What the BRE knew but did not say: part 2, the large-scale cladding test results of 1993-4.

In my last article, I stated that the BRE tested 6 AD B compliant cladding systems at full-scale in 1993-4, and that all failed the test according to the criteria set. This fact is not explicitly acknowledged in Ray Connolly's report of the tests, nor to my knowledge has it so far been recognised at the Grenfell Inquiry. The purpose of this article is to demonstrate, from the data given in the report, the veracity of my previous assertion.

The tests were conducted at Cardington Laboratory:

on a purpose-designed test facility with a range of 6 external walls, at least two of which had window openings on the 1st, 2nd and 3rd storey levels (p. 8):

The fire source was a timber crib, positioned either in the hearth at ground storey level to represent a fully developed room fire emerging out of a window, or 10 cm away from the wall to represent an external ground level fire. It was designed to give a heat output of 1.7 MW over 40 minutes, and the measured maximum flux on a non-combustible fa?ade at 1m above the hearth was about 80 kW/m2.

Ten rainscreen systems were tested. Systems 1 to 4 were tested on walls with windows, the remainder on walls without windows. Eight systems had polyester GRP sheeting (system 1 description shown):

and two (systems 4 and 6) had 'thermosetting resin bound cellulose reinforced' sheeting boards (system 4 description shown):

a description to which HPL appears to answer:

Although there are many types of resin bound cellulose cladding products used in construction, I have not yet definitely identified any apart from HPL that were being employed for external rainscreen cladding, and I will refer to these sheets henceforth for convenience as HPL[?] sheets.

It appears from the observations of collapse that both types of sheeting had an aggregate finish on at least one face (systems 2 and 6 shown):

The aggregate finish of the GRP panels can be seen in a photograph of test 1:

The HPL[?] panels of system 6 appear to have a smooth finish:

which seems hard to reconcile with the observation that some areas of the cladding collapsed as a shower of aggregate. Was the description of collapse of aggregate perhaps in error?

In her examination of Sarah Colwell, Inquiry Counsel cites Dr Connolly's (not yet released) Inquiry witness statement to the effect that System 2 was designed 'to be representative of, though not to replicate exactly', the Knowsley cladding system:

Knowsley Heights was clad with GRP Stenni (now 'Steni') panels with 'a surface finish of natural aggregate' (p. 39):

presumably on the outward facing side only, since it is hard to see what function it would have served on the cavity side. According to the New Civil Engineer, the resin was polyester (p. 29):

They had Class 0 fire performance, as reported by BRE:

but probably on one side only, as one of the 'Implications for Building Regulations' that the BRE identified from the Knowsley fire was that AD B should require Class 0 for the cavity facing side of rainscreen panels, as well as for the outward facing side:

If they were not Class 0 on the inside face, and assuming there was no finish on that face, then it appears to follow that the GRP material was not inherently Class 0 and that the panels achieved Class 0 on the outside face because the GRP was partially protected by the aggregate.

In the Connolly tests, both types of cladding sheets were Class 0 on both sides:

the GRP sheets thus differing from the Knowsley sheets in this respect. How had they achieved Class 0 on the inside face? Had the manufacturer begun to employ aggregate on the inside face also, in order to comply with the 1992 AD B? Was some other surface treatment employed? The report is silent on the matter. If fire retardants had been added to the GRP, then I am not sure that the system could accurately have been said to be 'representative' of the Knowsley system.

The Class 0 requirement for the inside face of the outer cladding had indeed been incorporated into AD B(1992) in accordance with BRE's suggestion:

Systems 3 to 8 had cavity barriers at every floor level:

as required by AD B(1992) 9.2 and Table 13:

The insulation in all systems was mineral wool, as shown for example in the system description for test 3:

thus satisfying 12.7:

Even the caution in the first paragraph of 12.7 about the use of combustible materials for cladding framework:

was heeded, aluminium support rails being used in all systems, as shown for example in the system description for test 3:

Systems 3 to 8 were thus fully compliant with current AD B guidance. If any of these six systems were shown to be hazardous, then the adequacy of the cladding guidance would be called into question.

Fire safety objectives and pass/fail criteria

Four aspects of cladding system fire hazard were distinguished:

or to put it another way, the implicit fire safety objectives were to prevent:

• i) fire spread from storey to storey through windows;
• ii) undue spread of fire either on the surface of or within the cladding system;
• iii) collapse of large sections of cladding.

Accordingly the following criteria were set for acceptable fire safety:

1) The heat flux on the centre of the windows should be ≤ 80 kW/m2:

2) Flame spread should not rise to the second floor:

Flame spread on the surface was detected by visual observation. Invisible spread was detected by thermocouples within the cavity, on the aluminium fixing rails, and behind the insulation. Fire was considered to be present where temperatures exceeded 400o C above ambient:

3) No debris should fall which might injure a fire fighter with a helmet:

Results

The four non-compliant systems failed by very wide margins. Of the six compliant systems, four failed the test by very wide margins, and system 4 by a somewhat narrower, though still considerable, margins. System 3 failed by a wide margin according to its fire spread diagram, which might possibly be in error, however, as I explain below. It certainly failed by a small margin at least.

Taking the compliant systems in order:

System 3

System 3 had GRP cladding, a 90mm cavity, a 1mm steel plate cavity barrier, and local venting through an 8mm gap either side of each barrier.

