The battery in the garage

On this page

1. A battery has two sizes
2. You never get out what you put in
3. Value 1: the evening the sun does not reach
4. Value 2: buy it cheap, use it dear
5. Value 3: grid services, and why we carry zero
6. The demand charge that is no longer required
7. Quarters instead of hours
8. What it costs, and why nobody knows exactly
9. Fifty per cent — but tied to your own production
10. Check yourself

A battery has two sizes

A battery in a garage is almost always described with a single number: 10 kWh. That number answers only half the question, and it is not always the half that decides whether you will be happy.

Think of a water butt. One measure is how many litres it holds. A second, entirely separate measure is how wide the tap at the bottom is — how fast you can get the water out. A large butt with a thin hose empties slowly, however large it is.

The two numbers are linked by time, exactly as in module 01. The capacity divided by the power is the number of hours the battery lasts if it gives everything it has.

10 kWh ÷ 5 kW = 2 hours

Where did 5 kW come from? From a convention. Our own engine assumes a home battery can give half its capacity in kilowatts — 10 kWh becomes 5 kW. That is an industry rule of thumb, not a measured property, and we say so plainly in the code. Your battery may sit above it or below.

Why the power matters: the house can draw far more than the battery can give. In module 02 we did the arithmetic on the fuse. A house with 20 A per phase and three phases has a ceiling of

3 × 20 A × 230 V = 13 800 W = 13,8 kW

So a 5 kW battery can never carry the whole house when everything comes on at once. Nor should it. The grid fills in the rest, quietly and automatically — but a salesperson who says the battery runs the house during an outage is saying something that depends entirely on the power, not on the kilowatt-hours.

You never get out what you put in

Two losses stand between what you charge in and what you get back. They are different in kind, and they are usually muddled together.

The first is the part you are not allowed to touch. A battery suffers from being emptied completely, so the control system leaves a floor untouched. The share you can actually reach is called the depth of discharge. Our engine reckons with 0.95 — of a 10 kWh nameplate, 9.5 kWh is usable.

The second turns into heat. Every time electricity goes into a battery and back out, some of it is lost on the way, in the cells and in the electronics. The share that comes back is called the round-trip efficiency. Our engine reckons with 0.90, and splits it evenly: the square root of 0.90, about 0.949, on the way in, and the same on the way out.

Put them together and one full lap through a 10 kWh battery looks like this:

10 kWh × 0.95 × 0.90 = 8.55 kWh

So nearly one and a half kilowatt-hours out of ten never became useful. There is nothing wrong with the battery — that is what storage costs in physics.

Why efficiency is not a single number

A datasheet can state efficiency measured in two places: between the battery's own terminals, or between the wall socket and the wall socket. The second is the one your meter sees, and it is always the lower of the two, because the inverter is counted twice.

The figure also falls when the battery works hard and when it is cold. A garage in February is not a laboratory. So do not ask only for the number — ask where and at what power it was measured.

Value 1: the evening the sun does not reach

This is the most certain of the battery's three income streams, and the only one that works by itself. The words kWp, kWh and self-consumption come from module 07 and are used here as they were explained there. The mechanism is a mismatch in time, and it shows best in a picture.

The sun delivers in the middle of the day, when the house is empty. The house spends electricity in the morning and above all in the evening, when the sun has gone. Without a battery the surplus goes out to the grid at midday, and the same house buys electricity back six hours later — at an entirely different price.

A summer day: the sun, the house and the gap between them

What is one such move worth? The difference between what you avoid buying and what you would have been paid for selling. We reckon with our own engine's 2026 price stack and assume the spot price in that hour is 50 öre. The buying side first:

(0.50 spot + 0.05 mark-up + 0.36 energy tax + 0.40 network) × 1.25 VAT = 1.64 kr/kWh

And the selling side, which since 1 January 2026 is short: the spot price plus the compensation for network benefit, which our engine sets at 4 öre.

