Book a demo
  1. Home
  2. The Home Energy School
  3. Modul 06

Part 2 · Your house · Moderate

The house that leaks

We have called the house a bucket with holes in it. Now we look at the bucket itself. How fast does it leak, what sets the speed, and what really happens when you add another twenty centimetres of insulation? The answer to that last one is less than you would think, and it is the most useful line on this page.

In brief

  • Heat leaves a house two ways: through walls and windows, and riding on the air.
  • The U-value says how fast a wall leaks. Low is good.
  • The first twenty centimetres of insulation do the most work. A second layer gives far less.
  • Ventilation may not be blocked. Damp, stale air and radon have to leave.
  • With a heat pump a saved kilowatt-hour is worth less. The pump makes heat cheap anyway.
On this page
  1. Two ways out
  2. The U-value, from the bottom up
  3. The second layer does much less
  4. What Swedish houses actually have
  5. The hole you are not allowed to plug
  6. From a leak to a yearly bill
  7. The same kilowatt-hours, different money
  8. Check yourself

Two ways out

The heat leaving your house takes two entirely different routes. They look identical on the bill, but they are fixed in entirely different ways, which is why they have to be kept apart.

Through the material. Heat walks straight through a wall, a roof, a floor, a pane of glass. No air moves. It happens silently and without pause, day and night, all winter.

Riding on the air. Warm indoor air leaves and cold outdoor air takes its place — through the ventilation, and through every crack. Here the air itself carries the heat out.

Let us put numbers on it with a worked house: 150 m² on one floor, built around 1970. That is the most common house type in Sweden. It measures 15 × 10 metres, the walls are 2.5 metres tall, with 20 m² of window and 4 m² of door. The dimensions are chosen so the sums can be followed, not taken from a register.

We need a measure of leak rate, and it is watts per degree, W/K: how many watts run out for every degree of difference between inside and outside. A 100 m² wall with a U-value of 0.40 leaks 40 watts per degree. That is the whole operation: area times U-value.

Part of the house Area U Leak
Walls101 m²0.4040.4 W/K
Roof150 m²0.3552.5 W/K
Floor on ground150 m²0.5075.0 W/K
Windows20 m²2.448.0 W/K
Doors4 m²2.08.0 W/K
Thermal bridges, +20 %——44.8 W/K
The air being replaced189 m³/h—62.4 W/K
Total 331 W/K. The material accounts for 269 of them, the air for 62 — roughly four fifths against one fifth. The U-values are typical for 1960–1975 from our own guide; thermal bridges are an uplift of 15–30 % the standard method allows for, and we chose 20 %.

Two rows deserve a comment. The windows are 20 m² and leak nearly as much as 101 m² of wall — the most striking line in the table, and the reason old windows get so much attention. The floor is the shakiest row. The ground under a slab is not as cold as January air, so in reality a floor leaks less than this multiplication says. A careful calculation gives the ground its own, milder temperature.

Where the 62.4 W/K for the air comes from

The rules require at least 0.35 litres of outdoor air per second per square metre of floor. For 150 m² that is 52.5 litres a second, which is 189 cubic metres an hour.

0,35 × 150 = 52,5 l/s → 189 m³/h

Warming a cubic metre of air by one degree costs about 0.33 watt-hours. The standard method writes the leak as 0.33 times the flow, times whatever the recovery does not take back.

0,33 × 189 × (1 − 0) = 62,4 W/K

The U-value, from the bottom up

The U-value appeared in that table without explanation. Time to build it up from the bottom, because then the rest of the module becomes obvious. There are three numbers, and they are one fact seen from three sides.

λThe material. Lambda says how well a substance conducts heat, whatever its thickness. Mineral wool sits around 0.04 W/mK. Wood conducts worse, metal far better — which makes metal far worse to build with.
RThe layer. Divide the thickness in metres by lambda and you get the resistance. It is the material plus how much of it you used. High R is good.
UThe whole wall. Add up every layer's R, and turn the number upside down. Low U is good.

R = thickness / λ    U = 1 / Rtotal

High R and low U are the same claim, said twice. U is R upside down.

An example with numbers. Twenty centimetres of mineral wool is 0.20 metres divided by 0.04, so a resistance of 5.0. A wall made of nothing but that wool would have a U-value of 1 divided by 5.0 — that is 0.20.

