Solar panels on the roof

On this page

1. kWp is not kWh
2. The Swedish year is brutally uneven
3. Which roof?
4. Self-consumption — the figure that decides the money
5. What the surplus is worth, and what changed
6. The green deduction as it stands
7. What it costs
8. Check yourself

kWp is not kWh

This is the most misunderstood pair in the whole solar trade, and it is worth stopping at. The quote talks about a 10 kWp system. Many people hear ten thousand kilowatt-hours. Those are two entirely different things, and the gap between them is large.

We have already met this difference in this course, in another setting. kWp is power — how fast the roof can deliver when everything is perfect. kWh is energy — how much actually came out of it over a year. Just as with a car: the size of the engine says nothing about how far you drove.

The exchange rate between them has a name: specific yield, kilowatt-hours per installed kilowatt-peak per year. And for Sweden there are two figures worth knowing, because they answer different questions.

So for that ten-kilowatt system, using what Swedish roofs actually deliver:

10 kWp × 850 = 8 500 kWh

Not ten thousand. Eight and a half thousand, and that is an ordinary roof rather than a poor one. A genuinely good south-facing roof lands nearer 10 000 kWh — but then you have also done away with the shade from the neighbour's birch.

The surprise: north and south lie close together

Almost everyone believes a roof in Skåne gives twice as much as one in Norrbotten. It does not. For a south-facing roof at a good tilt, the difference across the whole country is about one tenth.

Why latitude matters so little

Two things cancel each other out. Further north the sun stands lower, which is a drawback. But the summer days are far longer, and the northern inland air is clearer — fewer clouds, more direct sun.

Profu's report therefore splits Stockholm and Malmö into coast and inland, because cloud matters more than latitude does. The coast wins in both cases.

What does not cancel out is the shape of the year. Further north the same annual total is spread far more unevenly — a stronger summer and an almost empty winter. That is the next section.

The Swedish year is brutally uneven

Here is the figure that does the most work in this whole module. It shows one year from 1 kWp in Stockholm, month by month, calculated in the European Commission's PVGIS.

One year from 1 kWp in Stockholm, kWh per month

Do the arithmetic on the three pale bars yourself. November, December and January together:

23 + 13 + 19 = 55 of 991 → 5,5 %

And the summer half, April to September:

124 + 139 + 141 + 136 + 117 + 90 = 747 → 75 %

So three quarters of the year's harvest arrives in one half of the year, and the three darkest months together carry about a twentieth. The further north you go, the sharper the contrast gets — even though the annual total barely moves.

This is the section that puts to rest the dream of unplugging the house from the grid in winter. Take the ten-kilowatt roof in Stockholm again. December gives 13 kWh per kWp:

10 × 13 = 130 kWh ÷ 31 ≈ 4,2 kWh per day

Four point two kilowatt-hours a day, on average, from a large roof. That is roughly what a ten-minute shower uses. A whole December day's harvest, equal to one shower — and the house still has to be heated.

How we worked out the shower

A shower head delivers about 12 litres a minute. The water arrives at roughly 10 °C and has to reach 40 °C — a lift of 30 degrees. Heating one litre of water by one degree takes 1.16 watt-hours.

12 × 30 × 1,16 × 10 = 4 176 Wh

So just under 4.2 kWh for ten minutes. The 1.16 is water's heat capacity, a physical constant — everything else is calculated here on the page.

The conclusion is not that solar is pointless in Sweden. It is that solar panels and winter heating are two separate questions. The roof delivers when the house needs least, and that uncomfortable truth is what all Swedish solar economics rests on.

Which roof?

Two properties of the roof matter: which way it points and how steeply it slopes. One of them is forgiving. The other is not.

The tilt is forgiving. Profu's report for Energimyndigheten concludes that production is much the same for anything between roughly 20 and 50 degrees facing south. Only when the panel approaches vertical, as on a wall, does it drop materially.

The direction is not. But even there the difference is less dramatic than many people think.

