BRENNEN

Industries / Industrial equipment and services / Industrial services / Thermography

Buildings

Building thermography scans walls, ceilings and floors from inside and outside to find missing insulation, air leaks, thermal bridges (places where heat takes a short cut through the structure) and damp.

7

Milestones achieved, from 1972 to 2018.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1972

    A manual of heat pictures

    I. Paljak and B. Pettersson of Sweden's National Institute for Building Research in Stockholm published Thermography of Buildings (Termografering av byggnader), a manual on how to find insulation defects with an infrared camera: how the camera works, how to read thermograms (heat pictures), rules for field work and sources of error. It ends with a catalogue of more than 700 thermograms of four wall types, each built both to specification and with common building and insulation defects. Two of the institute's researchers, Sven-Ake Ljungberg and Rolf Norberg, wrote in 1980 that Paljak and Pettersson "presented the first operational method using thermography to indicate energy loss in buildings", and that Sweden later produced a handbook for training camera operators.

    aivc.orgisprs.org

    Credit

    I. Paljak and B. Pettersson, National Swedish Institute for Building Research

  2. 1978

    Heat slips around insulation

    From 1972 a team at Princeton University's Center for Environmental Studies measured energy use in occupied townhouses at Twin Rivers, New Jersey. Richard Grot, then at Princeton and later at the US National Bureau of Standards (NBS), scanned a bedroom ceiling with NBS's infrared camera and found a cold patch that turned out to be a missing batt of attic insulation; cold corner patterns, from outside air flowing behind interior walls next to the masonry firewalls, showed up in more than ten townhouses. The team concluded that heat loss through an average attic was about five times the predicted value and carried 35% of the energy leaving a house in winter, though the attic held about 9 cm (R-11) of fiberglass, because air flow and the party walls between houses carried heat around the insulation.

    LessonInsulation on the drawings is not insulation at work; a camera shows the paths heat takes around it.

    socolow.princeton.edu

    Credit

    the Twin Rivers team, in Saving Energy in the Home, edited by Robert H. Socolow (Ballinger, 1978)

  3. 1983

    Contractors found half the voids

    For the US Department of Energy, NBS sent infrared contractors into homes it had already scanned itself for a federal weatherization program, and compared what they found. In 1978–79, at 12 homes in three cities, contractors found between 45% and 86% of the defects in wall areas, and most did not cover the whole house. In 1980–81, at 20 homes in five cities, the share of defective wall area each of 18 contractors found ranged from 10% to 90%, with an average of 48.7%; only ten found more than half. Firms whose cameras resolved temperature differences finer than 0.5 °F (0.3 °C) generally did better, but the report blames incomplete scanning, poor records and misreading as much as the equipment.

    LessonThe camera is not the survey; coverage, records and interpretation decided how much was found.

    nvlpubs.nist.gov

    Credit

    Y. M. Chang and R. A. Grot, NBS, report NBSIR 83-2701 (May 1983)

  4. 1983

    One method for envelopes

    ISO 6781, published in December 1983, set an international qualitative method for finding thermal irregularities in building envelopes (the walls, roof and floor that separate inside from outside) by thermography, including poor airtightness; it states that it does not measure how well insulated or how airtight a structure is. ISO 6781-1, published in August 2023, replaced it. The new part covers heat, air and moisture irregularities and sets requirements for thermographic services: the extent of the work, the condition of the equipment, the qualifications of camera operators, image analysts and report writers, and how results are reported.

    iso.orgboutique.afnor.org

    Credit

    ISO, committee TC 163/SC 1 for the 1983 edition

  5. 2010

    Party walls get a number

    In co-heating tests (heating an empty house to a steady temperature and metering the energy used) on new terraced homes at Stamford Brook, England, from 2005 to 2007, Leeds Metropolitan University found that air moving up the cavity in masonry party walls, the shared walls between houses, carried away up to 30% of total heat loss. The tests put the bypass at an effective U-value (heat loss per square meter per degree of difference) of about 0.5 to 0.7 W/m²K, where energy models had assumed zero. In a 2009 follow-up on two terraced houses in Bradford, infrared images showed plumes of heat at the tops of the party walls, and the plumes were gone once the cavity was filled with mineral wool. SAP 2009, the UK government's method for energy-rating homes, used to check new homes against the building regulations, in its version dated March 2010, gives an unfilled, unsealed cavity party wall a U-value of 0.5 W/m²K and a filled and sealed one 0.0.

