Академический Документы
Профессиональный Документы
Культура Документы
18,313-325 (1994)
Polyester, epoxy and phenolic glass-reinforced laminates were compared in terms of their fire properties in the cone
calorimeter. The presence of a gel coat on the polyester and epoxy products had some effect, mainly in the FR grades.
The FR polyester with a brominated flame retardant showed a marked decrease in rate of heat release and in rate of
smoke production. Aluminium hydroxide (commonly called alumina trihydrate or ATH) caused a significant delay in
ignition time for the epoxy product and some reduction in rate of heat release but rate of smoke release was still
relatively high. Phenolic laminates showed a long ignition time and relatively low rate of heat release (RHR). The peak
RHR was further reduced by the presence of aluminium hydroxide and by changing the type of glass reinforcement.
The influence of thickness was explored for a series of phenolic laminates in both cone calorimeter and the
International Maritime Organization (IMO) surface flammability test. Ignition delay time was significantly affected
and the cone calorimeter results showed that the shape of the RHR curve was also changed as the thickness increased.
However, the heat release measurements in the IMO test were not sensitive enough to pick up the differences in this
property. The type of substrate under the laminate significantlyaltered the burning rate but this may have been because
they were not glued or mechanically fixed.
INTRODUCTION EXPERIMENTAL
*Address for correspondence: 64 Spencer Road, Strawberry Hill, Twickenham, Middlesex, TW2 5TQ, UK.
GP polyester
GC Scott Bader 65 PA 500 gm-’ Cray Valley Total QC 3OLSE 3 x 450g CSM (PPG Ind.) 1 Week at room
Catalyst: Butanox M50, 1% Catalyst: Butanox M50, 1% Resin :glass 1.1 : 1.O temperature then 8
hours at 75°C
NGC None As GC sample As GC sample As GC sample
FR polyester
GC As GP polyester Scott Bader 356 PA 3x450g CSM (PPG Ind.) As above
Catalyst: Butanox M50, 1YO Resin :glass 2.5 : 1.O
NGC None As FR GC sample As FR GC sample
GP epoxy
GC Ciba Geigy SW2402B 100 pts Ciba Geigy CY219 100 pts 3x450gCSM As above
Ciba Geigy HY2402 20 pts Ciba Geigy HY219 50 pts Resin : glass 1.8 : 1.O
Coverage 1 kg m-’ Ciba Geigy DY219 4 pts
NGC None As GC sample As GC sample As above
FR epoxy
GC As GP epoxy Ciba Geigy CY219 100 pts As GP epoxy As above
ATH Grade FRF80 100 pts
Ciba Geigy HY219 50 pts
Ciba Geigy DY219 4 pts
NGC None As GC sample As GP epoxy As above
GP phenolic
CSM Cellobond J26/047L Cellobond J2018L 4 x 4 5 0 g CSM 2 hours at 60°C
Catalyst: Phencat 10 then 4 hours at 60°C
WR66 As phenolic CSM 66wtYo WR As phenolic CSM
WR44 As phenolic CSM 44W% WR As phenolic CSM
Samples for evaluating the effect of thickness on fire RESULTS AND DISCUSSION
properties in the cone calorimeter and IMO flame spread
tests were phenolic laminates made with chopped strand Polyester laminate
glass mat. The thicknesses were as follows:
Panel no. No. of layers Thickness (mm) Results are shown in Table 3 and Figs 1-7. The gel coat
1 1 1.o applied to the surface of the general-purpose laminate
2 3 3.0 tended to slightly lengthen the ignition delay time at
3 6 5.5 external flux levels of 35 and 50 kW m-’ (Fig. 1)but there
4 10 9.5 was little or no significant effect on rate of heat release
(Fig. 2). In contrast, the gel coat applied to the FR
The samples used to evaluate effect of substrate were also laminate tended to have a marginally negative effect on
phenolic GRP and of thickness 3 mm. The surface was the ignition delay time at 50 and 75 kW m-’ and increase
coated with Trimite AP211/AE245, a flame retarded rate of heat release. This is not unexpected since the gel
paint system developed for phenolic laminates. coat itself (Scott Bader 65 PA resin) did not contain a
flame retardant additive. The presence of the gel coat had
Sample preparation no significant effect on smoke production (Fig. 6).