0.50 + 0.04 = 0.54 kr/kWh  →  1.64 − 0.54 = 1.10 kr/kWh

A little over one krona for every kilowatt-hour the battery manages to move. Of the five figures above, four are our engine's 2026 defaults and one — the spot price — is our example. The energy tax of 36 öre excluding VAT is Skatteverket's 2026 rate, the network charge of 40 öre is a typical value for a house in price area 3 with a documented spread between 15 and 55 öre, and the 5 öre mark-up is uncited in our own code.

That gap was much narrower two years ago. Up to 31 December 2025 anyone feeding electricity in received 60 öre per kWh as a tax credit, and it disappeared on 1 January 2026 (Skatteverket, prop. 2024/25:109). Export today is worth between a third and a half of what it was, depending on the spot price that hour: 0.54 kr instead of 1.14 kr in our example above, so 47 per cent. That is exactly why interest in batteries rose: keeping the electricity became worth more than selling it.

How many kilowatt-hours can be moved? That is where the certain answers run out. Sweden publishes no official statistic for self-consumption, and the estimates that exist partly contradict each other. A house without a battery uses something in the order of 30 to 50 per cent of its own production itself, and what drives it is the size of the system relative to consumption — not where in the country the house stands.

A battery adds roughly 15 to 30 percentage points at the sizes a house actually buys. The most solid Swedish study — Nyholm and co-authors, Chalmers, in Applied Energy 2016, modelled on 2 104 real Swedish households — puts the practical ceiling at 18 to 48 percentage points, but that ceiling is reached only with batteries far larger than the domestic market's. The study's most useful finding is a different one: houses that already use a lot of their own sunshine gain the least from a battery.

The rule of thumb about one kilowatt-hour per kilowatt

The industry often repeats that a battery of 1 kWh per installed kW of solar adds about 20 percentage points of self-consumption. The figure is a reasonable summary of the research above, but we have not been able to trace it to any primary source — it circulates on installers' own pages.

Treat it as a rule of thumb and nothing more. If you want to know what your own house gains, there is only one honest way: your own hourly readings, day by day, against a production curve for your own roof.

Value 2: buy it cheap, use it dear

The second income stream needs no sun at all. The battery charges when electricity is cheap, at night or early morning, and discharges when it is dear. It is real, it is modest, and it requires two things not every house has.

The first is a contract that prices the hour. On a variable monthly price every kilowatt-hour in the month costs exactly the same, whenever in the day you used it — and then moving electricity within the day is worth precisely nothing. According to SCB, 61.0 per cent of Swedish households were on a variable monthly price in August 2026, against 14.1 per cent on a quarter- or hourly-price contract. So for most households this income is zero today, not because the battery is poor but because the contract cannot tell one hour from another. Module 09 is about exactly that choice.

The second is a spread worth chasing. It is not the average price that pays, but the distance between the day's cheapest and dearest hours. During 2024 the annual average sat between 0.28 and 0.57 kr/kWh depending on the price area, while the dearest single hour reached 8.16 kr/kWh in price area 4 and the cheapest fell to −0.69 kr/kWh. Negative prices in the middle of a summer day are now routine — and that is precisely when solar exports.

Work through a day where the night sits at 30 öre and the evening at 90 öre of spot. Tax, network charge and mark-up are paid on every kilowatt-hour bought, whatever the hour, so they appear on both sides:

night: (0.30 + 0.81) × 1.25 = 1.39 kr  ·  evening: (0.90 + 0.81) × 1.25 = 2.14 kr

But to deliver one kilowatt-hour in the evening you must buy more than one at night, because a tenth of it turns into heat. That makes the arithmetic this:

1 ÷ 0.90 = 1.11 kWh  ·  1.11 × 1.39 = 1.54 kr  ·  2.14 − 1.54 = 0.60 kr

Sixty öre on a spread of sixty öre. Notice what the loss does: the extra you must buy, 0.11 kWh, is paid for at the full price with tax and VAT — about 15 öre. So until the spot price differs by around 12 öre between night and evening, the battery has not earned a krona by moving anything. On some days the spread is several kronor. On other days it is not there at all.