0.20 / 0.04 = 5.0 m²K/W  →  1 / 5.0 = 0.20 W/m²K

Now the anchor, which makes the numbers physical. An old single-glazed window has a U-value around 4.5. A modern high-performance triple sits around 0.8. Turn both upside down and compare the resistances:

Window U R = 1/U As much resistance as At a 20-degree difference
Single glazing4.50.229 mm of mineral wool90 W per m²
High-performance triple0.81.255 cm of mineral wool16 W per m²
The resistances are 1 divided by U, and the thicknesses are R times 0.04. The watts are U times 20 degrees. The difference is a factor of 5.6 — which is why an old window feels cold to sit beside.
Why no U-value can rise without limit

The air closest to a surface stands almost still and insulates a little by itself. The standard method counts 0.13 m²K/W on the inside of a wall, 0.10 on the inside of a roof, and 0.04 outside.

Those two films are always present. Even an infinitely thin sheet therefore has at least 0.17 of resistance, and cannot reach a U-value above roughly 5.9.

1 / (0.13 + 0.04) = 5.9 W/m²K

That explains why single glazing's 4.5 sits close to the ceiling on how bad a thing can be. Much worse than that is barely possible to build.

The second layer does much less

Here comes the most important line in the module, and the reason many people over-buy insulation. Resistances add up in a straight line, but resistance is not what you pay for — the U-value is. And U is one divided by the total.

Take the worked house's wall. Its U-value is 0.40, so its resistance is 1 divided by 0.40, which is 2.5. We add twenty centimetres at a time, and each helping is worth 5.0 of resistance.

Insulation R U This step gave
The wall as it stands2.50.400—
+ 20 cm7.50.133−0.267
+ 40 cm12.50.080−0.053
Same thickness, same price, same tradesman — and the second helping delivers 0.053 against the first one's 0.267. That is one fifth as much.

0.053 / 0.267 = 0.20 → one fifth of the benefit, for the same money

Resistance grows in a straight line. The U-value does not.

U 0.40 U 0.20 U 0.133 U 0.10 U 0.080 now +10 cm +20 cm +30 cm +40 cm −0.200 −0.067 −0.033 −0.020

The lower box is the wall you already have, the upper one is the wool you buy. Every helping bought is as tall as the last — but the bottom row shows what each one took off the U-value, and that figure nearly collapses.

What Swedish houses actually have

The construction year says more about the envelope than anything else you can know without opening a wall. The watershed is Svensk Byggnorm 1975, which followed the oil crisis and raised insulation requirements sharply. A house from 1974 and one from 1977 differ more than 1977 and 1990 do.

Element Pre-1960 1960–1975 Post-2010 BBR 29's ceiling
Walls0.50–0.700.35–0.500.12–0.180.18
Roof0.50–0.900.30–0.400.08–0.130.13
Floor on ground0.50–1.000.40–0.600.12–0.150.15
Windows3.50–4.502.00–2.800.70–1.201.2
A summary. The full table, with 1976–1990 and 1991–2010, what sits inside the wall in each period and what added insulation realistically delivers, is in the guide. The figures are typical per period, not measurements of one house.

The last column is what BBR 29 demanded of a new build, plus a combined requirement of 0.30 for the house as a whole. Note that a house built after 2010 already sits at or below the ceiling — the requirements followed practice rather than the other way round. Note too that Boverket's rulebook is being replaced, so a date always belongs with a requirement figure.

Open the guide: U-values by construction period

The hole you are not allowed to plug

A fifth of the worked house's leak went out with the air. Anyone who has just understood this chapter gets a very natural idea: block it up. That idea is the most dangerous one in the whole course, and it has made more houses sick than it has saved money.

Damp comes first. A household pours water into its own air all day long — cooking, showering, laundry, wet outdoor clothes, houseplants, and the people themselves breathing. It is litres a day, not decilitres. We have found no Swedish source that puts a figure on it, so treat that as an order of magnitude and nothing more.

That air has to leave, or the water condenses wherever the surface is coldest: in a window reveal, in an outside corner, behind a wardrobe against an outer wall. That is where mould starts. Mist on the inside of the pane on a cold morning is exactly that water becoming visible.

Then carbon dioxide. A bedroom with the door shut and a taped-over vent turns stale before morning, and the air in it shapes how you sleep. Radon last. Radon rises out of the ground and out of blue concrete, and the only everyday defence is air being replaced. A house sealed up without its ventilation being reviewed concentrates whatever was already there.