The Class 1 polyester GRP was manufactured from
The cone calorimeter specimens were wrapped in alumi- Scott Bader brominated polyester resin 356PA. Very
nium foil as specified in the standard6 and placed in the pronounced differences from the general-purpose (NFR)
specimen holder over a ceramic fibre blanket at least 13 grade were observed in all the measured fire properties
mm thick. All the tests were performed with the retainer except the smoke specific extinction area. Ignition delay
frame in place, as recommended in I S 0 5660-1. The IMO time was doubled for the non-gel-coated product and
test specimens were prepared according to the standard almost double but from a better base line for the product
and tested on the specified backing board with no addi- with the gel coat (Table 3). Peak RHR was more than
tional substrate. halved in both cases. This is clearly seen by comparing the
GLASS-REINFORCED PLASTIC LAMINATES 315
Table 2. Sample thickness and glass content Table 3. Effect of flame retardant and gel coat - polyester GRP
Description Mean thickness Glass content With gel coat Without gel coat
(mm) (WtYo)
General-purpose (NFR) NFR FR NFR FR
polyester with gel coat 2.38 38.2 Ignition time (s)
General-purpose (NFR) 35 kW m-’ 54 95 41 93
polyester without gel coat 2.0 43.7 50 kW m-2 31 53 25 62
Class 1 Flame retarded (FR) 75 kW m-2 12 24 13 31
polyester with gel coat 3.13 43.4 RHR peak (kWm-2)
Class 1 Flame retarded (FR) 35 kW m-2 347 144 327 112
polyester without gel coat 2.75 43.5 50 kW m-2 37 1 176 374 159
General-purpose (NFR) 75 kW m-2 439 206 47 1 174
epoxy with gel coat 3.25 32.6 RHR t,s to 180 s (kWm-2)
General-purpose (NFR) 35 kW m-’ 190 65 168 24
epoxy without gel coat 2.88 29.6 50 kW rn-’ 202 68 187 52
Flame retarded (FR) 75 kW m-2 199 103 193 100
epoxy with gel coat 3.88 39.5 RHR av. (kWm-2)
Flame retarded (FR) 35 kW m-’ 79 74 78 38
epoxy without gel coat 2.88 42.9 50 kW rn-’ 108 66 115 49
General-purposephenolic 75 kW m-’ 133 78 109 83
woven roving glass mat 3.5 66 EHC ( M J kg-’)
General-purpose phenolic 35 kW m-2 21.6 7.2 21.2 4.9
chopped strand glass mat 3.5 44.3 50 kW m-2 21 6.3 20.8 6.3
General-purpose phenolic 75 kW m-2 20.3 8.1 18 8.3
woven roving glass mat 3.0 40 SEA (m2kg-’)
FR phenolic chopped 35 kW m-’ 935 701 1035 843
strand glass mat 3.7 c 40 50 kW m-2 966 1002 1001 975
75 kW m-2 891 967 971 1004
Note: All the polyester and epoxy samples contained chopped SMK PARAM. (MW kg-’)
strand glass mat as reinforcement. 35 kWm-’ 324 101 338 94
50 kW m-2 358 176 374 155
75 kW m-’ 391 199 457 175
35 w/mZ
50 W/m2
75 w/m2
100 w/ml
g 100
50
. .
1 2 3 4 5 6 ‘ 7 ’ 8 ’ 9 ‘10
Lamblatr
Figure 1. The effect of gel coat and flame retardant. Ignition time in cone
calorimeter.
heat release curves shown in Figs 4 and 5 with Fig. 3. The (Fig. 3), suggesting that the lower layers were decom-
FR product burned in two distinct phases under external posing and producing combustible volatiles at the same
heat flux levels of 35 and 50 kW m-2 although a rather time as the upper layer was burning. The FR species were
more continuous burn pattern was experienced under an presumably able to delay this effect at 35 and 50 kW m-2
exposure of 75 kWm-2. Repeatability between speci- but not at the very high flux of 75 kW m-2.
mens was very good-this is shown for the 75 kWm-’ The effective heat of combustion (EHC), calculated
burn in Fig. 5 but similar close correspondence was also from total heat released and total mass loss, was signific-
shown at the other two heat flux levels. It is interesting to antly reduced from an overall average of 20.5 MJkg-’
note that the NFR laminate showed a single RHR peak for the NFR samples to 6.9 MJ kg-’ for the FR ones.
316 M. J. SCUDAMORE
350 1 I
1.R6CwR
ZR6CR I
3001
:
0
200
c
9
+ 150
g 100
50
n
1 . . . . . .
1 2 3 4 5 6 7 8 9 10
Lamlnate
Figure 2. The effect of gel coat and flame retardant. Early RHR in cone
calorimeter.
I \\UkW,d
I . .*.........._.
.......