The same 1 kWp in Stockholm, different roofs — kWh per year

• South, 40°990

• South, 20°927

• Flat roof, panels laid flat780

• East, 30°760

• West, 30°753

• North, 30°493

PVGIS 5.3, Stockholm, 1 kWp with 14 per cent system losses, retrieved 23 September 2026. Profu's 2018 report measures the same thing slightly differently and finds an east–west roof gives 81–84 per cent of a south-facing one. Both sources say the same thing: an east–west roof is a modest loss, not a refusal.

So an east–west roof is no disaster. It gives about four fifths of a south-facing one, and it has a property that may be worth more than the last twenty per cent: it spreads production across morning and evening instead of piling it all around lunchtime. That is exactly when the house uses electricity. Why this matters for the money comes in the following section.

A flat roof is treated in Profu's report as a first-class position, because there the panels are assumed to be mounted on frames at the best tilt and direction. A flat roof is therefore not a production question — it is a cost and space question, because the rows have to stand apart so they do not shade one another. Lay the panels straight down on the roof instead and you land at about 79 per cent.

A north-facing roof is where it stops adding up: between roughly half and three quarters of a south-facing one, depending on tilt. And shade is worse than any of this — but shade shows up in no table, only on your own roof.

Self-consumption — the figure that decides the money

Now comes the thing that decides whether a solar roof pays or not, and it is neither the direction nor the kilowatt-peak. It is what share of the production the house uses itself, in the same moment it is made. This is called self-consumption.

The reason is simple, and it has become far more important since 2026. A kilowatt-hour you use yourself is one you do not have to buy — and a bought kilowatt-hour costs you the whole stack of energy price, markup, energy tax, network charge and VAT. A kilowatt-hour you sell earns you the spot price and a few öre on top. The figures come in the next section; the difference is in the order of six to ten times.

A summer day: what the roof makes, and what the house uses

The picture also explains why an east–west roof can be the better deal than a south-facing one despite producing fewer kilowatt-hours. It moves part of the hill towards morning and evening, where the house's own peaks sit — and so turns a larger share of the production into self-consumption.

How large is self-consumption? Nobody knows for certain

Here we have to be honest in a way a salesperson rarely is. There is no official Swedish statistic for self-consumption in a house. Energimyndigheten publishes no such figure. On its Solelportalen it asks the homeowner instead to analyse their own hourly readings — season, month, week, day — and offers only one rule of thumb: build roughly so that annual production matches annual consumption.

The estimates in circulation come mainly from a research project at Mälardalen University, funded by Energimyndigheten. They look like this — and they do not quite agree with one another, which we think you should be told:

What drives self-consumption is therefore not where you live, but two other things: how large the roof is relative to what the house uses, and how the house is heated. A house heated with electricity has more load to shift but also a summer surplus no night can swallow. A house heated another way starts high and reaches the ceiling quickly.

The strongest lever on self-consumption is not the battery. It is not building bigger than the house needs.

A battery helps, and a Swedish study from Chalmers has measured how much: it modelled 2 104 real Swedish households and found a battery raises self-consumption by 18 to 48 percentage points, depending on the household's load pattern. The upper end of that range needs batteries considerably larger than houses buy. The study's most important observation may be this: the households that already use a lot themselves gain the least from a battery.

Why the figure fell on 1 October 2025, with nothing changing in the house

Since 1 October 2025 electricity has been traded per quarter of an hour instead of per hour. Imbalances had already been settled per quarter from March 2025. Self-consumption is therefore netted quarter by quarter.

Previously a lunchtime hour in which the house both exported and imported was evened out into a single figure, and the export never appeared. Now some of those mismatches surface as simultaneous import and export within the same hour.

Measured self-consumption therefore falls somewhat, and measured export rises — with no physical change at all. The direction is clear. We have found no source quantifying the size, so we put no percentage on it.

What the surplus is worth, and what changed

This is the most important current fact in the whole module, and it is recent enough that a great deal of the material still out there on the internet has it wrong.

