Reaction to Fire versus Fire Resistance

A reaction-to-fire class and a fire resistance classification come from different machines and answer different questions. Written into one clause, they let a supplier evidence half a facade and call it compliant.

The cavity barrier line in the facade specification reads: mineral wool, A1 to EN 13501-1, minimum EI 120, non-combustible throughout. The tenderer sends back a reaction-to-fire certificate showing A1 and a technical data sheet showing EI 120, and the submittal is approved.

Those two documents came from different tests and answer different questions. The A1 certificate says nothing about holding anything for 120 minutes, and the EI 120 report says nothing about combustibility. Neither says the barrier is the right type for the cavity it is going into. If the clause never said which component each requirement lands on, you find out on site which one the contractor priced.

Fig 1. A non-combustibility furnace beside a fire resistance furnace, with the result each one produces. The class and the classification come from different exposures, so one certificate cannot evidence the other.

Two questions, two machines

Reaction to fire asks how much the product itself contributes to a fire. EN 13501-1 answers with a class letter from A1 to F, plus sub-classes for smoke (s1 to s3) and flaming droplets (d0 to d2). The class is assembled from up to four bench tests, depending on the class being claimed: EN ISO 1182 for non-combustibility, EN ISO 1716 for gross heat of combustion, EN 13823 for the single burning item, and EN ISO 11925-2 for ignitability. None of them measures separation between two spaces. There is no time in an A1 classification and no assembly.

Fire resistance asks how long an element keeps fire on one side of it. EN 13501-2 answers in minutes against R for loadbearing capacity, E for integrity and I for insulation, using the EN 1363 heating conditions and the relevant part of EN 1364, EN 1365 or EN 1366. The I criterion is a defined number: a mean temperature rise of 140 °C on the unexposed face, and a rise of 180 °C at any single point. That result belongs to one construction, at one dimension, with one set of fixings.

A1 is a property of the material. EI 120 is a property of the system the material was tested inside. Written into the same clause, they describe two things that no single certificate can evidence, and a supplier who provides one of the two has technically complied with something.

British practice has the same trap in older wording. Class 0 is a building regulation classification, not a test outcome, and no laboratory issues a Class 0 report. It is assembled from a Class 1 surface spread of flame result to BS 476-7 together with fire propagation indices to BS 476-6, index i not more than 12 and sub-index i1 not more than 6. A submittal quoting Class 0 with only the BS 476-7 report attached is one test short, which is easy to miss when the covering letter is confident.

Where each requirement lands in a ventilated facade

Cut a section at a floor line through a drained and back-ventilated rainscreen and at least five things carry a fire requirement, each governed by a different document:

the insulation and the cladding panel carry a reaction-to-fire class
the cavity barrier carries a fire resistance classification
the joint between the slab edge and the backing wall or curtain wall carries a separate one
the support brackets and rails carry a combustibility requirement but no rating of their own
the wall as a whole, if anyone asks for it, carries a system test result that none of the component certificates can produce

A specification that lists A1 and EI 120 without saying which of those five it means has not specified the facade. It has specified a mood.

Fig 2. Section through a ventilated rainscreen at a floor line, with the standard that governs each component. Read it as five separate pieces of evidence rather than one rating for the wall.

`EI 120` for a cavity barrier is at least three different tests

EN 1366-4 covers linear joint seals, and it is where closed cavity barriers are tested. It assumes the barrier fills the space and stays filling it.

An open-state cavity barrier does not do that. In service it has to leave the cavity clear so the facade drains and ventilates, and it closes only when heated. Until recently there was no European test standard for it and the market ran on ASFP Technical Guidance Document 19. BS EN 1364-6 was published in May 2025 to cover open-state barriers, with an optional flame criterion for the facade case where a window breaks and a developed fire reaches the outside face. Its scope is explicit about what it excludes: closed cavity barriers stay with EN 1366-4, and barriers containing penetration seals go to EN 1366-3.

Worth knowing before anyone waves the new standard at you. The national foreword records that the UK committee voted against approving it, and one of the stated reasons sits in the introduction: no European reference scenario has been established for open-state barriers in facades.