100 200 300 400 500 1 10
Rnq-
This indicates that the combustion mechanism is being expressing the EHC for the product as a whole would be
interfered with, probably in the vapour phase, by the to divide the total heat released by the original mass of the
brominated species from the flame retardant resin. The specimen, including the glass layers.
above figures represent the EHC for the polyester mater- Average specific extinction area (SEA)results were not,
ial within the laminate product. A more useful way of in general, significantly affected by the presence of the
GLASS-REINFORCED PLASTIC LAMINATES 317
40-
.
20-
.
0 100 200 300 400 so0 800
h q d
1200
1
1000
35 W/mZ
50 W/m2
r” 800
2E 75 W/ml
. 600
>:
100 M/m2
d
d
# 400
200
0 . . . . . . .
1 2 3 4 5 6 7 8 9 10
Laminate
Figure 6. The effect of gel coat and flame retardant. Smoke SEA in cone
calorimeter.
flame retardant (Fig. 6) but is important to note that SEA but with ATH FR agent added. This decomposes at the
takes no account of the rate at which the smoke is relatively low temperature of approximately 200°C and
produced. One suggestion for overcoming this problemg the large volume of water vapour released is the active
is to express the smoke in terms of smoke parameter agent in the flame retarding mechanism. It seems clear
which is defined as the product of the mean SEA and the from the averaged results in Table 4 that this water
peak RHR. When smoke generation is expressed in this release was having a significant role in increasing ignition
way (Fig. 7) the FR grade polyester is seen to be superior delay time in comparison with the NFR product.
to the general-purpose grade because the smoke is being RHR was reduced as a result of ATH incorporation
produced at a much reduced rate. (Fig. 2) but the reduction was much less than that
produced by the brominated FR polyester resin. Effective
heat of combustion was somewhat reduced for the FR
Epoxy laminate grade but it appears that the ATH is having a much less
significant effect on the combustion mechanism than is
The gel coat used for the epoxy laminates (Ciba Geigy the halogenated FR agent in the polyester.
SW2402B with HY 2402) had little effect on the ignition The effect of flame retardant on the RHR curve is
delay time when used on the NFR product but this time illustrated in Figs 8-10. The effect is not as striking as in
was significantly reduced in the case of the gel-coated FR the corresponding polyester comparison and the double
laminate at 35 and 50 kWm-’ (Fig. 1). There was no peak observed for the FR product was also present in the
marked effect of the gel coat on RHR or smoke genera- NFR grade. Specific extinction area was slightly less in
tion. FR laminates (Fig. 6) and this difference is increased when
The FR epoxy product was made from the same resin smoke production is expressed in terms of smoke para-
(Ciba Geigy CY 219) as for the general purpose laminate meter (Fig. 7) because of the reduction in RHR.
318 M.J. SCUDAMORE
600
U
500 1 I 35 W/m2
P
2 400 50 Wjm2
Fw/m2
300 100 W/m2
s
#& 200
cn
100
n . . . . . . . . .
1 2 3 4 5 6 7 8 9 10
Laminate
Figure 7. The effect of gel coat and flame retardant. Smoke parameter in cone
calorimeter.
J
A75 W/na Table 5. Effect of flame retardant - phenolic GRP
NFR FR
11 I:, \ 50 kW m-’
75 kW m-’
25
60
21
37
8
12
8
18
100 kW m- ’ 17 17
50
50 W/nQ
40-
20 -
O.6 ’ 100 O* 300 400
Time, seconds
500 600 700 840 oi 100 260 360 460
Time, seconds
560 660 760 6O
:
Figure 12. FR phenolic laminate. Rate of heat release. Figure 13. FR phenolic laminate. Rate of heat release.
as the external heat flux level increases (Fig. 15). Surpris- was 50 kWm-’ the thinner laminates (1 m m and 3 mm)
ingly, the thinnest laminate (one layer only) did not showed a sharp, clearly defined peak but the thicker (5.5
exhibit flaming combustion within the 900 seconds limit and 9.5 mm) products showed a very pronounced double
put on the test at the lowest flux level of 35 kWm-’ peak. As the external flux was increased further to 75 and
although when the thickness was increased to 3 mm (three 100 kW m-’ these initial peaks coincided and there was
layers) flaming ignition occurred at approximately 200 also some evidence of an initial peak from the 3 mm
seconds. The RHR curves also show a significant depend- product in close proximity to those from the other two.
