Professional Documents
Culture Documents
ABSTRACT
E-glass /fire retardant Vinyl ester composite is one of the most commonly used face-
sheet/laminate in sandwich and composite structures. Evaluating its post-fire mechanical
properties is critical to predict the residual strength of structural systems after compartment fires.
This paper presents an experimental approach for assessing the residual capacity of
environmentally exposed and fire damaged composite laminates. The approach involves
capturing the post-fire response of a typical Biax-±45° E-glass/Vinyl ester laminate exposed to
different heat fluxes (10, 20, and 30 kW/m2) using the cone calorimeter (ASTM E1354). A total
of 96 samples were tested for shear, tension, and compression. Strength and stiffness reductions
are proposed to safely predict the post-fire structural performance of these structural systems.
Response and failure mechanisms were clarified using optical microscopy and thermomechanical
analysis on the ambient and exposed laminates.
1. INTRODUCTION
Composite structures are currently being used in buildings, emergency temporary housing, and
modular shelters. These structures are constructed of highly insulated light weight sandwich wall
panels and composite structural frame members. The monolithic laminates and face sheets are
commonly made of reinforced woven roving or non-crimped knitted E-glass with fire retardant
vinyl ester and phenolic resins. Due to their inherent shear load demand, shear walls are usually
constructed of double Bias [±45] fabric construction. When exposed to marine environment,
structures are often subjected to salt spray exposure that is either transported to the structure by
direct splashing or airborne currents. Like any construction materials, fiber reinforced composite
materials are susceptible to degradation due to severe environmental exposure. In cold regions,
effect of Freeze-Thaw (F/T) and thermal cycling on FRP composites is a major design concern
for most structural engineers. In addition, standards and building codes dictate that all
construction materials as related to building functions and occupancy, meet certain flame spread,
smoke and toxicity requirements. These limitations are established by the governing
building/non-building code and often require some sort of protective coating such as intumescent
and phenolic coatings. Recent advancement in composite materials can meet these design
challenges related to flammability and durability. However, a frequent question always arise, “if
a low-to-medium intensity fire was developed in the vicinity of a structural wall, could the wall
be still reliable to resist the demanding load or should the structural member be repaired or
replaced?”, in addition, “how would the environmental effect magnify the degradation in the
composite material?”
Copyright 2017. Used by CAMX – The Composites and Advanced Materials Expo.
CAMX Conference Proceedings. Orlando, FL, December 11-14, 2017. CAMX – The Composites and Advanced Materials Expo
Related Post-fire mechanical properties on GFRP in marine and naval ships have been
extensively studied and reported in the literature [1-13]. Simplified models to predict the post-
fire mechanical properties were developed. These models were based on the rule of mixture,
specifically predicting the residual tension and compression strengths and moduli. Most of the
correlated experimental data and validation included unidirectional lamina, woven roving and
quasi-isotropic laminates. None of the experimental work included angle ply or double bias
construction. In addition, and to date, no single experimental characterization combined the
accelerated weathering and post-fire structural performance of GFRP. Further work is needed to
assess the post-fire properties of pultruded [14] and sandwich structures failure mechanisms [15].
The capability to understand the mechanism of failure and predict the residual mechanical
properties of laminates following low-intensity fire is still needed.
This experimental work investigates the degradation and post-fire damage mechanism of a
composite laminate before and after F/T exposure. Using state-of-the art
destructive/nondestructive evaluation techniques, this research paper tends to fill this gap and
examine the failure modes induced due to accelerated weathering and fire-induced damages. A
double bias laminate construction using non-crimped knitted E-glass Vinyl ester commonly used
in shear-web girders and shear walls are the main focus of this research.
2. EXPERIMENTATION
2.1 Materials and Samples Fabrication
Vacuum infusion molding process under high vacuum was used to mold all composite laminates.
This process is generally used for large civil engineering structural applications. The materials
selected to manufacture the composite panels are shown in Table 1. The +45/-45 fiber
architecture was used in this study due its commonality in structural shear walls and shear web
applications. The measured thickness of the E-glass/ vinyl ester was 3.0 ±0.1 mm thick with 52
% fiber volume fraction. Panels were then cured for 30 days at room temperature (22.5 °C and
50 % RH), then cut to specific ASTM test standards and prepared for environmental and
mechanical testing, both for pre and post-fire mechanical testing.
Fibers
Samples Fabric Construction Resin Catalyst Molding Process
Direction
2
E-BXM-1708 E-Glass, 883 gr/m DERAKANE
2 1.25%
total. [+45°(304 gr/m ), -45°(304 610 C-200 Organic VARTM Infusion with
2 2
EBXM1708 gr/m ), Chopped Mat (275 gr/m )] Epoxy Vinyl Peroxide, ±45° 28 Hg Vacuum
® Ester Resin ® Pressure
(Vectorply Corporation ) (Ashland Inc.) Cadox
2.2 Environmental Exposure:
In presence of water saline solution (3% NaCl + 97% distilled water), freeze/thaw cycling (-20
°C to 40 °C) was conducted on 42 samples. Figure 1a depicts the 24-hour period of F/T cycle
used in this study. The total aging process for the exposed samples lasted 120 days. Samples
were submerged in a sliane solution inside the stainless steel pans as shown in the below Figure
1b.The freeze/thaw (F/T) slab tester by Qualitest TM (Model number EN 196-1) shown in Figure
1b was used to environmentally expose all samples.
