How did PE ACM panels achieve Class 0, given that 'Zerobond' failed to do so?

Since 2017 I have gathered documentary evidence for PE ACM products achieving the national reaction to fire classification Class 0 through testing to BS 476 parts 6 and 7, and found the following nine products:

Reynobond 160. May 1997. (Classification report, part 7 report, part 6 report.)

PE ACM panel tested by BRE for government under contract cc1924 in 2001 or 2002. (Analysis of test data.)

Alucobond. Certificate dated 24 March 2005. (BBA Certificate 2005, BBA Certificate 2013.)

Reynobond 33, 3.4 mm core, 0.3 mm facings. September 2006. (Classification report, part 7 report, part 6 report.)

Reynobond 33, 2.4 mm core, 0.3 mm facings. September 2006. (Classification report.)

Reynobond 33, 1.4 mm core, 0.3 mm facings. September 2006. (Classification report, part 7 report, part 6 report.)

Larson. Certificate dated 22 April 2008. (BBA Certificate.)

Alubond [Uncertain whether PE or FR]. Certificate dated 1 March 2012. (Certifire Certificate.)

Larson. Certificate dated 22 January 2014. (Certifire Certificate.)

Despite the extreme danger associated with using this type of product for cladding high buildings, it was not at all surprising that they could achieve Class 0 through testing, for the simple reason that in both the underlying tests the attack is directed against the aluminium facing rather than against the PE core.

In May 2019 however, as described in an article in Inside Housing, Ian Abley commissioned tests of samples of Reynobond PE ACM, under the fictitious name 'Zerobond', at Warrington Fire, to determine whether they could achieve Class 0. The product achieved Class 1 to BS 476-7 (Surface Spread of Flame), but failed to achieve the requisite indices to BS 476-6 (Fire Propagation), the tests being invalidated through the escape of flaming molten polyethylene through the air inlet valve of the combustion chamber:

the invalidation being triggered by the provisions of BS 476-6:1989+A1:2009 at 9.2(b):

This provision seems logical, since flaming material escaping from the chamber will not contribute to the temperature of the flue gases, and thus to the temperatures at the thermocouple (between chimney and cowl), from which the indices are derived.

Three specimens of the same product are tested, and each test lasts twenty minutes. In Zerobond test 1, no molten material was visible through the air inlet in this photograph taken at 'around 17 minutes', according to Ian Abley's records:

although the yellow light may perhaps be suggestive of flaming material out of sight.

Shortly afterwards, flame became visible in the chamber:

and increasingly so:

before beginning to escape from the chamber:

According to the Laboratory Record Sheet, this escape of material occurred at 19:38, just 22 seconds before the end of the test:

but Abley considers it possible that material had begun to escape a short time before the escape was observed.

A photograph of the panel specimen taken after the test suggests that most of the PE core was gone:

The molten PE that escaped the test apparatus fell through a hopper into a sand box, where the uncombusted portion resolidified:

Abley tells me that the PE continued to burn in the chamber even after the test, and perhaps even after the apparatus had been placed in the 'cooling cupboard'. He has described it as 'a horrible burning viscous liquid fire'.

In test 2, flame was visible in the chamber 'before 17 minutes', according to Abley's observations:

and grew intense:

before, again, molten flaming material escaped from the air inlet:

This photograph, however, appears to have been taken after the test, since Abley is certain that the gas supply was disconnected after the completion of the test. The Laboratory Record Sheet for test 2 contains no invalidation observation about the escape of molten material:

The test report states that all three tests were invalidated because of material either escaping through the air inlet or being confined to the recess of the specimen holder:

A photograph of the holder for specimen 2 taken after the test shows little sign of PE having escaped into it:

If escape into the holder was not the cause of invalidation, and the invalidation was properly triggered, then it follows that there was escape from the air inlet before the end of the test. Abley's recollection also is that there was an escape from the air inlet before the end of each test.

