BRENNEN

Industries / Mining and materials / Insulation manufacturing

Fire safety

Fire safety is testing and classing how insulation products burn, alone and built into walls, and the rules that changed after fires.

6

Milestones achieved, from 1974 to 2024.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1974

    Foam makers settle with FTC

    In 1974, 25 companies that made or sold cellular plastics (plastic foams) or their ingredients signed a consent order with the US Federal Trade Commission. It set up a trust of $5 million, spent over five years by a Products Research Committee of trustees from universities, government and industry, to fund research on how these foams behave in fire, in four areas: basic research, small-scale tests, large-scale tests and smoke toxicity. The committee's final report, in April 1980, urged standards bodies such as ASTM and NFPA to review the test methods it recommended. A trade article of November 1974 reported that the consent orders called for hazard alerts to product users.

    firedoc.nist.govfiredoc.nist.gov

    Credit

    US Federal Trade Commission; Products Research Committee

  2. 2000

    Europe's fire classes

    Commission Decision 2000/147/EC of 8 February 2000 set up one European system for classing how construction products react to fire, with classes from A1 to F. The classes carry extra grades for smoke (s1 to s3) and flaming droplets (d0 to d2), set by tests that include non-combustibility, heat of combustion, the single burning item test and ignitability, and pipe insulation has its own classes, A1L to FL. Delegated Regulation (EU) 2026/331 kept the same classes under the new Construction Products Regulation, in force from 11 May 2026.

    eur-lex.europa.eueur-lex.europa.eu

    Credit

    European Commission

  3. 2010

    A retrofit feeds a fire

    On 15 November 2010 fire swept a 28-story apartment block in Shanghai that was being retrofitted to save energy, killing 58 people and injuring more than 70. Sparks from welding are believed to have started it, and it spread to bamboo scaffolding and nylon netting; Luo Lin, head of China's State Administration of Work Safety, blamed the disaster partly on flammable polyurethane insulation used in the retrofit. A preliminary State Council panel pointed to illegal layers of subcontracting, poor contractor management and weak safety supervision.

    LessonWork meant to save energy put fuel on the outside of an occupied tower.

    bjreview.com

    Credit

    China's State Administration of Work Safety and a State Council investigation panel

  4. 2016

    Germany adds fire barriers

    Germany's conference of state building ministers judged that properly installed external thermal insulation composite systems (ETICS, insulation boards fixed to an outside wall and covered with render) were safe enough against fires inside rooms, and new rules aimed at fires directly in front of a facade. From 1 January 2016 the approvals that the German Institute for Construction Technology (DIBt) issues for rendered systems insulated with expanded polystyrene (EPS) required bands of non-combustible mineral wool at least 200 mm high: at the base of the system, at the first-floor and third-floor ceiling levels, and at most 1 m below combustible parts such as a roof.

    LessonA wall judged safe against a fire from a window also had to be protected against a fire at its foot.

    baulinks.de

    Credit

    Bauministerkonferenz (conference of German state building ministers); DIBt

  5. 2024

    Inquiry exposes rigged tests

    The Grenfell Tower Inquiry's final report, published on 4 September 2024, set out how two insulation makers won a market for combustible foam boards on tall buildings. Celotex's first BS 8414 test (a full-scale fire test of a whole wall system) of its polyisocyanurate board was stopped after 26 minutes in February 2014 when flames overtopped the rig; the test it passed in May 2014 had thicker cladding panels and fire-resistant magnesium oxide boards at critical points, which it hid when it marketed RS5000 as "the first PIR board to successfully test to BS 8414". Kingspan tested a system with an earlier formulation of its K15 phenolic board in May 2005, changed the product's chemistry and facers by 2006, and kept making the same fire claims although no BS 8414 test was run on the new version until 23 March 2015.

    LessonA pass for one built test wall was sold as a rating for a board, and the board sold was not always the board tested.

    assets.publishing.service.gov.uk

    Credit

    Grenfell Tower Inquiry, chaired by Sir Martin Moore-Bick

  6. 2024

    Europe agrees one facade test

    The European Commission approved the final report on a single European method for testing how fire spreads up a facade and what falls from it. Input came from 29 laboratories in 20 countries, and a closing round robin tested four facade systems at three laboratories in 24 tests, at large and medium fire exposures, trying propane burners in place of wood cribs. The Commission plans to ask the European standards body CEN (working group CEN/TC 127/WG 6) to turn it into formal standards.

    pinfa.euinternationalfireandsafetyjournal.com

    Credit

    RISE, BAM and Efectis, for the European Commission

Watch for

  1. Facade tests rank like chance

    In 25 experiments, five facades (two ETICS and three rainscreen systems) were each tested under five large-scale standards: PN-B-02867, ISO 13785-1, DIN 4102-20, NFPA 285 and BS 8414. The authors report that the standards "could not rank the flammability of the facades more consistently than random chance" by flame spread, while the average heat release rate they reconstructed did give a consistent ranking.