So the rules set a floor, and it is not negotiable: at least 0.35 litres of outdoor air per second per square metre of floor area, and at least 4.0 litres a second per person in a room. That is where the worked house's 189 cubic metres an hour came from.

The Swedish system types

Type What it does What happens to the heat
Natural draught No fan at all. Driven by temperature difference and wind. All the heat goes with it. The draught is weakest in summer, when it is needed most.
Mechanical exhaust A fan pulls air out. Supply air arrives through vents and gaps. All the heat goes with it. Cold outdoor air is drawn straight in.
FTX Fans both ways, with a heat exchanger between them. The warm exhaust pre-warms the cold supply. The two streams never mix.
Exhaust-air heat pump A heat pump sitting on the exhaust flow. The heat is lifted out of the exhaust and sent to radiators and hot water. It is a heat pump, not a second air path.
The definitions are Boverket's, from its regulation on ventilation function checks. FT without an exchanger also exists: controlled flow, no recovery.

The exchanger does one thing only, and it is easy to picture. Exhaust air at 21 degrees passes close by supply air at −1 degree, and hands over most of its heat without the streams mixing. If the exchanger takes 80 per cent of the temperature difference, the air enters the room at 16.6 degrees instead of −1.

−1 + 0.80 × (21 − (−1)) = 16.6 °C

The same airflow, with and without an exchanger

Exhaust only: no exchanger FTX: heat exchanger 80 % −1 °C −1 °C 21 °C 21 °C arrives at −1 °C arrives at 16.6 °C 189 m³/h 189 m³/h 62.4 W/K 12.5 W/K the radiator covers 22 degrees the radiator covers 4.4 degrees Difference: 49.9 W/K, which in Stockholm is 4 311 kWh a year

Computed as 0.33 × 189 × (1 − 0.80) = 12.5 W/K against 62.4 without an exchanger. The annual figure is the difference times 3 600 degree days times 24 hours.

Does a house need a ventilation inspection?

A building's owner must have the ventilation checked before first use and regularly after that. But one- and two-dwelling houses are exempt from the recurring check.

On top of that the requirement does not apply at all to a house with natural draught or plain exhaust without recovery. The net: an ordinary house with natural draught or exhaust needs no check whatsoever. With a recovery unit it needs a first inspection, but no recurring one.

If the owner suspects a fault it must be put right as quickly as possible. Supervision sits with the municipal building committee.

From a leak to a yearly bill

We have a leak rate: 331 watts per degree. It says nothing about money until we know how many degrees, and for how long. That is exactly what degree days are for.

The idea is simple. Take each day of the year. If the day's mean temperature sat below 17 degrees, count how many degrees below — at 5 degrees that day scores 12 degree days. If it sat above 17 the day scores zero. Add up the whole year, and you have one number describing the winter of a place.

Why 17 rather than the 21 degrees you actually keep indoors? Because people, lamps, the fridge, the cooker and the sun heat the house for free. The last few degrees up to 21 come as a bonus. Those four degrees are the method's crude way of subtracting the free heat, and crude is the word — anyone wanting precision has to model solar gain and how many people live there.

kWh = W/K × degree days × 24 / 1000

Locality Degree days, 17 °C base We compute with The worked house, kWh/yr
Malmö2 800–3 2003 000~23 800
Göteborg3 200–3 5003 350~26 600
Stockholm3 500–4 0003 600~28 600
Sundsvall4 500–5 0004 750~37 700
Östersund5 000–5 5005 250~41 700
Luleå5 500–6 0005 750~45 700
Kiruna6 000–6 6006 300~50 000
The right-hand column is 331 W/K times the chosen figure, times 24, divided by 1 000. We take the middle of each band except for Stockholm, where we use 3 600 — what our own guide computes from the typical-year files, and low within the band. The same house costs a little over twice as much to heat in Kiruna as in Malmö — without a single wall being different.

Three caveats belong with those figures, and they belong in the same sentence as the figures. The bands are wide because a degree-day number depends on which normal period and which base temperature were used — our own guide computes a table per locality from typical-year files, and it agrees with station data within 4.3 per cent. Hot water sits outside: degree days are only about warming rooms. And the worked house is untouched since 1970: 28 600 kWh over 150 m² is 191 kWh per square metre, which is deep in the worst energy class. A house with a new attic and new windows lands far lower.