The commercial consequence is immediate. Write EI 120 on a rainscreen cavity barrier and say nothing about open state, and a compliant offer is a closed EN 1366-4 product. It is cheaper, it is easier to install, and it is evidenced against your clause. It also fills the drainage path the facade engineer designed. That argument is unwinnable at submittal stage because the specification did not ask for the right thing, and the variation to correct it lands after the brackets are set.

The slab edge joint is tested on a different rig

The joint between the floor slab and a curtain wall is not a linear joint in a wall. Fire reaches it from the compartment behind and from a plume running up the outside face once the glass fails, and the joint has to survive both while the curtain wall deflects.

Perimeter fire barriers therefore have their own test. ASTM E2307 uses an intermediate-scale, multi-storey apparatus with a room burner and a window burner, so the specimen sees both exposures, and it reports F and T ratings, integrity and temperature, rather than E and I. UL 2079 adds movement cycling and air leakage, reported as an L rating. The exposure, the failure criteria and the reported letters are all different from a furnace test.

So an EN 1366-4 report is not evidence for a perimeter fire barrier, and an ASTM E2307 report is not evidence of an EI classification. Neither is stricter than the other. They are answers to different questions, and a consultant who accepts one against a clause asking for the other has approved something untested for its actual condition.

Fig 3. Linear joint seal exposure beside the perimeter fire barrier rig. The perimeter condition adds an external plume and a deflecting curtain wall, which the furnace test never applies.

No stack of component certificates makes a wall test

Facade fire behaviour is a system property. ISO 13785-2 tests it at large scale, with a simulated compartment fire venting through a window opening onto a specimen built with a 90 degree re-entrant corner, and ISO 13785-1 is the intermediate-scale screening test that filters candidates before that. BS 8414 and NFPA 285 do the same job in their own regulatory contexts. All of them burn a built-up wall rather than a material.

The 2025 EN for open-state cavity barriers says as much in its own introduction: work on a full-scale facade test is ongoing, and results to the document cannot for the time being be used to imply the performance of the external facade system the barrier sits in.

Read as a specifier, the conclusion is uncomfortable but useful. A correctly classified barrier, correctly classified insulation and correctly classified cladding do not add up to a tested wall. If the project needs system-level evidence, ask for it as a separate line, with the substrate, the bracket type and the cavity depth stated, because those are what the test was built around.

Fig 4. Component classification plotted against system-level test evidence, with the compliant region shaded. Good classes on every component still sit outside the region if the wall itself was never tested.

Before you approve a facade fire submittal

Which of the five components in the section does this certificate actually cover?
Is the cavity barrier evidence open state or closed state, and does that match the cavity the facade needs?
Does the tested joint width and movement range bracket the surveyed slab edge, or only the nominal one on the drawing?
If the clause says Class 0, has the BS 476-6 fire propagation result been supplied alongside the BS 476-7 result?
Is the perimeter joint evidence from a perimeter rig, or from a furnace test being offered as equivalent?
Does anything in this package test the wall as built, and if not, has that been asked for separately?

A rating belongs to a tested assembly, not to the material inside it. The class tells you what the product is made of. The classification tells you what one build-up did in one furnace. Neither tells you what the wall does. Name the component each requirement applies to, and submittal review stops being an argument about certificates and becomes a check against a drawing.

Cavity barriers, linear joint seals and firestopping are what Agnimatic supplies, matched to the test evidence for the condition on site rather than to the clause.

Agnimatic Technologies LLP, passive fire protection. Standards named for reference: BS 476-6, BS 476-7, BS 8414, EN 13501-1, EN 13501-2, EN 1363, EN 1364-6, EN 1365, EN 1366-3, EN 1366-4, EN 13823, EN ISO 1182, EN ISO 11925-2, EN ISO 1716, ISO 13785-1, ISO 13785-2, ASTM E2307, UL 2079, NFPA 285. Values in the diagrams are illustrative and must not be used for design. This article is general guidance, not a specification for any particular building.