    Achieved whenanother group repeats the comparison on different facades and gets the same result, or a standards body changes its pass criteria in response.

    citedrive.com

    Claimed by

    Matthew Bonner and Guillermo Rein (Imperial College London) with Wojciech Węgrzyński and Jakub Bielawski (Instytut Techniki Budowlanej, Warsaw), in Fire and Materials, June 2026

Open

  1. 01

    Fire claims checked by machine

    A tool that reads insulation makers' datasheets, brochures and product pages, lists every fire claim (a Euroclass, "suitable above 18 m", a BS 8414 pass) and checks each against the report it cites: the same product, the same wall build-up, the same standard, and a report still in date.

    The brief
    Credit · unclaimed
  2. 02

    A fingerprint for every board

    A public library of infrared spectra (FTIR, which identifies a material from the wavelengths of infrared light it absorbs) and densities for insulation boards on sale, with each product bought again every six months, so that a change in formulation shows up as a change in fingerprint.

    The brief
    Credit · unclaimed

Open 01 · the brief

Fire claims checked by machine

Credit · unclaimed

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

What to build

A tool that reads insulation makers' datasheets, brochures and product pages, lists every fire claim (a Euroclass, "suitable above 18 m", a BS 8414 pass) and checks each against the report it cites: the same product, the same wall build-up, the same standard, and a report still in date.

Why it is still open

The Grenfell inquiry found that two makers sold foam boards for tall buildings on tests of a different system build-up or an earlier formulation. The UK's voluntary Code for Construction Product Information (CCPI), started by the Construction Products Association after Grenfell, verifies a firm's product information by independent third-party assessment, firm by firm; searches for this brief found no automated checker.

assets.publishing.service.gov.ukdesigningbuildings.co.uk

Built with what exists

A large language model that reads long PDFs, such as Claude (the Note "Seven more flagged" describes a Claude-based reviewer of drug-making records); PDF text extraction; makers' public datasheets and classification reports; the class rules in Delegated Regulation (EU) 2026/331 and the UK's BR 135 criteria.

  1. First step

    Collect 20 datasheets for PIR, phenolic, EPS and mineral wool boards, mark every fire claim in them by hand, and use that as the first test set.

  2. Done looks like

    200 datasheets from at least 10 makers in the UK and EU read; every fire claim tagged supported, unsupported or unclear; a chartered fire engineer checks a random 50 claims, and the tool's agreement with the engineer is published with the list of claims it could not match to any report.

Open 02 · the brief

A fingerprint for every board

Credit · unclaimed

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

What to build

A public library of infrared spectra (FTIR, which identifies a material from the wavelengths of infrared light it absorbs) and densities for insulation boards on sale, with each product bought again every six months, so that a change in formulation shows up as a change in fingerprint.

Why it is still open

Kingspan changed K15's chemistry and facers in 2005 and 2006 and kept selling it on a test of the earlier version, which the market did not know. A 2024 thesis by George Edwardes of the Fire Protection Association, at the University of Central Lancashire, developed a "chemical fingerprint" for cladding products from microscale combustion calorimetry and infrared spectroscopy, as a method; no public, repeated library of insulation board fingerprints was found.

assets.publishing.service.gov.ukclok.uclan.ac.uk

Built with what exists

A bench ATR-FTIR spectrometer of the kind found in university chemistry teaching labs; a precision balance and calipers for density; open spectral-matching code in Python; boards bought from builders' merchants.

  1. First step

    Buy samples of five PIR and five phenolic boards, take three spectra from each, and check whether the spectra tell the ten products apart.

  2. Done looks like

    40 products from at least eight makers, each bought twice six months apart; in a blind test where someone relabels 20 pairs of samples, the library matches at least 18 to the right product; any product whose fingerprint changed is listed publicly with the maker's reply.