecasts built from your site's own history, weather and calendar patterns, corrected against live readings all day. It gets better the longer it runs on your site.
In one sentence: shaving is the battery stepping in for the exact minutes your site would otherwise set a new peak, and forecasting is what makes it reliable.
Most commercial solar systems make their best power at midday and spill the excess to the grid for next to nothing, then the site buys expensive power all evening. A battery closes that loop: the midday surplus goes into storage instead of the grid, and comes back out when the sun is gone and the tariff is not.
Feed-in rates for commercial solar are a fraction of what you pay to buy power in the evening. Every kilowatt-hour you export at midday and buy back at 7pm is the same energy traded at the worst possible exchange rate. Storage flips that: the surplus keeps its full retail value because it never leaves your site. And as networks move to dynamic export limits, there are hours where exporting is not just poorly paid but capped, which makes the store-it case stronger again.
The right battery size falls out of two curves: how much solar you spill at midday, and how much load you carry after sunset. Too small and you leave surplus on the table; too large and capital sits idle. Designer reads both curves from your interval data and sizes the system where they meet.
How Designer sizes a system →The famous duck curve is what solar does to the whole grid: a deep midday dip in demand, then a steep evening ramp as the sun leaves and everyone switches on at once. Your site with solar has a private version of the same shape. Midday, your grid draw collapses and may even go negative. Evening, it surges back exactly when power is dearest. The battery is the tool that flattens your private duck: it swallows the midday dip and feeds the evening ramp, so your grid draw looks calm all day even though your site is not.
Self consumption is seasonal. In summer the midday surplus is generous and the battery fills easily from your own roof. In winter, shorter days and heavier heating loads can shrink the surplus toward zero, and a battery that only knew one trick would sit underused. This is why the working modes are a portfolio rather than a setting: on thin-solar days the optimiser leans harder on arbitrage and shaving, and the asset keeps earning either way. A battery bought for one mode alone gets judged unfairly by August.
A common fork in the road. If your existing solar already spills a healthy midday surplus, a battery usually beats more panels: you are wasting energy you already own. If your solar barely covers daytime load, extra panels might come first, or the two together. The wrong answer is guessing. Designer models both paths against your interval data and shows which sequence pays back faster, with the assumptions on the page.
Model both paths in Designer →Because the feed-in rate is a fraction of the evening purchase rate. Exporting at midday and buying back at 7pm is trading your own energy at the worst exchange rate on offer.
Without storage, mostly yes: light load means maximum spill. With storage, quiet days become the best charging days, banked for Monday morning's ramp-up.
The system exports what the network allows, within any dynamic export limits, and the optimiser has usually planned the morning so the battery fills late rather than early, leaving room for the afternoon.
In one sentence: self consumption means your own solar keeps its full retail value by being spent on site after dark, instead of being sold for cents at noon.
Electricity prices move through the day, and the gap between the cheapest and dearest hours is the opportunity. In Australia the cheapest hours are now the middle of the day, when rooftop and grid-scale solar flood the market. The battery fills then and carries the site through the expensive evening. No behaviour change on site, no one turns anything off. The saving comes purely from when the energy was bought.
Prices are shaped by supply and demand across the day. Through the middle of the day so much solar is generating that wholesale prices fall to their lowest, and sometimes below zero. In the early evening solar output falls away while businesses and homes ramp up together, and prices spike. That daily spread between the cheap hours and the dear ones is the arbitrage opportunity, and it exists whether you are on a time-of-use tariff or exposed to wholesale prices.
On most commercial sites arbitrage earns less than demand reduction and self consumption, and it should never compromise them: a battery that sold everything into the evening price and then met your peak flat has cost you more than it made. This is exactly the trade-off the optimiser weighs every day, across all objectives at once, so each kilowatt-hour goes where it earns most.
How the optimiser ranks the jobs →Arbitrage value depends entirely on the spread you can trade across. A flat-rate tariff offers nothing: every hour costs the same, so timing is worthless. A time-of-use tariff offers a dependable daily spread between off-peak and peak windows. Wholesale exposure offers the biggest and wildest spreads, including the occasional price spike where a battery earns a month's value in an afternoon, and negative-price hours where you are effectively paid to charge. The dearer your peak window relative to your cheap one, the more this mode matters to you.
Cycling is what batteries are for, and the LFP chemistry in our systems is rated for thousands of full cycles under warranty. The real question is whether a given cycle earns more than it costs in battery life, and that is not a judgement a human needs to make at 6pm: the optimiser weighs the value of each discharge against degradation and warranty terms as part of the daily plan. Thin spreads that are not worth a cycle simply do not get chased.
Third fiddle is the usual seat, not the permanent one. Sites with a flat, steady load and little demand charge, sites without solar, and sites on aggressive time-of-use or wholesale pricing can all see arbitrage climb the order. Sites that run hardest in the evening are the classic case: they buy most of their power in the dearest window, so shifting that consumption onto midday energy is where the widest spread sits.
No. The optimiser reads the prices and plans the cycles. You see the result in the monthly report, not a trading screen.
No. Peak protection and reserve levels outrank arbitrage in the plan. A cycle that would compromise the demand defence does not happen.
Yes. It is the one mode that needs no solar at all: it trades purely on the difference between cheap hours and dear ones.
In one sentence: arbitrage is buying your energy at the day's cheapest hours to use in its dearest, and it is worth exactly as much as the spread on your tariff.
