Battery cells have become a commodity. Two systems from different manufacturers, both lithium iron phosphate, both around 13 kilowatt hours, both carrying ten year warranties, will store and release energy in almost exactly the same way.
What separates them is the code deciding when to do it.
That distinction gets almost no attention in quotes, which tend to list capacity, brand and price. Yet the control logic determines whether a battery captures the expensive hours or sleeps through them, whether it has charge left when the grid drops out, and whether it keeps working when a vendor’s cloud service does not.
What the software is actually deciding
A battery makes several decisions continuously. Whether to charge from solar or hold. Whether to discharge into household load or preserve capacity. Whether to import cheap grid power overnight rather than wait for tomorrow’s sun. How much reserve to hold back for a possible outage.
None of those are hardware questions. They are scheduling problems, and the quality of the answer depends on what information the controller has and what rules it follows.
The simplest systems run a fixed priority: solar to load, surplus to battery, surplus beyond that to grid, then battery to load after dark. That works, and for a household on a flat tariff it may be close to optimal.
It stops being optimal the moment pricing varies by time of day, which is now the normal case. A controller that knows your tariff can charge from the grid during a cheap midday window and hold that energy for the evening peak, which is a completely different behaviour from one simply reacting to whatever the panels are doing. A solar and battery Adelaide install now involves more decisions about software behaviour than about which cells are inside the cabinet.
The network is already talking to your inverter
This is the part most people do not realise has happened.
In South Australia, every new or upgraded exporting system since 28 September 2020 must have a nominated Relevant Agent and be capable of being remotely disconnected by that agent. From 1 July 2023, dynamic export requirements went further, requiring all new exporting generation systems to be capable of having their export limits updated remotely.
In practice that means choosing a connection type. New solar and battery installations select either a fixed export limit, commonly 1.5 kW or zero, or Flexible Exports, where a site-wide limit varies dynamically between zero and 10 kW per phase based on the network’s assessment of local capacity.
Flexible Exports requires the inverter to maintain an internet connection so limits can be adjusted automatically. During the SA Power Networks trial, systems fell back to the 1.5 kW fixed limit if communications dropped out, which gave customers a direct incentive to maintain reliable connectivity.
The practical implication is worth stating plainly. Your home network is now part of your energy infrastructure. An inverter sitting on marginal wifi at the far end of the house is not a minor commissioning detail.
Forecast-based control
The more capable energy management systems now incorporate forecasting: solar generation predictions from weather data, and in some cases wholesale price signals.
The logic is intuitive once you see it. If tomorrow is forecast to be overcast, there is value in arriving at morning with more charge than usual, because the panels will not refill the battery. If a heatwave is expected, pre-cooling the building while power is cheap and holding battery capacity for the afternoon is worth more than discharging early.
Forecast-aware control is genuinely useful in a state where daytime prices frequently collapse and evening prices spike. It is also where implementations vary enormously, from sophisticated optimisation to a marketing claim attached to a fixed schedule.
The cloud dependency problem
Here is the question almost nobody asks: what happens if the manufacturer’s servers go offline?
For some systems, very little. Scheduling runs on the device, the app is only a window into it, and losing connectivity means losing visibility rather than function. For others, the optimisation logic lives in the vendor’s cloud, and an outage drops the system back to basic behaviour.
The longer term version of the same question matters more. Solar and storage brands have exited the Australian market before. When that happens, the hardware keeps working, but firmware updates stop, the app degrades, and any cloud dependent features become unreliable.
Two things reduce that exposure. Ask whether core control logic runs locally on the inverter or gateway. Ask whether the system supports an open protocol or third party monitoring, so a future owner is not entirely dependent on one vendor staying interested.
Questions worth asking before signing
Before capacity or brand, ask what the system will actually do.
What charging modes does it support, and can it charge from the grid during a cheap window? Can schedules be set by time of day, and changed when your tariff changes? Does it hold a reserve for backup, and is that reserve adjustable?
Does the control logic run on the device or in the cloud? What happens during an internet outage?
Is the system dynamic export capable and compatible with the network’s flexible connection option, and who is nominated as the Relevant Agent?
What does the monitoring show: instantaneous flows only, or historical data at a useful resolution? You cannot evaluate whether a system is behaving sensibly without data.
And if backup matters, confirm it is included and which circuits it covers, since plenty of installations have no backup capability unless specified.
The quiet shift
Rooftop solar used to be a hardware purchase. You bought panels, they made power, and the relationship ended there.
Storage turned it into something closer to a networked appliance, with firmware, connectivity requirements, remote control obligations and a vendor relationship that has to outlast the warranty. That is a meaningful change in what people are actually buying, and the quote in front of them rarely mentions it.
