Batteries
The mode a battery is left in matters more than another kilowatt-hour
Storage hardware is inert without a control strategy, and the default strategy shipped in the box is designed for the average household rather than for yours.
By Arjun Nair3 min read
The hardware does nothing on its own
A battery is a container. What decides whether it earns anything is the software choosing when to fill it and when to empty it, and that decision is made every few minutes by a controller most owners never look at. The specification sheet describes the container. It says nothing about the decisions.
Most systems ship with a default strategy that is sensible for a typical household on a flat tariff: store any generation the house isn’t using, then discharge to cover consumption until the battery is empty. It requires no configuration and no forecasting, and it does a reasonable job.
It is also the least sophisticated thing the hardware can do, and on some tariffs it is close to the worst.
Self-consumption and time-shifting want opposite behaviour
A strategy aimed at using your own generation wants the battery empty by morning so there is room for the day ahead. A strategy aimed at exploiting a cheap overnight period wants it full by morning so the expensive daytime hours are covered. Those two goals conflict directly, and a controller cannot pursue both at once.
Which one is right for a household depends entirely on the pricing it faces, and pricing structures vary so much between countries that no general recommendation survives contact with a real tariff. What is general is the conflict itself, and the fact that the box has to be told which side to take.
Systems that offer a blended mode try to reserve part of the capacity for each objective. That works, and it does each job with a smaller battery than either alone would have.
Forecasting is what separates the good controllers
The interesting decision is how much cheap overnight energy to buy, and answering it properly requires a guess at tomorrow. Charge too little and the house imports at an expensive time. Charge too much and the battery is still full when the sun comes up, so generation is exported instead of stored.
Better controllers take a solar forecast for the location and a model of the household’s recent consumption, then choose an overnight target from the two. When the forecast is right this is clearly superior to a fixed rule, and when it is wrong the system behaves worse than a simple one would have.
There is a second forecast buried in the same calculation, and it is the household one. A controller that assumes tomorrow will resemble last week is wrong on a public holiday, wrong on the day guests arrive, and wrong throughout the fortnight everybody is away. Some systems allow those periods to be flagged in advance, and hardly anyone does it.
Forecasts are least reliable in exactly the changeable weather where the decision matters most. That is worth knowing before treating an optimisation feature as a guaranteed improvement.
Reserves and priorities are the settings people forget
Most systems allow a portion of capacity to be held back for a power cut, and every unit reserved is a unit not available for ordinary use. Holding a large reserve in a region with a stable network is a continuous cost paid against a rare event, and it is worth setting deliberately rather than leaving at whatever the installer chose.
Priority order matters too where a system also feeds a hot water diverter or a car charger. Which load gets the surplus first is a configurable decision with a real consequence, and the default order is a guess about a household the manufacturer has never met.
None of these settings are difficult. They are simply buried, and almost nobody revisits them after the first week.
Reviewing the strategy is worth more than upgrading the hardware
A household whose circumstances have changed — a new tariff, a car, someone now working from home — is very likely running a strategy chosen for the household it used to be. That mismatch costs more than the difference between two battery products, and correcting it costs nothing.
A practical way to judge whether the current strategy is working is to look at how often the battery reaches either extreme over a month. A unit that never fills is larger than the array can serve, one that never empties is larger than the household needs, and one that does both on most days is doing exactly the job it was bought for.
The honest caveat is that fiddling has diminishing returns fast. Once the mode matches the tariff, the reserve is set sensibly and the priorities reflect the actual loads, further adjustment is entertainment rather than saving. Set it properly, then look at it once a year, and spend the rest of your attention elsewhere.
Common questions
Should I let the battery charge from the grid overnight?
Only if your tariff has a genuine difference between periods that is wide enough to cover the losses of storing and retrieving the energy. On a flat rate it achieves nothing and consumes cycles for no return. Where a real difference exists, overnight charging can be the largest single benefit the battery provides.
Does changing modes frequently harm the battery?
Switching mode is a software decision and harmless in itself. What affects the battery is how much it is cycled and how deeply, so a strategy that puts it through more full cycles each day will age it faster regardless of what that strategy is called. That extra wear may still be worth it.
Is a third-party optimisation service worth paying for?
It depends on whether your tariff and consumption give it room to work. On a flat rate with a simple household there is very little for an optimiser to find, and any ongoing fee is hard to justify. On a complex time-varying tariff with a large battery and an electric car, the potential is real, though it should be measured rather than assumed.
Staff writer, Power Your Roof
Arjun has been reporting on solar basics, batteries, bills & tariffs since long before it was fashionable and thinks most subjects are more interesting once you know how they work.