For some businesses the battery's biggest number is not on any bill: it is the outage that did not stop production, spoil the cold room or drop the servers. When the grid fails, the system islands the site and carries the critical load until the grid comes back, then reconnects cleanly.
Backup is a sizing decision. Carrying an entire facility through a long outage takes serious capacity; carrying the circuits that actually cost you money, the cold rooms, the servers, the controls, the security, takes far less. Most sites land in between: the battery islands the site, sheds what can wait, and holds what cannot. That priority list is set during commissioning, with you, not discovered during the first blackout.
The bill savings from the other four modes are easy to model. Outage value is lumpier: spoiled stock, stopped production, restart labour, missed orders, and the occasional insurance excess. If your business has ever added up a blackout afternoon, you already know whether this mode matters to you. Bring that number to a load assessment and we will size the backup case honestly alongside the savings case.
Talk through your outage risk →Grid-connected inverters normally follow the grid's rhythm. When the grid disappears, an islanding-capable system does two things fast: it disconnects the site at the point of supply, so it never backfeeds a dead network, and its inverter switches from following the grid to forming one, setting the voltage and frequency for the site itself. Your equipment keeps seeing normal power because, inside the fence, the battery has become the grid. When utility supply returns and holds stable, the system synchronises and reconnects without anyone touching a switchboard.
Battery capacity buys hours. For outages that run longer, the islanded site can keep harvesting: solar continues charging the battery through the day, stretching the reserve well beyond its nameplate hours. Sites with an existing generator can integrate it as the backstop: the genset charges the battery and supports load while the battery handles the instant response and smooths the ride. That combination beats a generator alone, which takes seconds to start and hates partial loads.
A battery that earned all day and sat empty at 6pm would be useless in an evening blackout. That is why backup reserve is an explicit setting: a slice of capacity the daily optimisation is not allowed to touch. Protecting that reserve is one of the objectives the optimiser weighs every day, so the earning happens with the insurance intact. How big the slice should be is a conversation about your site: what must stay on, for how long, and what an outage costs you.
The changeover is fast enough that most equipment rides through it without noticing. Truly zero-gap requirements, like some medical or data loads, are a specific design conversation, so raise them early.
You trade a slice, not the lot. The reserved capacity sits ready while the rest of the battery keeps earning through the other four modes.
If it is sized for that, yes. Most sites choose to back critical circuits instead, which is far cheaper and covers what an outage actually threatens.
In one sentence: blackout protection means your site forms its own grid the moment the network fails, running the loads you chose to protect until supply returns.
Every other tab on this page describes a battery you own, working to cut your bill. A Power Purchase Agreement flips the model: a funder owns, insures and operates the system, you provide the site and the grid connection, and no capital leaves your business. In most of these deals the battery sits in front of the meter, on the grid side of your connection point, and that changes everything about what it does all day.
Under a PPA or lease structure, the owner pays for the hardware, the installation, the insurance and the maintenance for the life of the agreement. Your business signs a long-term contract instead of a purchase order: typically you either buy the energy the system produces at an agreed rate, or you are paid rent for hosting the asset on your land and connection. Your balance sheet stays clean, and the performance risk sits with the party who owns the machine.
Everything else on this page is behind the meter: the battery sits on your side of the connection, your loads see it, and it earns by changing your bill. An in-front-of-the-meter battery connects on the grid side. It is registered as a market asset, the market operator can see it, and your site's loads never draw from it directly. Same hardware, completely different job description.
It earns from the electricity market itself. Energy arbitrage at wholesale scale: charging when prices are low or negative in the middle of the day, discharging into evening peaks when the grid is short. FCAS, the frequency control ancillary services: the battery is paid to stand ready and respond within seconds when grid frequency drifts, injecting or absorbing power to hold the system at 50 hertz. And network support: soaking up local solar exports and easing congestion so the network can defer building poles and wires.
This is the mindset shift. An in-front-of-the-meter battery does not shave your peaks, does not back up your site in a blackout, and does not care what your factory is doing. It is there for the grid: trading energy, holding frequency, supporting the network. Your benefit arrives through the contract instead: lease income for hosting it, or energy at a better price than your retailer offers, with none of your capital at risk. If you want the battery working for your site as well, that is a different design, and we model both so you can compare them on numbers rather than instinct.
In one sentence: with a PPA, someone else pays for a battery that works for the grid rather than your site, and you earn from hosting it or buy its energy cheaper, with no capital outlay.
Demand charges can be 30 to 50% of a commercial bill, and one bad half-hour sets them. Here is how that works and how a battery caps it.
Peak demand reduction →Most C&I sites land between 4.5 and 7.6 years. What moves a site within that range, what a defensible model includes, and where the value really comes from.
ROI and payback →From 1 September 2026 the NSW scheme covers commercial batteries, and our systems are certified for it. The rest of the picture, honestly told.
Incentives →The battery is the muscle. The intelligence layer decides when to charge, when to shave and what to chase each day.
Ampaura Connect →Cell chemistry, fire suppression, Australian standards and the documents your insurer will ask for.
The hardware →Probably. Six brands tested and running today, open protocols, and up to four weeks to onboard something new.
Compatibility →The commercial battery storage guide covers the full journey: what a battery does for a business, which industries are already benefiting, whether your site is a candidate, and how buying works.
Commercial battery storage, the complete guide →Ask it. A person in Adelaide reads these, and if the question is good we will probably write the answer into this section for the next reader too.