    LessonLowe and colleagues note that the same bypass was seen at Twin Rivers in the late 1970s and reported in the UK in the mid-1990s, but nobody confirmed it or changed the calculation rules until their measurements.

    leedsbeckett.ac.ukleedsbeckett.ac.ukfiles.bregroup.com

    Credit

    R. J. Lowe (University College London) with J. Wingfield, M. Bell and J. M. Bell (Leeds Metropolitan University); J. Wingfield, D. Miles-Shenton and M. Bell for the 2009 tests; BRE for the UK government

  6. 2015

    Pictures move householders

    Researchers at Plymouth University tested whether seeing heat escape changes what people do. In one study of 43 English households, those shown thermal images of their own homes used less energy a year later, while households given a carbon footprint audit, and a control group, showed no change. In a second study of 87 households, those shown images were nearly five times as likely to fit draught proofing, and the effect was strongest for problems visible in the images.

    pmc.ncbi.nlm.nih.gov

    Credit

    Julie Goodhew, Sabine Pahl, Tim Auburn and Steve Goodhew, in Environment and Behavior, December 2015

  7. 2018

    Rules limit what cameras prove

    In the US, a home energy rating under the RESNET standard grades how well insulation was installed, from Grade I, the best, to Grade III, and the grade goes into the energy model behind the rating. A 2010 draft RESNET infrared standard described grading by camera, with Grade II for 0.5% to 2% of the area missing insulation and Grade III for 2% to 5%. RESNET's rules of 2018 settled it otherwise: a thermographic inspection may show that a hidden wall is insulated, which counts as Grade III; it may support Grade II only if the inspector is an ASNT NDT Level III (the highest US nondestructive testing certification) or a licensed engineer, or works under one; and it may never be used to show Grade I.

    LessonThe rating system decided that a camera cannot prove the absence of small voids.

    resnet.usresnet.usefficiencyvermont.com

    Credit

    RESNET's standards committee SDC 300, interpretation 301-2014-15 (effective 29 July 2018) and Addendum F-2018 to ANSI/RESNET/ICC 301-2014

Watch for

  1. Heat maps on 3D scans

    Chao Yang, Haoyuan Zheng and Yue Ma describe Thermo-LIO, a system that fuses a thermal camera with lidar (a laser scanner that measures distance) and tracks its own movement, so that surface temperatures are laid onto a 3D model of a whole building or bridge across repeated inspections. In their own case studies on a bridge and a hall building, they report that it found thermal anomalies and structural defects more accurately than traditional methods; the abstract gives no figures.

    Achieved whenanother team scans a building whose defects are confirmed by opening or coring, and publishes how many defects the system and a handheld camera survey each found and how many they flagged wrongly.

    arxiv.org

    Claimed by

    Yang, Zheng and Ma, in an arXiv preprint, January 2026

Open

  1. 01

    Insulation grades, checked by camera

    A set of wood-frame test walls with insulation voids of measured size, from none to 5% of the wall area, each forming one face of an insulated box kept warm inside while the other side is cold. Certified thermographers, and any AI tool that claims to grade insulation, grade each wall blind against RESNET's three grades, and their calls are scored against the voids actually built in.

    The brief
    Credit · unclaimed
  2. 02

    Repeat the 1981 contractor test

    A blind comparison of building thermographers, and of AI report tools given the same images, on houses that are about to be stripped for deep renovation, so that every defect a surveyor reports, and every one they miss, can be checked when the plaster and cladding come off.