A photograph of Specimen 2 after the test suggests that little of the PE core remained:

There appears to be a hole in the facing through which molten PE could possibly have entered the chamber. It is not certain whether this was formed before or after the end of the test. If the molten PE flowed out of the side edges of the panel then, since its dimensions (225 mm x 225 mm) are greater than those of the rear of the chamber (190 mm x 190 mm) which it forms the back to,:

it is not obvious (at least to me) why it would flow into the chamber in preference to the holder. It could perhaps have been the intensity of the PE fire in the chamber that caused the damage to the facing.

At 'around 17' minutes in test 3, there is perhaps some indication of flaming beginning out of direct sight:

Fuller flaming is visible just before the end of the test:

As in test 2, we have no photograph or laboratory record demonstrating conclusively an escape from the air inlet before the end of the test, but only the fact of invalidation, Abley's recollection, and certain inferences which may be drawn from the temperature data, as discussed briefly below.

For test 3, there is a photograph of PE that has escaped from the joint between the chamber and the holder:

It seems to me at least arguable therefore that test 3 was not conducted according to BS 476-6:1989+A1:2009 since section 5.4 makes reference to 'obtaining an adequate seal' between chamber and holder:

Zerobond temperature data

The laboratory record sheet for each specimen shows the thermocouple temperatures at time intervals, alongside the calibration temperatures, which are recorded before the test of the specimen with a specified calcium silicate non-combustible sheet in its place. In test 1:

the 'Specimen Temperature' (not the temperature of the specimen, but the thermocouple temperature for the specimen under test) is less than or equal to the calibration temperature for the first 16 minutes. At 18 minutes, however, it is 305 - 236 = 69° C higher than the calibration temperature, and at 20 minutes it is 87° C higher.

What is generating this rise in temperature towards the end of the test? Aluminium is non-combustible at normal fire temperatures. The contribution from combustion of the coating may be expected to be too small to raise temperatures above calibration temperatures (see some evidence for this below). The PE can hardly combust while it remains in place between the facings of the panel for lack of oxygen, and it seems to me that the same is probably true of PE that escapes into the recess of the holder, where in any case, being protected by the intact panel from the heat sources, temperatures would presumably be too low for ignition and combustion. It seems to follow that the only plausible cause is that molten PE is flowing out of the panel into the chamber, where it ignites and combusts. This also is what was observed.

The contribution to the indices is given by (Ts - Tc)/10t (with only positive values being used for the calculation) so that the contribution made by a positive temperature differential is inversely proportional to the time elapsed. In consequence, substantial rises in temperature late in the test lead only to quite modest values for the specimen index of performance.

It may be noted that the temperatures recorded for the specimen are up to 11° C lower than calibration temperatures:

Given that the calcium silicate sheet used for calibration is non-combustible, how does one account for lower temperatures being observed for the specimen? One possibility would be normal experimental variation. But a similar pattern is observable in tests 2 and 3, in all three tests of Reynobond 160, in all three tests of Reyobond 33 (3.4 mm core), but in none of the 3 tests of Reynobond 33 (1.4mm core), suggesting that some common factor may account for the negative temperature differentials with the thicker panels. One possibility, from a suggestion by Ian Abley, is that the panels absorb heat from the chamber as the PE melts, the latent heat of fusion being given in one article as 102J/g. Another possibility is that the greater thermal conductivity of aluminium causes greater heat loss from the chamber, but this would not account for the difference in pattern between the 3.4 mm and 1.4 mm versions of Reynobond 33.

Temperatures rise somewhat earlier in test 2, rising 35° C above calibration temperature at 16 minutes, 117° C above at 18 minutes, and 96° C above at 20 minutes:

If molten PE entered the chamber earlier in test 2 than test 1 then, other things being equal, one would expect it to begin escaping through the air inlet earlier also. But, as explained above, there is no direct evidence for this.

The temperature data for specimen 3 were not supplied in the Laboratory Record Sheet, but are visible in a photograph taken by Abley:

Temperatures rise 85° C above calibration temperatures at 18 minutes, and 100° C above at 20 minutes.

