Your lateral thinking and knowledge of lab kit could help us solve a tricky measuring problem in building-physics.

One of the problems we get in modelling the heat demands of buildings, is getting an accurate number for the building's thermal conductivity as a whole. Even when we know the materials it's built from, variations in material quality, thickness, and build-consistency, mean that the calculation of the composite thermal conductivity is frequently wrong. Now, yes, if we could get all of the very specific information on the exact materials, and the exact locations of every thermal bridge, we could calculate the building's thermal conductivity from first principles. But that's very rarely possible. (incidentally, we also need to know the thermal mass, but the calculation tends to be a bit more forgiving about measurement inaccuracies, and close enough may be good enough)

So, it strikes me that there is almost certainly some kit in use by physicists in labs or industry, that could be co-opted for this purpose. Something that could be taken to different dwellings, to measure the thermal conductivity of specific walls, floors, or whole buildings, and that would allow the measurement to be done within one day. [edit 1:] Note that in general, we will have access to the inside as well as the outside of a building.[end edit 1]

It seems that the use of thermal cameras hasn't been entirely successful, due to problems with calibration, and with variations in reflectivity and emissivity of different materials.

Buildings are almost never in thermal equilibrium, and often have high thermal mass, so simply releasing a known quantity of low-grade heat in the dwelling and recording time-series of temperature of inner air, wall, outer wall, as well as recording times series of external air temperatures, could be doable, but it would typically be difficult to establish what the initial and final conditions are (e.g. how much heat is stored in the building, at the start and at the end).

Testing must be non-destructive: that is, we can't submit surfaces to tests that would damage them. So no super-heating or deep freezing, if it would cause permanent damage.

[edit 2: supplementary information prompted by a question about using historic information about energy consumption to deduce conductivity:]

Historic energy use of dwellings does provide some information; however, it is often hard to obtain for a lot of dwellings (being commercially sensitive); and, as generally we don't know the long-term pattern of internal temperatures, we can't deduce the building's thermal conductivity. We may know how much energy was used, but we don't know how much warmer than external temperatures did it keep the building, nor for how long. It turns out that observed thermostat settings aren't useful for this, for several reasons.

[end edit 2]

[edit 3]We are interested heat losses from both conduction, convection and radiation. Convection is sometimes estimated using data from a pressurisation test, which isn't always easy to do. And we're interested in conduction and radiation losses as well as convection losses. Some heat flux sensors have been tried, but apparently without much success. If there have been recent advances in heat flux sensors, I'd love to hear about them.[end edit 3]

I'm not inviting speculation: I'm looking for a factual answer about specific tests to yield specific measurements about the physics of dwellings.

  • 1
    $\begingroup$ Wow, extremely well thought-out and interesting question. I have no idea how you would tackle the initial/final conditions problem in general. If the building is a black box for you, whatever you measure will be dependent on what's happening inside the box. $\endgroup$
    – Marek
    Commented Jul 29, 2011 at 8:56
  • $\begingroup$ Thank you! It's not entirely a black box, in that we can estimate the thermal mass, and we typically will have access to the inside - so we can place measurement devices either side of a wall, for example. $\endgroup$
    – 410 gone
    Commented Jul 29, 2011 at 9:08
  • $\begingroup$ Aha, there is a conspriracy not to give that secret "kit" to the green saviours of the world? There is a very simple tool for existing buildings: look at the energy (electricity, gas, hot water etc) needed to heat that building. $\endgroup$
    – Georg
    Commented Jul 29, 2011 at 10:15
  • $\begingroup$ @Georg - thanks for the prompt - I've updated the question with some information about the usefulness (or otherwise) of historic energy usage. $\endgroup$
    – 410 gone
    Commented Jul 29, 2011 at 10:35
  • 1
    $\begingroup$ I think you have an additional problem, that is knowing the history of the temperature imposed on the outside of the walls and roof. These are affected by longwave and shortwave thermal radiation. So dependent upon atmospheric conditions other than the air temperature, your actual boundary conditions may be seriously wrong if these aren't accounted for. $\endgroup$ Commented Jul 30, 2011 at 22:22

2 Answers 2


In one of our first physics lab we did an experiment to characterize the thermal properties of a simulated house (it actually was a box with 4 sides made of different materials).

One just periodically heated the inside to 50 degree celsius. The temperature time series on the outside wall will be damped and phase shifted relative to the inside.

I googled "praktikum thermohaus periodisch" and came up with this description: http://www.uni-jena.de/unijenamedia/Downloads/faculties/physik_astro/phys_gp/V_210_1.pdf

And conveniently there is also a students protocol with some data: http://www.personal.uni-jena.de/~p1erra/physik/protokolle/seite%20129-%20versuch%20210.pdf

The following image displays the dynamic temperature data. It indicates how the insulating properties of Styrofoam are superior to glass.

enter image description here

I don't see a reason why this shouldn't work with a real building.

  • $\begingroup$ The problem in a real building is that it is made of meany materials and the "heat likes" are much harder to detect. $\endgroup$
    – Fortunato
    Commented Jul 31, 2011 at 21:34
  • $\begingroup$ I guess you mean heat leaks. So you say: If you measure the temperature at one point of the outside wall energy is actually leaving the building somewhere else, you won't pick it up. I agree, this seems to be a major issue of this method. $\endgroup$
    – whoplisp
    Commented Aug 1, 2011 at 17:22
  • $\begingroup$ Furthermore, thermal inertia is so large that the time needed to reach thermal equilibrium is way larger than the natural day-night temperature cycle. $\endgroup$
    – Pere
    Commented Jun 2, 2020 at 12:40

You asked for an instrument you can take around the building to detect the heat leaks and I don’t think you can do this. There is a way to do it over several days. Setting up a 2 time-lapse Thermographic cameras taking pictures every 10 min over a period of 3 days. One camera is set perpendicular to building face ant the other parallel. Watching the 2 sequences side by side will show where the heat is leaking. Studying the sequences over the major temperature shifts such as morning and evening. When the outside of the building goes threw the day to night transition the areas of heat loss will tend to linger or change faster compared to the well insulated areas.

  • $\begingroup$ Thank you! Yes, detecting where the leaks are is relatively straightforward; and I'm looking for the step beyond that, to measure the building's (composite) thermal conductivity. Understanding where the thermal bridges and convective losses are, can be one way to build up the big picture from the individual elements. I do like the idea of time-lapse thermography: +1 for that. $\endgroup$
    – 410 gone
    Commented Aug 1, 2011 at 7:29
  • $\begingroup$ If I remember correct, thermographic cameras are not welcome answers to this question, unfortunately :=) $\endgroup$
    – Georg
    Commented Aug 4, 2011 at 10:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.