    The brief
    Credit · unclaimed

Open 01 · the brief

Insulation grades, checked by camera

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A set of wood-frame test walls with insulation voids of measured size, from none to 5% of the wall area, each forming one face of an insulated box kept warm inside while the other side is cold. Certified thermographers, and any AI tool that claims to grade insulation, grade each wall blind against RESNET's three grades, and their calls are scored against the voids actually built in.

Why it is still open

RESNET's 2018 rules let a camera support Grade II only in the hands of a Level III inspector or a licensed engineer, and never Grade I; its 2010 draft had tied grades to void shares of 0.5%, 2% and 5%. NBS's 1983 field test scored contractors against NBS's own scans, not against walls with known voids. Laboratory work has shown that infrared finds planted defects in insulated panels, for example Candoré, Bodnar and colleagues in 2008, who heated French plasterboard and polystyrene linings to show planted gaps, and BRANZ, New Zealand's building research body, which in 2010 imaged a test wall with gaps of 15 to 60 mm above the insulation and found the 15 mm gap only just visible, a surface difference of 0.3 K, with 5 K between the air on its two faces. Neither scored people grading walls, and searches in November 2026 (web, Crossref, GitHub, Hugging Face) found no test that scored camera grading against the RESNET void thresholds.

resnet.usefficiencyvermont.comnvlpubs.nist.govapi.openalex.orgbranz.co.nz

Built with what exists

Stud wall panels about 1.2 by 2.4 m (4 by 8 ft) lined with plasterboard, filled with fiberglass batts and blown cellulose, with voids cut to measured areas; an insulated box with a small electric heater and a thermostat; a cold side outdoors in winter or a walk-in cooler; a FLIR E-series camera and a phone-attached camera for comparison; thermocouples; Python with OpenCV to measure the marked areas.

  1. First step

    Build four panels with 0%, 0.5%, 2% and 5% voids and image them on one cold night with a 10 °C difference across the wall.

  2. Done looks like

    At least 30 panels imaged at temperature differences of 10, 15 and 20 °C and graded by at least eight certified thermographers; a published table of called grade against true grade, the smallest void found 80% of the time at each temperature difference, and every image released under an open license.

Open 02 · the brief

Repeat the 1981 contractor test

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A blind comparison of building thermographers, and of AI report tools given the same images, on houses that are about to be stripped for deep renovation, so that every defect a surveyor reports, and every one they miss, can be checked when the plaster and cladding come off.

Why it is still open

The last same-house comparison found is NBS's of 1980–81, in which 18 contractors found 10% to 90% of the defective wall area, 48.7% on average. Richard Grot also gave reasons for inconsistent results between inspectors of the same building in a 1983 SPIE paper. Cameras, the ISO 6781-1:2023 rules on operator qualifications and AI report tools have all arrived since: Maesin, a London firm funded by Innovate UK, has built AI that predicts condition ratings of building elements from external thermal surveys and had thermographers test it, but its project page reports no comparison against opened walls. A search in November 2026 found no later study that sent several surveyors to the same houses.

nvlpubs.nist.govapi.crossref.orgiuk-business-connect.org.uk

Built with what exists

Three to five houses due for deep retrofit, found through a housing association or a retrofit program; a survey brief written to ISO 6781-1; at least ten certified thermographers with their own cameras; a blower door (a fan in a doorway that lowers the indoor air pressure so leaks show) for the air leakage scans; a photographic log of each wall as it is opened; a scoring sheet that matches reported and found defects.

  1. First step

    Ask a housing association or a retrofit program for houses due to be stripped this winter, and draft the survey brief and the matching rules.

  2. Done looks like

    At least ten surveyors on at least three houses; for each surveyor, the share of the defects found on opening that they reported and their number of false calls, published with the opened-wall photographs; the same scores for at least one AI tool run on the surveyors' raw images.