Up to and including 2025, a micro-producer received a 60 öre per kilowatt-hour tax credit for electricity fed into the grid. On 1 January 2026 it disappeared. Skatteverket says so plainly, and the Riksdag decided it on 15 May 2025 through proposition 2024/25:109. The state expected to save 680 million kronor a year, and put the loss for an average household at around 3 300 kronor a year.

Two payments remain, and only two:

Put figures on it. Say your exported kilowatt-hour meets a spot price of 30 öre — on a summer midday it is often lower than that — and the grid company pays 2 öre of network benefit:

0,30 + 0,02 = 0,32 kr/kWh (2026)

0,32 + 0,60 = 0,92 kr/kWh (up to and including 2025)

The same kilowatt-hour, the same roof, two thirds of the value gone:

0,32 ÷ 0,92 ≈ 35 %

Now compare that with what you avoid paying when you use the kilowatt-hour yourself. In 2024 a Swedish household paid between 1.9 and 3.1 kronor per kilowatt-hour all in — network charge and taxes included. Part of that is fixed monthly fees that stay whatever you do, so what you truly avoid is somewhat lower. But the order of magnitude is clear:

1,90 ÷ 0,32 ≈ 6    3,10 ÷ 0,32 ≈ 10

A kilowatt-hour you use yourself is worth six to ten times one you sell. That is the entire argument for self-consumption, and it is also why interest in batteries rose sharply during 2025 — 78 000 applications for the green deduction for a battery that year, forty per cent more than the year before.

There is one more asymmetry, rarely mentioned, and it pulls the same way. Your production coincides with low prices: sunny summer middays when everyone else's panels are producing too, and the price sometimes goes below zero. Your consumption coincides with high prices. So the surplus sells below the annual average price, while the electricity you avoid buying would have cost above it. To see how that lands in your own house, the measure calculator works from your own hourly readings.

The green deduction as it stands

This is the support that remains, and the installer deducts it straight off the invoice — you do not have to apply for anything afterwards. Skatteverket's three rates:

The ceiling is per person, not per house. If you own the property together you can therefore reach 100 000 kronor — but only if the invoice states each person's share. If the whole invoice is in one name, the whole deduction is that person's too, and the 50 000 kronor ceiling bites. It is a formality that is easy to ask for before the work starts and awkward to repair afterwards.

We have found no decision changing the rates for 2027. But that is an absence of news, not a promise — the rates have been changed once already, and anyone calculating over the long term should check against the autumn budget before treating the figure as settled.

What it costs

There is one citable Swedish price, and it comes from IEA-PVPS's annual survey, in which Swedish installation companies report their typical turnkey prices. For what concerns us — a house roof of 5 to 10 kW — the 2024 average was 14.70 kronor per watt excluding VAT.

Take that all the way to what appears on your own paperwork, for a 10 kWp system:

10 000 W × 14,70 = 147 000 kr

147 000 × 1,25 = 183 750 kr

183 750 × 0,15 = 27 563 kr

183 750 − 27 563 ≈ 156 000 kr

Now the caveat, and it matters more than the figures. A national average is a band, never a quote. This is a mean across the country's installation companies; real quotes spread widely with the awkwardness of the roof — scaffolding, how many penetrations, how far to the consumer unit, whether the panels go up a ladder or on a lift.

Two honest checks on the arithmetic above. Installer sites in 2026 give between 15 000 and 20 000 kronor per kWp including VAT, and 125 000 to 155 000 kronor for ten kilowatts after the deduction — so our chain lands at the upper end of that band. And the 27 563 kronor deduction is probably a little generous, since the 15 per cent is calculated on labour and materials rather than on the whole turnkey price. The trade therefore tends to write 14.55 per cent; Skatteverket's figure is 15 per cent, and the effective share of a full quote is slightly lower.

What remains is the only calculation that means anything: your production, your self-consumption, your electricity price and your quote. Three of those four are known only to you. That is why this module has given you the bands and their sources rather than an answer — and why the measure calculator starts by asking about your house.

One more thing, for completeness: the inverter, the box that turns the panels' direct current into alternating current, is explained in the module on electricity. It is the one component in the system that will certainly need replacing before the panels are worn out.

Check yourself