We can calculate the indices of performance that this temperature data would have given rise to if the tests had not been invalidated:

i(1) would have equalled (0.0 + 0.0 + 0.4)/3 ≈ 0.13

i(2) would have equalled (0.0 + 0.0 + 0.0)/3 = 0.0

i(3) would have equalled (0.82 + 1.35 + 0.97)/3 ≈ 1.05

I would have equalled 0.13 + 1.05 = 1.18

Both I and i(1) would therefore have been more than ten times lower than the limits for Class 0 of I ≤ 12 and i(1) ≤ 6.

Temperature data and indices for other PE ACM products

Apart from 'Zerobond', three of the PE ACM products listed above have BS 476-6 test reports containing the temperature data. The pattern observed, of substantial rises in temperature in the last minutes of the test, is strikingly similar for many, though not all, of the specimens.

1) Reynobond 160. May 1997.

No temperature rise above calibration temperatures was recorded in Specimen 1:

One conclusion that may be drawn from this case is that the contribution from the combustion of the aluminium coating is too small to raise temperatures above calibration temperatures.

Specimen 2 had no rise above calibration temperature for the first 18 minutes, but a 72° C differential at 20 minutes:

What could cause this sharp rise in temperature other than PE entering the combustion chamber?

Specimen 3 saw a small rise of 10° C above calibration temperature at 18 minutes, and an 86° C delta at 20 minutes:

These late but substantial rises above calibration temperature for Specimens 2 and 3 yielded values for s(3) of just 0.36 and 0.48, and an Index of Performance of only 0.0 + 0.0 + (0.0 + 0.36 + 0.48)/3 ≈ 0.3:

2) Reynobond 33, 3.4 mm core, 0.3 mm facings. September 2006.

All three specimens exhibited the same distinctive behaviour.

For Specimen 1, there was no rise above calibration temperature for the first 18 minutes, and a 78° C differential at 20 minutes:

Specimen 2 is of particular interest since already at 16 minutes there was a 37° C rise above calibration temperature, followed by differentials of 119° C and 99° C at 18 and 20 minutes respectively:

This is remarkably similar to 'Zerobond' Specimen 2, which had differentials of 35° C, 117° C and 96° C at 16, 18 and 20 minutes.

Is it plausible that molten PE combusted in the chamber for more than four minutes without any material escaping through the air inlet?

The same question may be asked with respect to Specimen 3, which had differentials of 5° C, 45° C and 132° C at 16, 18 and 20 minutes:

The Index of Performance was just 0.0 + 0.0 + (0.39 + 1.39 + 0.94)/3 ≈ 0.9:

3) Reynobond 33, 1.4 mm core, 0.3 mm facings. September 2006.

Specimen 1 exhibited sizeable temperature differentials of 104° C and 124° C at 18 and 20 minutes. Small differentials of 3° C and 1° C at 4 and 5 minutes gave rise to a non-zero value for s(2):

Specimen 2 of the 1.4 mm core product similarly had some minor and probably insignificant rises of no more than 5° C above calibration temperature during the first 7 minutes, but in this case there was no temperature rise towards the end of the test:

Specimen 3 had rises of up to 8°C above calibration during the first 12 minutes, and then a differential of 13° C at 20 minutes. Given that the temperature was 5° C below calibration at 18 minutes, it could perhaps be argued that this represents an 18° C rise towards the end of the test, suggestive of the beginning of PE combustion in the chamber:

The Index of Performance was (0.00 + 0.28 + 0.45)/3 + (0.10 + 0.22 + 0.42)/3 + (1.21 + 0.00 + 0.08)/3 ≈ 0.24 + 0.25 + 0.43 ≈ 0.9:

Discussion and conclusions

By subjecting Reynobond PE ACM to the BS 476-6 method of test at his own expense, and adding photographs and observations to the temperature data required by the standard, Ian Abley has contributed greatly to our developing understanding of the causes of the Grenfell tragedy. He has shown:

a) that molten PE from the ACM core can enter the chamber during the last minutes of the test;

b) that this material can ignite and combust in the chamber;

c) that this material can escape the chamber through the air inlet, invalidating the test according to the provisions of the standard.

