What BRE knew but did not say: part 4, why Class 0 was the wrong requirement for cladding materials.

In this article, I attempt to demonstrate that the BRE must have been fully aware that Class O/0 was an unsuitable requirement for external cladding materials from the time that it was introduced into the national building regulations in something like its longstanding form in 1972 until its opportunity came in 1999 to give evidence to the ETR Committee.

History of the national cladding regulations and guidance

In the first national building regulations of 1965, cladding was required to be non-combustible if it was within 3 feet of a boundary:

If it was more than 3 feet from a boundary, but over 50 feet high, it was required to be Class O, except that timber of thickness ≥ 3/8 inch (≈ 9.5 mm) was allowed on the lower parts of such buildings:

Combustible cladding materials were considered to Class O if they were finished with a layer of non-combustible material at least 1/8 inch (≈ 3.2 mm) thick. Also, non-combustible materials with a combustible finish were classed as Class O if the finish was ≤ 1/32 inch (0.79 mm) thick, and if the whole product had a Class 1 Surface Spread of Flame to BS 476-7:

The aluminium facing of ACM panels is typically 0.5 mm and these therefore could not have been used at that time on high buildings under the regulations even if they had been available on the market. On the other hand, EPS insulation of any thickness with an inorganic render finish would have been compliant, making the regulations less than satisfactory, since such systems can present a severe risk if the finish is penetrated or comes away.

Class O was also required for the internal linings of the circulation spaces and protected shafts of blocks of flats and most other categories of buildings:

Not long after (being reported by Barbara Rogowski in a Fire Research Station (FRS) publication in 1970) there arose a view that the Class O requirement was unduly restrictive towards, in particular, homogenous lining materials that narrowly failed the non-combustibility test owing to a small quantity of organic material :

For this and other reasons, the Fire Propagation test was developed by FRS. It was adopted as BS 476 part 6 in 1968. In the 1972 Building Regulations, the definition of Class O was changed so as to be expressed in terms of BS 476-6 indices achieved in this Fire Propagation test (I have omitted further requirements specific to plastics with low softening points):

In 1976, Class O was renamed Class 0, and a Class 1 Surface Spread of Flame requirement was added to its definition on top of the BS 476-6 indices requirement:

giving it a form similar to that which it effectively retains today for legacy materials:

Hinkley et al (1968), in another FRS Research Note, make clear that BS 476-6 was designed and developed as a test for internal lining materials:

as also was the Surface Spread of Flame test, as explained in 'Fire Grading of Buildings' (1946):

Accordingly, AD B has described Class 0 as a classification for internal lining materials, for example in the 1992 edition:

Applied heat flux

The heat flux on the specimen in the Surface Spread of Flame test is specified in BS 476-7 at 32.5 ± 0.5 kW/m2 at 75 mm along the reference line:

with a maximum of about 3.7 W/cm2 (37 kW/m2), presumably closer still to the radiant panel, as reported by Rogowski. According to Ibrahim, this is not severe enough to indicate the real fire risk of certain materials with protective facings:

Applied heat flux to the specimen in BS 476-6, the Fire Propagation test is higher, rising to around 5 W/cm2 (50 kW/m2) according to Hinkley (a non-combustible material is used to eliminate a contribution from the specimen):

Only materials which meet what Ibrahim calls the 'stringent requirements' of this test (with, I think he should have added, the AD B criteria of I ≤ 12 and i? ≤ 6) are classified Class 0:

A note on the European EN 13501-1 system

In passing, it may be worth mentioning that the Euro EN 13501-1 Small Burning Item (SBI) test was likewise designed for internal linings:

with a design heat exposure to the sample of 30 kW/m2:

less severe than either of the national internal lining tests, while experimental work has shown a maximum heat flux of 40 kW/m2 with the prescribed 30 kW burner:

or even, in another study, an average of 44 kW/m2 near the burner early in the test:

indicating perhaps approximate equivalence with BS 476-6 with regard to severity of fire attack. A 1994 University of Ulster study 'confirmed the veracity' of BS 476-6 evaluations of internal linings, putting the lie to the recent canard that it has long been out of date:

BS 476-7 adds more detailed information about surface spread of flame than is provided by the EN 13501-1 classification system. EN ISO 11925-2 provides information about ignitability that is absent from the national system. Both systems require the product to be tested in conjunction with its end-use substrate. The national tests have a limit of 50 mm on total specimen thickness, which I suppose is probably sufficient for lining materials. It is not obvious to me that one system is superior or inferior to the other as a means to classify internal linings.