ence on thickness (Figs 16-18). At a heat flux level of The second peaks were well separated for the three
35 kW m-’ a single peak was observed at all thicknesses different thicknesses and the average times at which the
but this became broader and reduced in height as thick- peaks occurred are summarized in Table 8. A likely
ness increased from 3 mm to 9.5 mm. When external flux explanation of the above behaviour, largely borne out by
1
“he,,
be eliminated if smoke is expressed directly in terms of
visibility using units of optical density or of obscura”
3-too4 I where 1 obscura (Ob) is equivalent to human visibility of
75 W/ne
approximately 10 m. Thus,
20
0
0 100 200 300 400 5W 000 700 000 Where L-path length and I, and I are the light in-
Time, seconds tensities in the absence and presence of smoke, respect-
Figure 14. NFR phenolic laminate-woven roving. Rate of heat
ively. Swedish workers’ commonly express smoke pro-
release. duction rates as the product of smoke obscura and the
volumetric flow rate through the duct, i.e. Ob.m3s-’.
Results from the present study expressed in these terms
show that the rates of smoke production mirror fairly
the observations, is that the thicker products delaminate closely the rate of heat release curves. This is illustrated in
under heat flux conditions and volatiles released from the Figs 19 and 20 for the 50 kW m-* burns. Smoke ex-
separated lower layers escape through the side of the pressed in terms of smoke parameter (Table 7) show that
specimen, even though these are protected by means of the rate of smoke production tends to increase with
the edge frame, to cause flaming around the edge of the external flux at all thicknesses.
exposed surface. When the same samples were tested in the IMO surface
Smoke results, expressed as specific extinction area, flammability test the effect of thickness on ignitability was
were rather variable and no significant trend with thick- confirmed with the value of Qi, heat required for ignition,
ness could be detected. This is probably because SEA increasing as laminate thickness increased from 1 mm to
1000 -I i
900 -
800 -
{
53 700-
600-
500-
/-
-
:::w
400
2 300-
OO 2 12
Thldaress, mm
Figure 15. Phenolic laminate. The effect of thickness on ignitability.
im.
20.
n
-0
L& 200 400 600 800 1000 1200 1400 1600 11
Figure 17. The effect of thickness on RHR-external flux 50 kW m-'. Phenolic GRP.
5.5 mm but not thereafter (Table 9). There was no obvious calorimeter and because of the single heat flux profile
effect on flame spread, however, with critical flux for employed.
extinguishment (CFE) staying approximately constant.
At all thickness the product satisfied the requirements for
bulkhead, wall and ceiling lining in ships. No clear trend Substrates
was observed in the heat release measurements from
the IMO apparatus, probably because of the rather For the short series of cone calorimeter tests run with
cruder measurement system in comparison with the cone phenolic laminates over various substrates the laminate
GLASS-REINFORCED PLASTIC LAMINATES 323
200 'I 1
lmjn 'gp.
110
Tlirywc
Figure 18. The effect of thickness on RHR-external flux 75 kW m-2. Phenolic GRP.
35 1 - -
3 350 -
5.5 650 -
9.5 1300 -
50 1 80 -
3 190 -
5.5 200 420
9.5 300 900
75 1 50 -
3 90 100
5.5 100 320
9.5 1 00 750
100 1 25 - 2!
3 40 80
5.5 50 280
9.5 30 500 Figure 20. The effect of thickness on smoke production. Phenolic
laminatesexternal flux 50 kW m-2.
1 2 1 2 1 2 1 2
Ignition time (s)
50 kW -’ 1’39 97 157 206 265 60 95 176 240
75 kW m-’ 60 30 110 97 103
RHR peak (kWm-’)
50 kW m - z 95 89 85 39 60 72 48 130 152
75 kW m-’ 77 64 65 71 187
RHR t,, to 180 s (kWm-’)
50kWm-’ 26 53 59 31 41 48 103
75 kWm-’ 43 55 50 102
RHR t, to 360 s (kWm-’)
50 kW m-’ 39 39 43 26 32 65 93
75 kW m-’ 34 43 40 106
RHR av. (kW m-’)
50 kW m-’ 29 20 29 22 28 24 21 46 57
75 kW m - z 29 36 37 27 89
EHC ( M J kg-l)
50 kW m-’ 20.4 11.9 11.7 6.6 7.8 4..1 3.2 5.6 12.6
75 kWm-2 20.4 26.4 13.2 8.1 13.6
SEA (m’kg-’)
50 kW m-’ 167 96 89 17 24 35 7 33 67
75 kW m-’ 138 101 137 12 40
SMK PARAM. ( M W kg-’)
50 kW 21 11 10
75 kW m-’ 14 7
Smoke vol(m30D1)
50 kW m-’ 0.73 1.01 0.91 0.23 0.43 0.13 0.13 2 2.1
75 kW m-‘ 0.65 0.62 1.?2 0.16 1.44
200 - -
was used can be explained by the involvement of the
180- substrate itself in the fire after approximately 600 seconds
but the difference in the shapes of curves from the other,
largely inorganic, substrates is less easy to explain.