Open the guide: degree days and normal-year correction

The same kilowatt-hours, different money

Now comes the thing that makes two identical houses get two different answers to the same quote. It is the single most common slip in a measure calculation, and it is easy to correct once you see it.

There is a consolation in that. A house with a poor heat source has the most to gain from its envelope, and a house with a heat pump has already banked much of the prize. So the order matters: work out the envelope before the pump is bought, because a tighter house needs a smaller pump — and a smaller pump is cheaper.

Check yourself

Six questions. Getting one wrong is the useful part — the explanation is written for exactly that answer. Nothing is timed, nothing is saved, and nobody sees what you answer.

  1. Question 1One wall has a U-value of 0.40 and another 0.15. Which is better?
    • Tempting, because higher usually means better. But U measures the leak, not the protection: 0.40 watts per square metre per degree run out, against 0.15. It is the resistance R that rises when you insulate, and U is 1 divided by R.

    • Yes. The 0.15 wall leaks barely 40 per cent of what the other one does. If you want a figure where higher is better, use the resistance: 1 divided by 0.40 is 2.5, and 1 divided by 0.15 is 6.7.

    • A shrewd objection, but it applies to lambda, not U. Lambda describes the material with no thickness attached. U describes the whole wall, thicknesses included — which is precisely why U is the number you compare with.

  2. Question 2A wall has a resistance of 2.5. You add 20 cm of insulation, so 5.0 more. What is the U-value?
    • That would hold if you had added as much resistance as the wall already had, namely 2.5. But you added 5.0, which is double. The total is 7.5 and U is 1 / 7.5 = 0.133.

    • Exactly. 2.5 + 5.0 = 7.5, and 1 / 7.5 = 0.133. Notice how you did it: the resistances were added straight up, and only at the end was the number turned upside down.

    • Here the U-values have been added instead of the resistances, and that never works. Only R may be summed. The right route is 2.5 + 5.0 = 7.5, then 1 / 7.5 = 0.133. The figure 0.080 is only reached at 40 cm of insulation.

  3. Question 3You have put 20 cm in the attic. The contractor suggests another 20 cm, at the same price. What do you get?
    • That is the intuitive expectation, and it is why people over-buy. The resistance does grow by the same amount — but U is 1 divided by the total. The first layer gave −0.267, the second gives −0.053.

    • Yes, and it is the module's most important figure. 0.053 / 0.267 = 0.20. The money is the same, the benefit is one fifth. The question to ask is whether some other part of the house is still on its first layer.

    • Not that bad. There is no threshold where the wool stops working, and U keeps falling: 0.133 becomes 0.080. The benefit merely shrinks per krona spent, approaching zero without ever reaching it.

  4. Question 4A neighbour has taped over the bedroom vents to save heat. What is the main problem?
    • The arithmetic holds, and that is what makes it dangerous. The air carries out the damp a household produces every day, and if it stays in it condenses wherever the surface is coldest. Mould costs far more than the kilowatt-hours saved.

    • Yes, and in that order. Damp first, because it damages the building itself. So the rules set a floor of 0.35 litres a second per square metre. If you want that heat back, recovery is the route, never throttling.

    • A reasonable worry, but the wrong floor of the problem. The fan usually survives — it is the house and the people in it that suffer. And in a natural-draught house there is no fan to wear out in the first place.

  5. Question 5The same house is moved on paper from Malmö to Kiruna. What happens to its heat demand?
    • The house is identical, but demand is the leak rate times the weather. Kiruna has roughly 6 300 degree days against Malmö's 3 000, so twice as much to replace: about 50 000 kWh against 23 800.

    • Yes. 331 W/K times 6 300 degree days times 24, divided by 1 000, is about 50 000 kWh. In Malmö the same sum gives 23 800. Same walls, twice the bill — which is why a measure can pay in the north and fail in the south.

    • The cold feels worse than it counts. Degree days measure the whole year, and while Kiruna's winter is cold, Malmö's is also long. The figures are 6 300 against 3 000 — a factor of 2.1, not 4.

  6. Question 6A measure saves 4 300 kWh of heat a year. The house has a ground-source pump at SCOP 3.0. What does the owner save in money, at 2.00 kr/kWh?
    • That is the answer for a house with direct electric heating, and it is the commonest slip in a calculation. A pump at SCOP 3.0 buys only a third of the heat as electricity: 4 300 / 3.0 = 1 433 kWh, so 2 900 kr.