One thing pulls the right way here: since the exchanges trade per quarter instead of per hour, the peaks are sharper, because they are no longer averaged across sixty minutes. More on that further down.

Value 3: grid services, and why we carry zero

The third income stream is the one that sells batteries, and it is the one we treat most harshly. Here is what it is.

The grid has to hold its frequency near 50 hertz every second. Svenska kraftnät therefore buys readiness: someone promises to react within fractions of a second if the frequency slips. A home battery can do that, but not on its own — it happens through an aggregator that gathers thousands of batteries into one shared resource. You are paid for standing ready, not for delivered electricity. FCR-D is the most talked-about of these markets.

And this is what has happened to the price. The magazine Energi reported on 23 June 2026, drawing on Svenska kraftnät's forecasts:

The reason is not that the need disappeared. Svenska kraftnät's outlook to 2030 has demand for FCR fairly stable, and the need for mFRR growing from about 1 200 to 1 800 MW upwards and from 1 000 to 1 500 MW downwards. It is the price that has fallen, not the volume — and the cause is the home batteries themselves. Around 1 200 MW of battery storage is now prequalified for these markets. Supply ran away from demand, and an academic forecast points further down, towards about 4 euro per megawatt by 2030.

The home battery's own success is what destroyed the price of the thing the home battery was going to earn money on.

Our engine sets the income from grid services to 0 kronor, deliberately. Four reasons, and they all hold at once. The income is contested — the unit in the best source available is uncertain. It is falling — we know the direction but not where it stops. It is contract-specific — it depends on which aggregator you sign with and what terms apply that month. And it reaches nobody who has not signed up — a battery in a garage with no aggregator agreement earns exactly nothing on this market.

There is a figure in circulation worth seeing, precisely because it shows what is at stake. One aggregator published about 52 kr per kW per month excluding VAT from its customers' 2025 results — roughly 600 kr per kW per year, so in the order of 3 100 kr a year for a battery of 10 kWh and 5 kW. That is an aggregator's own figure about its own customers, not official statistics, and we do not carry it in the code. But notice how much it would move: 3 100 kr a year is in the same neighbourhood as the entire self-consumption value for many houses.

Three income streams, in order of how certain they are

The demand charge that is no longer required

A demand charge bills you for your highest power during a month, not only for your kilowatt-hours. A battery can shave such a peak, which is why demand charges are one of the commonest arguments in a battery offer. The argument rested for a long time on a requirement. That requirement no longer exists.

So say it carefully: demand charges are permitted, some grid companies have them, and there is today no rule saying yours must. We will not put a number on how many have introduced them — two figures circulate and cannot be reconciled from the sources we have read. The right answer is to look up your own grid company's tariff, not to generalise. Our own engine carries the demand charge as 0 kr per kW, and says in the code that it is not modelled.

Quarters instead of hours

Two dates have changed how a house with solar panels and a battery is measured. At midnight between 18 and 19 March 2025 imbalance settlement moved from 60 to 15 minutes. And from 1 October 2025 the European day-ahead market trades in quarters, which is when quarter-hourly prices for consumers begin to exist. All 96 quarters of the next day are published at 13:00 the day before.

Ei also requires every retailer with more than 200 000 customers to offer a quarter-price contract. Fixed prices and variable monthly prices are untouched — there you will notice nothing.

What does that mean for a house with a battery? The netting between what you feed in and what you draw out now happens per quarter instead of per hour. Under hourly metering a lunchtime hour in which the house both exported (while the oven was off) and imported (while it was on) could be netted into a single number, and the export never appeared. Now some of those mismatches break through as simultaneous import and export within the same hour.

