Grid charging: when it pays and when it doesn't
Filling a battery from the grid is the single most useful thing it can do on a smart tariff, and one of the most wasteful things it can do on the wrong one. There is one test that decides which, and one complication that solar adds.
What grid charging means
A home battery can fill from two places: the panels on the roof or the grid. Grid charging is the second. Energy is bought from your supplier, stored, and used later. Nothing about it is free; you pay the import price at the moment you charge, and you lose a little of what you bought in the round trip.
On a Powerwall, grid charging is a permission rather than a schedule. With it off, the battery fills only from solar surplus. With it on, and the Powerwall in Time-Based Control, the Powerwall may buy in periods its tariff shows as cheap and hold that energy for periods it shows as expensive. One exception is worth knowing: raising the backup reserve above the current charge makes the Powerwall top up to it, and owners report it will use the grid to do so even with grid charging off. How Powerwall time-based control works covers the settings.
The one condition
Grid charging pays when the price you would otherwise pay at the time of use is higher than the price you charged at, by more than the round-trip loss. Written out:
It pays if the price later ÷ the price now is greater than 1 ÷ efficiency.
With a round-trip efficiency of around 90 %, that means the later price has to be at least about 11 % higher than the price now. Anything less, and you have paid to store energy that would have been cheaper to buy when you needed it.
In practice the gap is either much bigger than that or it does not exist. On a tariff with a cheap overnight window, the day rate is several times the overnight rate, and the 11 % is a rounding error. On a half-hourly tariff like Agile, the evening peak is usually a multiple of the cheapest overnight slot, though on some days the gap narrows and the arithmetic deserves a second look. Round-trip efficiency: where the missing ten percent goes explains where the loss comes from.
Why it never pays on a flat tariff
On a flat tariff the price later and the price now are the same, so the ratio is exactly 1, and 1 is never greater than 1.11. Every kWh you store from the grid costs about 11 % more than buying it at the moment you need it, plus a little standby loss while it waits. Grid charging on a flat tariff is a small, steady way to lose money.
The one reason to do it anyway is security. Before a storm, a full battery is worth more than the few pence of losses, and Storm Watch does exactly this. That is insurance, not arbitrage, and it is fine to pay for it. Just do not leave it on all winter thinking it saves anything.
The export-tariff catch
A tempting extension is to buy cheap overnight and sell dear in the evening. Whether you are allowed to depends on the export tariff's terms. Some tariffs pay only for energy you generated yourself: Octopus Outgoing is one, and its terms exclude exporting energy that came from the grid. Others, Flux among them, pay for anything the meter sees leaving the house, and are designed for exactly this pattern.
The condition is the same one as before, with the export price standing in for the price later: buying at the cheap import rate and selling at the peak export rate pays only if the export rate beats the import rate by more than the round trip. On a Flux-style tariff at the evening peak it usually does. On a flat export rate it almost never does, because flat export rates sit below almost every import rate. Check the terms before you build a plan around it; UK export tariffs explained goes through the types.
The solar conflict
Here is the complication. If the battery is full at dawn, there is nowhere for the day's solar surplus to go. It is exported at the export rate, which is low, or if you have an export limit it is clipped and lost altogether. Either way you have replaced free energy from the roof with paid energy from the grid, and that is worse than not charging at all.
So grid charging and solar compete for the same space in the battery, and the right amount to charge overnight is the amount the day will need after the panels have done their part. On a bright day that may be nothing. On a grey day it may be everything.
The forecast arithmetic
The target for the overnight charge is roughly:
Charge to 100 % minus the solar surplus you expect tomorrow, and never below the reserve.
Expected surplus is the solar forecast less what the house will use while the sun is up. Some worked figures for a 13.5 kWh battery:
| Tomorrow's forecast | Daytime use | Expected surplus | Overnight target |
|---|---|---|---|
| Bright, 18 kWh | 6 kWh | 12 kWh (about 90 %) | Reserve only |
| Mixed, 9 kWh | 6 kWh | 3 kWh (about 20 %) | Around 80 % |
| Grey, 2 kWh | 6 kWh | None | 100 % |
Forecasts are wrong more often than you would like, so leave a margin in the direction that costs least. On a cheap-window tariff, over-charging by 10 % costs you the round trip on a kWh or two; under-charging costs you a kWh or two at the day rate. The second is usually worse, so err towards charging a little more than the arithmetic says. Why your solar forecast is wrong explains how much to trust the number.
Making it happen
What the Tesla app's permission does
The grid charging switch in the Tesla app, where your region offers it, is a permission. It does not tell the Powerwall when to charge or how much. In Time-Based Control the Powerwall decides both from the tariff it has been given and its own forecast, and it will only buy in periods its tariff marks as cheap. In Self-Powered mode the switch is largely moot, because that mode fills from solar and does not plan around prices.
That means the switch alone cannot implement the forecast arithmetic above. The Powerwall does not know your reserve should be 80 % tonight and 20 % tomorrow. To get that, something outside the app has to change the settings each evening.
A rule that charges only when tomorrow looks poor
The simplest useful automation is a pair of rules:
- When the clock reaches 23:30, only if solar tomorrow is below 8 kWh, then grid charging on and reserve 100 %.
- When the clock reaches 05:30, then grid charging off and reserve 20 %.
Adjust the threshold to your own array and household. A second pair with a middle threshold and a middle reserve gets you close to the table above. The point is that the decision is made fresh every night from the forecast, not once in October and forgotten.
In SunHarvest
The solar tomorrow signal is available to any rule, so the pair above can be written as described. The force charge until action does the same job without touching the Powerwall's tariff by hand: it makes the Powerwall fill flat out until the time you set, keeps battery export off meanwhile, and then puts every setting back on its own. The AI planner takes the arithmetic further: its context pack carries the forecast, the expected surplus above your export cap, the kWh needed to fill the battery and the measured charge rate, and the sums are done in code before the planner sees them. The load model learns how much your home usually uses at each hour, so the daytime-use figure in the table comes from your own history rather than a guess.
General information, not financial or electrical advice. Tariff terms, prices, warranties and connection rules change and differ by supplier, region and installation; check the current documents for yours. SunHarvest is not affiliated with Tesla or Octopus Energy.