    • Yes. The heat saved is the same, but the pump was already making it cheaply: 4 300 / 3.0 × 2.00 = 2 900 kr. So the same quote can pay in the neighbour's electric house and not in yours. SCOP 3.0 is an assumption — change it and the answer moves.

    • The pump makes heat cheaper, but it does not make it free. Every kilowatt-hour of heat saved is still a third of a kilowatt-hour of electricity that need not be bought. 2 900 kronor a year is real money — it merely carries a large investment less well.

Your answers live only in your browser, and vanish when you leave the page.

Sources

  1. U-values by construction period and BBR 29's requirements for new builds (wall 0.18 · roof 0.13 · floor on ground 0.15 · window and door 1.2 · 0.30 combined for houses above 50 m²): Kronwatt's own research note on insulation, summarised in the U-value guide. Typical per period, not measurements. Boverket's rulebook is being rewritten, so a requirement figure always carries its date.
  2. The calculation method is ISO 13790's monthly quasi-steady-state one: ventilation loss as 0.33 times the airflow times (1 − efficiency), thermal bridges as an uplift of 15–30 % (we chose 20 %), and surface resistances of 0.13 m²K/W inside a wall, 0.10 inside a roof, 0.04 outside.
  3. Minimum outdoor air flow of 0.35 l/s per m² of floor area and 4.0 l/s per person: Boverket's regulation BFS 2024:8, chapter 3, section 5, in force with the new building rules from 1 July 2025. The well-known figure survived the rewrite, but the legal reference is new.
  4. The system types S, F, FT and FX/FTX, and the inspection rules: Boverket's regulation BFS 2011:16, sections 2 and 3, and the Planning and Building Ordinance chapter 5, section 1, which exempts one- and two-dwelling houses from recurring inspection.
  5. No Swedish rule requires a minimum heat-recovery efficiency in dwellings. We checked Boverket's current and former building rules. The 75 % and 85 % figures are Upphandlingsmyndigheten's procurement criteria for apartment blocks, reproduced in Å. Wahlström, Guide FTX (revised 2024-05-27, published in RISE's DiVA archive), which also records earlier recommendations of 70–80 % for plate exchangers, above 80 % for rotary ones and around 60 % for run-around coils. EU regulation 1253/2014 sets 73 % and 68 % from 2018, but only for ventilation units in non-residential buildings.
  6. Energimyndigheten (Husguiden, ventilation page, updated 8 July 2022) writes that recovering heat from exhaust air "halves" the energy for heat and hot water in energy-efficient houses, depending on the house and the place. That is the agency's wording, not a measured constant, and we have used it in no calculation on this page.
  7. Degree days: base temperature 17 °C, without seasonal thresholds since 2015, normal period 1991–2020. The bands per locality come from Kronwatt's own research note; a computed table per locality from SMHI's and Boverket's typical-year files sits in the degree-day guide and agrees with station data within 4.3 %. A degree-day figure without its base and its normal period says nothing.
  8. Sveby's standard values are 21 °C indoors and internal gains of 4–5 W/m², plus hot water at 20 kWh/m² per year or 781 kWh per person per year. Hot water is in none of the degree-day calculations on this page.
  9. The electricity price of 2.00 kr/kWh is our own calculation: SCB's measured total price for the second half of 2025 was 209.21 öre/kWh, and the electricity tax was cut in 2026 from 43.9 to 36.0 öre, which with VAT is 9.9 öre. A worked example to swap for your own invoice, never a forecast.
  10. SCOP 3.0 is an assumption, not a measurement. No official Swedish figure exists for a typical seasonal performance factor: Energimyndigheten's field measurement of ground-source pumps in 2012–2014 gave 2.7–2.9, lab tests of newer pumps 4–5, and the EU label's class boundaries describe a third thing. We use 3.0 and say so.
  11. Lambda of 0.04 W/mK for mineral wool is a rounded working value we declare here on the page, not a cited constant. It is printed on the packaging and differs between products.
  12. How much water a household adds to its own indoor air: we found no Swedish source that puts a figure on it. "Litres a day" is an order of magnitude, and is used in no calculation here.
  13. The worked house's dimensions, areas, watts per degree, kilowatt-hours and kronor are multiplied out here on the page from the figures above. None of them is taken from a register.