The consequence is that measured self-consumption falls somewhat and measured export rises — with nothing at all changed physically in the house. For the battery it pulls the other way: the value rises, both because the price spikes are sharper and because a battery that can react within a quarter can now capture the netting that used to be free.

What it costs, and why nobody knows exactly

We must start here with an uncomfortable sentence. There is no primary source for what a home battery costs per kilowatt-hour in Sweden. IEA-PVPS, otherwise the citable series for Swedish solar, states plainly that there are no official statistics on battery storage in Sweden. It estimates 650 to 800 MW of battery capacity installed during 2024, but publishes no price for the domestic market.

What exists is the installers' own pages, 2025 and 2026: roughly 3 500 to 7 000 kr per kWh of usable capacity including VAT and installation, and a 10 kWh battery at around 60 000 to 100 000 kr before the green deduction. The direction has been clearly downwards from 9 000 to 10 000 kr/kWh in 2020 — but an installer survey for the first half of 2025 reported that premium battery prices rose about 10 per cent, while panel prices fell 11 to 14 and inverters 13. These are commercial pages, not statistics.

Our own engine carries 8 000 kr per kWh turnkey installed. That figure comes from Swedish consumer guides for 2026 that put the range at 7 000 to 9 000, and 8 000 is the middle. In the code it is described as a national band, not a quote — and it sits above the installers' per-kilowatt-hour range above, which is precisely what "no official statistics" means in practice. Two sets of figures about the same thing do not agree, and neither is wrong enough to throw away.

Work through an example anyway, with the figures in the open. A 10 kWh battery at our band value, then the green deduction of 50 per cent of 97 per cent of the invoice, which is Skatteverket's simplified method for turnkey fixed-price contracts:

10 × 8 000 = 80 000 kr  ·  80 000 × 0.97 × 0.50 = 38 800 kr  ·  left: 41 200 kr

41 200 kr sits inside the range installers give after the deduction, 30 000 to 50 000 kr, which is a comfort but not a proof. Now set that sum against the value we worked out earlier. One full lap gave 8.55 kWh, and every moved kilowatt-hour was worth 1.10 kr:

8.55 × 1.10 = 9.41 kr per lap  ·  41 200 ÷ 10 years = 4 120 kr/yr  ·  4 120 ÷ 9.41 = 438 laps/yr

438 full laps a year is more than one every day of the year — and no Swedish year offers that, because the surplus does not exist in November. This is not a payback period. It is the size of the hurdle, under assumptions that are all written out here: no interest, no ageing, a single price level, and a ten-year life taken from a typical LFP warranty. Our engine, for that matter, carries no default lifetime at all — it is the advisor's figure, not ours. Change any of the assumptions and the hurdle moves. Module 11 shows how to do that properly.

And notice what the hurdle says about the rest of this module. Self-consumption alone rarely carries a battery at our band price. It is the arbitrage, a demand charge at your own grid company, a price lower than 8 000 kr per kWh — or a grid-service market that turns — that decides the case. Three of those four are uncertain. That is the entire point.

Fifty per cent — but tied to your own production

The green deduction for a battery is 50 per cent of labour and materials, against 15 per cent for solar panels. The solar rate was cut from 20 to 15 per cent on 1 July 2025 through prop. 2024/25:109; the battery and the charging point were untouched and remain at 50.

Four rules decide whether the deduction actually lands with you. The ceiling is 50 000 kr per person per calendar year, so a couple owning jointly can reach 100 000 kr if the invoice states each share. Private individuals only — never a housing association, never a company. The final payment decides which rules apply, not when the work started. And the most important of the four: the half rate applies to lagring av egenproducerad elenergi, storage of self-produced electricity — so only where the storage is connected to an installation for electricity you produce yourself.

The last rule is worth reading twice, because it hits exactly the case this module has described. A battery bought purely to trade on the spot price, in a house with no production of its own, is a different case from a battery behind a solar roof. If you are thinking in that direction: ask Skatteverket before the invoice is written, not after.

Check yourself