Specific extinction area and effective heat of com-
bustion results would be expected to be misleadingly low
for those substrates which evolve water or other volatiles
on heating. In these cases a non-mass-related smoke
value is preferable and in Table 10 total smoke volume in
m3 of smoke of unit optical density is also shown. The
greatest volume of smoke resulted from the specimens
tested on the particleboard but calcium silicate board and
paper-faced plasterboard showed low smoke volumes.
Overall, it appears that more consideration needs to be
given to ways in which thin materials should be tested on
their end use substrate in the cone calorimeter. It is
probably not sufficient to place the material on the
substrate as in these examples; some form of giueing or
mechanical fixing may be necessary to prevent the under-
side of the specimen or the top of the substrate becoming
(in duplicate) and at 75 kW m-’ (singly). The results in involved in the fire even though the edge frame is in
Table 10 show that repeatability was not very good position. The selection of glue is, of course, very import-
aIthough some significant differences can be discerned. ant since this may also become involved in the fire and
Ignition time was shortest for the runs with plasterboard have a bearing on the results.
substrate followed by Rockwool and air whiie specimens
with calcium silicate board and wood particleboard
substrates took longest to ignite.
The substrate had a significant effect on the rate of heat CONCLUSIONS
release curves (Fig. 21). This graph shows the most severe
burn from each replicate pair tested at 50 kW m-2. The The reaction-to4re properties of glass-reinforeed thermo-
large double peak obtained when the wood particleboard set plastic laminates depend on the type of resin and flame
GLASS-REINFORCED PLASTIC LAMINATES 325
retardant, the type of glass reinforcement and, for thin The cone calorimeter edge frame as normally used is
laminates, the thickness. The type of substrates can also probably not capable of preventing delamination and
have a significant effect for thin laminates. escape of combustible volatiles to the sides of the specimen
Brominated or inorganic flame retardants were re- and hence to the exposed surface. This phenomenon was
quired for polyester and epoxy laminates in order to most clearly observed for the phenolic laminates. For this
reduce the rate of heat release and rate of smoke produc- type of product a redesigned frame which prevents lateral
tion. In the case of the phenolic laminates, however, no escape of volatiles is needed to obtain meaningful cone
FR additive or treatment was required to obtain results calorimeter results.
for these properties which were already lower than for the
FR polyester. Incorporation of ATH in the phenolic resin
resulted in an even lower RHR and smoke production.
REFERENCES
1. T. Ohlemiller, Assessing the flarnmabjlity of composite mater- 7. IMO Resolution A653 (16). Recommendation on improved fire
ials. NET Report NlST JR 89-4039 (1989). test procedurefor surface flammability of bulkhead, ceiling and
2. J. E. Brown, J. J. Loftus and R. A. Dipert, Fire characteristicsof deck finish materials. Adopted on 19 October 1989. Inter-
composite materials-a review of the literature. NlST Report national Maritime Organization, London.
NBSlR 85-3226 (1986) 8. BS 476: Part 7: 1987. Fire tests on building materials and
3. J. E. Brown, E. Braun and W. H. Twilley, Cone Calorimeter structures, Part 7. Method for classification of surface spread of
evaluation of the flammability of composite materials. NlST flame of products. British Standards Institution, London.
Report NBSIR 88-3733 (1988). 9. V. Babrauskas, J. Fire & Flammability12, 51 (1981).
4. V. Babrauskas, Developmentof the Cone calorimeter-A bench 10. E. Mikkola, On the thermal ignition of combustible materials.
scale heat release rate apparatus based on oxygen consump- Fire and Materials 14 (1989).
tion. Fire and Materials 8. 81-95 (1984). 11. D. J. Rasbash and R. P. Philips, Fire and Materials 2, No. 3,
5. ASTM E-1354-90, Standard test method for heat and visible 102-9 (1978).
smoke release rates for materials and products using an oxygen 12. 6.A. L. dstman and L. D. Tsandaridis, Smoke production in the
consumption calorimeter. American Society for Testing and cone calorimeter and the room corner test. Fire Safety J. 17,
Materials, Philadelphia. 27-44 (1991).
6. IS0 5660-1:1993 Fire tests-reaction to fire. Rate of heat
release from building products (cone calorimeter method). In-
ternational Standards Organization, Geneva.