Cladding fire exposure

The fire exposure of cladding materials, and especially of rainscreen cladding materials, is very different to that of internal lining materials, for two main reasons:

• 1) the reference fire scenario is different;
• 2) in rainscreen systems, the edges of boards and panels are likely to be exposed to the fire.

The usual reference fire scenario for internal linings is a small fire in a room. One is interested in whether and how quickly it will develop to flashover. The edges of internal linings are generally not exposed, or at least should not be if well maintained. The SBI (Small Burning Item) test, for example, which features prominently in the European classification system, models such a fire with a 30 kW burner.

The usual reference fire scenario for external cladding systems is a fully developed room fire that reaches flashover and breaks a window, with resulting external fire plume. In rainscreen cladding systems, gaps are left between the panels to allow ventilation, with the consequence that the panel edges are exposed in the event of fire. The BS 8414 test models this scenario with a 3 MW fire source, 100 times the power of the SBI test.

The BR 135 (1988) experiments also used a 3 MW heat output, with heat flux on to the facade of at least 100 kW:

In 1989, I. Oleszkiewicz of Canada's national Fire Research Section showed heat flux on a facade from a wood crib fire reaching about 90 kW/m2 above a square window:

and even 140 kW/m2 above a tall narrow window (caption should read '.. 1.5m high'):

The Connolly tests , conducted by BRE in 1993-4, used a 1.7 MW heat output which subjected the lowest part of the facade to a 80 kW/m2 heat flux, which he said was representative of actual fires:

Connolly included another 1989 paper by Oleszkiewicz in his bibliography, showing that he knew the work in Canada, as well as papers by Ondrus and Petterson from Lund University, who had measured heat fluxes of well over 100 kW/m2 with fire loads representing rooms with combustible linings and synthetic furnishings:

as reported also by A. G. Moss in a major 1990 study for the Building Research Association of New Zealand on 'Facade Fire Spread in Multi-Storey Buildings', which FRS must presumably have known:

BRE's Fire Note 3 (1998), a precursor to Fire Note 9 and BS 8414, specified a heat flux of 90 ± 20 kW/m2 at 1 metre above the opening:

The BRE was thus consistent over a period of ten years in expecting heat flux on the facade to be in the region of 80 - 100 kW/m2. Having themselves developed BS 476-6, and being accredited to test to BS 476-7, they would have been fully aware in 1999 that the tests underlying Class 0 applied a much lower of heat flux to the specimens,

Edges

In BS 476-7, the edges of the specimen butt against water cooled steel plates:

Ibrahim explains, citing a paper by Janet Murrell, that one of the purposes of this arrangement was to reduce the influence of edge burning:

It is being assumed that in the end-use application the edges will not be exposed. Likewise, in BS 476-6, the perimeter of the specimen is to be prepared in such a way as to:

i) prevent flame entering a cavity;

ii) prevent ignition of the underlying layers of a product with a flame-retardant coating.

It is not clear to me whether paragraph 4.3 applies in the case of ACM, or indeed GRP with an aggregate surface finish. Neither the aluminium facing nor the aggregate finish are a 'coating' in the normal sense of the word, but it might be possible to argue that the same principle should apply. This could be reasonable if the product were to be used as an internal lining - ACM has been used in foyers for example, without edges exposed - but surely not if the application is rainscreen, with exposed edges.

Many PE ACM cladding products achieved Class 0. But when Ian Abley tested a sample of Reynobond PE ACM, all three specimens failed BS 476-6, molten PE flowing out of the sample into the chamber, igniting, and flowing out of the air inlet:

invalidating the test according to BS 476-6 9.2:

The edges of the specimens were not protected in the Abley tests. I understand that the observations that invalidated the tests were made at 16 and 17 minutes, sufficiently close to the end of the 20 minute test to make it plausible that PE ACM products with exposed edges could sometimes pass the test, given normal experimental and product variability. But another possible explanation for the past achievement of Class 0 by PE ACM is that it has in fact been normal practice to protect the edges in BS 476-6 tests of ACM panel products. The Grenfell Inquiry has not yet investigated this question to my knowledge.