The observed combustion of PE in the chamber towards the end of each 'Zerobond' test corresponds well with the observed rise of thermocouple temperatures towards the end of each test, and serves to explain it, in the seeming absence of plausible alternative explanations.

A comparison with temperature data for Reynobond 160, Reynobond 33 (3.4 mm core) and Reynobond 33 (1.4 mm core) suggests strongly, at the least, and to my mind demonstrates conclusively, that molten PE also entered the chamber and combusted in at least six, and perhaps seven, of the total of nine individual specimen tests.

The potential significance of this behaviour for the fire safety of PE ACM can hardly have escaped those observing the tests. As an internal lining test, BS 476-6 is specifically designed to test the surface only, with efforts made to 'prevent ignition of the underlying layers':

If the material of the 'underlying layer' did in fact ignite and combust despite these efforts, how much more would this be so in a real fire? Were the technicians struck by the intensity of the fire once the PE became involved? Did they report their observations to their superiors at Warrington Fire? And did other testing laboratories also test PE ACM to BS 476-6? I am not sure whether these questions have been asked at the Inquiry but, in case not, the Metropolitan Police should have the opportunity to remedy any omission.

Were there any failures to invalidate?

In Zerobond test 1, a positive temperature differential was first recorded at 18 minutes, and an escape of material through the air inlet was recorded at 19:38. In Reynobond 33 (3.4 mm core) test 2, a temperature differential of 37° C was recorded at 16 minutes, indicating that molten PE was already in the chamber by that time. Is it plausible that there had been no escape through the air inlet before the end of the test at 20 minutes?

I do not know the answer. Abley has said that the molten material is viscous, so that it might travel relatively slowly over the floor of the chamber. Combustion of the material would reduce its mass, and one can perhaps conceive of a kind of steady state in which flow into the chamber was matched by the loss in the exhaust gases. But is that plausible? It seems to me that a little experimentation would probably provide an answer one way or another.

The Inquiry should have investigated the possibility that one or more test laboratories failed to implement the invalidation clause on one or more occasion and wrongly gave a Class 0 classification when it was not merited. Again, it seems to me to be a line of investigation that the police must pursue.

Where is WF 157531?

The Inquiry has obtained both classification and test reports for the 3.4 mm core and 1.4 mm core versions of Reynobond 33, but apparently the classification report only for the 2.4 mm core version. This classification report reveals that the Fire propagation index was 2.0 and that the values of both i(1) and i(2) were zero:

meaning that the positive differentials all occurred in the last ten minutes of the test. This is highly suggestive of the same pattern of temperature rises occurring towards the end of the test, caused, as has been argued, by molten PE entering the chamber towards the end of the test and combusting.

The value of i(3) at 2.0 is about 90% higher than the value of i(3), 1.05, which would have been obtained by 'Zerobond' if the tests had not been invalidated. It is more than twice the result obtained for Reynobond 33 (3.4 mm core) of 0.9. A higher value of i(3) could be obtained by higher differentials, or earlier differentials, or a greater number of differentials. It seems to indicate either more molten PE flowing into the chamber, or an earlier entry of molten PE into the chamber, or both. And the greater the amount of molten PE in the chamber, and the earlier the time, the sooner it will begin to flow out of the air inlet, or so it seems to me.

It seems odd that the Inquiry only has test reports for two of the three Reynobond 33 products that received a Class 0 classification. It seems especially odd that the product for which the test reports are missing is the one which had by far the highest value for the part 6 Fire Propagation Index. I wrote to the Inquiry on 23 April to ask, among other things, why they had not obtained these test reports, but have not so far received a reply that addressed this question.

The Inquiry reference identifiers for the other test reports in the series indicate that they were supplied by Arconic. Has Arconic explained why they did not supply the test reports for Reynobond 33 (2.4 mm core)? Has the Inquiry sought to obtain them from Warrington Fire, who issued them?

It is not acceptable to allow potential miscreants to fail to supply key evidence without any explanation to the public, and without any sort of comeback or protest.

Andrew Chapman