Maximum heat flux at the 1st floor window was 46 kW/m2, well below the limit set of 80 kW/m2:

Cladding collapse was limited to a shower of aggregate, without large debris falling.

External flame spread extended to the second floor window, failing to meet the criterion of no spread to the second floor, but by a modest margin.

According to the caption to Plate 8 (p. 16) there was no fire spread in the cavity above the second floor barrier:

but the fire spread diagram shows 'internal' fire spread up to the third floor window:

If the diagram is correct, then the fire spread criterion was breached by a large margin. If both caption and diagram are correct then the fire spread must presumably have been behind the insulation, rather than in the cavity. If only the caption is correct, and there was in fact no internal fire spread beyond the 2nd floor barrier, then the fire spread criterion was breached by a small margin, externally.

System 4

System 4 had HPL[?] cladding, an 80mm cavity, and a 1mm steel plate barrier with 8mm gaps.

Fire spread was contained to the 1st floor by the barrier between 1st and 2nd floors:

There was no collapse.

The sheeting directly above the fire, however, burned quickly and noisily:

and heat flux on the 1st floor window was 110 kW/m2, well above the limit set of 80 kW/m2:

System 5

System 5 had the same cladding system as System 3, with GRP cladding, a 90mm cavity and the 1mm plate barrier with 8mm gaps. The fire source was external, however, and there were no windows.

According to the caption to Plate 12, fire passed the 2nd floor barrier, and the 3rd floor barrier too on the outside surface, reaching the top of the structure:

The apparent implication is that internal fire did not pass the 3rd floor barrier. The fire spread diagram, however, shows surface spread up to around the 3rd floor barrier, and cavity spread up to the top of the structure:

It seems hard to reconcile the diagram with the Plate 12 caption. The fire spread criterion was breached by a very wide margin, whether externally or internally.

System 6

System 6 had HPL[?] cladding, an 80mm cavity and 1mm plate barriers with 8mm gaps. The fire was external.

The cladding surface burned very rapidly, up to the top of the rig:

and the diagram shows surface spread almost to the top of the rig:

Strangely, however, Connolly then says that fire spread 'did not reach the top of the test building', and that the fire hazard could be considered 'acceptable':

even though the fire spread criterion he had set had been breached by a very wide margin!

System 7

System 7 had GRP cladding, a 90 mm cavity, and intumescent cavity barriers at each floor level. Ventilation was at the top and bottom of the fa?ade only.

Flames emerged from the top of the building:

as indicated also in the diagram:

The cladding sheet material was 'largely destroyed' and 'fell from the building':

Under the 'Collapse' performance indicator, only a falling shower of aggregate is recognised:

System 8

System 8 had GRP cladding, a narrow 35mm cavity, and 1mm plate barriers with 8mm gaps.

Fire spread on the surface beyond the 2nd floor barrier, and in the cavity to the top of the building, emerging from the top:

as shown also in the diagram:

The cladding sheets 'fell away' up to the level of the 2nd floor barrier:

which might appear to indicate the collapse of large debris. Under the 'Collapse' Performance Indicator, however, only the shower of aggregate is acknowledged:

Connolly explains in his Discussion that on occasion, despite the the resin having burned off, the glass fibre held sections of the sheeting together, which then fell in 'larger masses'. It was considered unlikely, however, to cause serious injury to fire fighters, and therefore not to be a breach of the collapse criterion:

Reasons for failure

In tests 3, 5, 6 and 8, the 1mm steel plate cavity barriers failed because of the melting of the aluminium rail (which supported the barriers by design):

allowing fire spread through the cavity.

Intumescent barriers, which were employed in System 7, and which are intended to close the cavity at fire temperatures, but allow ventilation at normal temperature, were found to be ineffective in that test:

from which Connolly felt able to draw the conclusion that intumescent barriers were inadequate:

Fire also spread over the surface, bypassing the barriers, even up to the top of the rig in some tests, despite the cladding sheets having achieved Class 0, as Connolly himself observed:

Both GRP (System 5), and HPL[?] (System 6) sheets exhibited this excessive surface fire spread.

The action required

The failure of just one AD B compliant cladding system should have been enough to prompt an urgent reappraisal of the guidance. The failure of all six compliant systems demonstrated its complete inadequacy, and should have result in prompt action, consistent with the deficiencies that had been identified.

Provisional guidance should have been given not to use intumescent barriers, until such time as they products had been proven to be effective in large-scale testing.

Second, guidance should have been given to prevent cavity barriers failing due to melting or weakening of aluminium rails. Connolly concluded that they should be 'substantially fixed' to the masonry substrate, and thus not dependent on the rails for support:

Third, the Class 0 requirement should have been replaced by something more stringent. It had been shown by these large-scale tests of 1993-4 to be dangerously inadequate. Only BRE and the DoE were privy to this information:

The power to take immediate action lay with government. It seems extraordinary that having commissioned the research, and having thereby identified extremely serious deficiencies with their guidance, they apparently did nothing to remedy them, and left industry in ignorance that they could be designing and installing dangerous cladding, despite following the recommendations of AD B.

BRE had the opportunity in 1999 to inform the ETR Committee of what they knew, and failed to take it. More than this, they indicated their assent to the government submission to the Committee, by saying that they had 'nothing.. to add' to it. The government claimed that the risk of serious fire spread with AD B compliant cladding systems was 'minimal'. The BRE knew that this was not true.

Andrew Chapman