In BRE's April 2000 report (p.22 ff) for North Ayrshire Council on the fire tests conducted on samples of Garnock Court's GRP spandrel panels (p. 25):

Penny Morgan points out (p. 24) that the edges of the spandrel panels had been exposed to the fire:

as shown in this photograph (p. 34) of the living room window of Flat 2:

As part of the test programme, the GRP cladding was subjected (with a minor variation) to what was then still a draft version of the Ignitability test EN ISO 11925-2, which forms part of the EN 13501-1 classification system. The material was tested on both faces, with six specimens subjected to surface exposure and six to edge exposure. Of the six front face surface exposure specimens, two ignited but none exhibited flame spread ≥ 150 mm, whereas of the six accompanying edge exposure specimens, all ignited and exhibited flame spread ≥ 150 mm (p. 47):

None of the rear face surface exposure specimens ignited:

suggesting that the ignition of two of the front face surface exposure specimens may possibly have been due to the yellow coating, which was on the front only.

There is no indication that the rear face had any form of surface finish, and there is no discussion in the report as to why it did not ignite. Morgan devoted a page of her report (p. 28) to photos of the flame spread under edge exposure:

and points out (p. 26) that this test, which was about to be introduced into the UK under the Euro classification system, could become a crucial indicator of the fire behaviour of materials whose flat surface resists flame spread but whose edge does not:

Sadly, it has since become apparent that the attack on the specimen edge in EN ISO 11925-2 is not strong enough to reveal the dangers of edge exposure of PE ACM. Even when subjected to 30 second edge exposure, the PE ACM specimens (strangely described here as 'aluminium sheets') did not ignite in BRE's BR 135 revision test series:

and many PE ACM products have achieved Euro Class D or better, indicating that any flame spread was < 150 mm under 30 second edge exposure.

The prominence given by Morgan to the exposure of panel edges as a contributory factor to fire spread at Garnock Court was sufficiently evident as to be noted in the short summary of her Phase 2 report contained in Cathy Crook's 2002 BRE Report on 'External Fire Spread via Windows':

Discussion of the effect of panel edge exposure is absent however from the BRE's first report to North Ayrshire Council of August 1999, of which Morgan was a co-author along with Brian Martin and Tony Morris. It only pointed out (p. 10) that both sides of the GRP panels were exposed to fire:

It was not that there had been no awareness of the exposure of the edges at the time of the first report. In a 24 June 1999 email (p. 11) to Martin, Morgan reported Dougal Drysdale, longtime Professor of Fire Engineering at Edinburgh University, as telling her that the cladding appeared to overlap the window pod with the result that flames may have acted on 'both sides of the exposed edge of the cladding':

A hand-written note (p. 39), sandwiched in the DETR file between emails dated 5 July and 8 July 1999, describes the 'Moulded GRP around each window. On face [?] between windows, moulded pieces - they overlap - not sealed - fire gone [?] up - caught on [?] exposed edge':

which seems to show that its author believed, even before the panels were tested, that the fire ignited the panels at their exposed edge. It seems odd that this possibility was not discussed in either the BRE's first report to the Council or in their report to government, which merely stated that the 'plume will have ignited the GRP':

The DETR Garnock file also contains notes of a 'discussion on the reason for the rapid vertical spread' with Professor Drysdale on 22 July:

He estimated the heat flux at 80-90 KW/m2:

There was a discussion about Class 0:

whose content was not recorded. Can we imagine Professor Drysdale possibly pointing out that when a cladding panel is tested to BS 476-6 and BS 476-7:

• 1) the applied heat flux is much lower than the level to which it may be subjected in a fa?ade fire; and
• 2) its edges are protected;

with the consequence that a Class 0 classification was of little reliability as an indicator of the panel's performance in a fa?ade fire?

Andrew Chapman