b)
a)
Figure 1. a) Freeze-Thaw Cycle Profile (1 cycle =24 hours), b) Samples inside the
Environmental chamber
2.3 Fire Exposure:
Fire test on the composite laminates were conducted using a FTT Dual cone calorimeter (East
Grinstead, United Kingdom). The specimens tested inside the cone calorimeter were heated on
one side, and placed flat 25 mm away from the radiant heat on aluminum foil wrapped ceramic
wool pad. A fresh ceramic wool pad was used for each test. The center portion of each coupon
test was symmetrically centered below the cone. All unexposed (baseline) and exposed (F/T
samples) composites were fire tested at incident heat fluxes of 10, 20, and 30 kW/m2 for 7
minutes (Table 2). This heating time was approximated as the time for the fire fighters to engage
in extinguishing the fire. When/if the sample ignited, it was immediately removed and placed
under an exhaust hood and covered with a stainless steel snuffer which was flushed with nitrogen
to quickly extinguish the sample. The time of heat exposure was recorded for each sample
tested. If the sample did not ignite after 7 minutes of exposure, it was removed and placed under
the exhaust hood and allowed to cool to room temperature. Char thickness was then measured for
each burnt sample (Figure 2 and Figure 3) and then prepared for the post-fire mechanical testing.
Tensile, shear and compression testing were performed on a total of 96 samples (Table 2). To
monitor the strains up to failure, strain gauges were installed back-to-back for all compression
samples, and on one side for the tension/shear samples. For the burnt samples, bonding the strain
gauges was difficult, especially for the 30 kW/m2 samples. For this specific batch, the gauges
were installed on the unburnt side of the specimen. All samples were tested under room
temperature (20 °C) using a 250 kN MTS machine (Alliance 300).
Group
Number of
Tests Exposure and Incident Heat Test Method Exposure Time (a) (Average)
Samples
Flux
Baseline - 6
F/T Exposed - 6
2
10kW/m + 7 min 6
2
20kW/m 6 min and 9 sec 6
Tension and ASTM D3039
Shear 2
ASTM D3518
30kW/m 2 min and 1 sec 6
2
F/T + 10kW/m + 7 min 6
2
F/T + 20kW/m 6 min and 1 sec 6
2
F/T + 30kW/m 1 min and 6 sec 6
Baseline - 6
F/T Exposed - 6
2
10kW/m + 7 min 6
2
20kW/m + 7 min 6
Compression ASTM D6641
2
30kW/m 2 min and 1 sec 6
2
F/T + 10kW/m + 7 min 6
2
F/T + 20kW/m 6 min and 2 sec 6
2
F/T + 30kW/m 2 min and 2 sec 6
(a): Ignition time specified for samples with less than 7 minutes threshold
Figure 2. Char thickness of unexposed and F/T exposed cross-section samples after fire testing.
Char Thickness Vs. Heat Flux
3
2.5
1.5
1
Unexposed-kW/m2
0
10 20 30
Heat Flux, kW/m2
3. RESULTS
All Freeze/Thaw exposed samples reported in the following sections were removed from the
environmental F/T chamber and allowed to dry for 7 days before fire and mechanical testing.
3.1 Mechanical Testing:
Figure 4, 5 and 6 show the failure modes of all baseline, F/T exposed and burnt samples. Both
baseline and F/T exposed samples failed at the center of the specimens under compression load,
whereas under tension load, most of the baseline and the 10kW/m2 samples failed at the ends of
the specimen in the vicinity of the MTS grips. This premature failure mode is still acceptable by
the ASTM D3518/D3039 and tends to underestimate the strength/stiffness of the samples. Most
fire F/T exposed samples subjected to several fluxes failed at the localized burnt damaged area as
expected. This is specially noted for the 20 kW/m2 and 30 kW/m2 heat fluxes.
a)
b)
Figure 4. Baseline (unexposed) and exposed Failure modes a) Failed by compression, b) Failed
by Tension/Shear
Figure 5. Failure modes of burnt samples - Compression Test
a) b)
Figure 7. Post-fire strengths of a) F/T exposed samples and b) baseline (unexposed) samples
a) b)
Figure 8. Post-fire modulus of a) F/T exposed samples and b) Baseline (unexposed) samples
Most of the post-fire tests reported in the literature are on Woven and quasi-isotropic laminates.
With these types of fiber architectures, the reduction in compression modulus was the most
detrimental. This is due to buckling/micro-buckling of the fibers aligned with the axial
compression load at the fire damaged zone. For a double Bias construction, the behavior is
significantly different. Figure 8 clearly shows that the compression moduli of all samples are
significantly higher than the tension moduli. This response could be explained by a simple truss
mechanism where the square bars simulate the +45/-45 fiber direction constructing the lamina.
When such lamina is under axial tension load, the sides of the truss (or fibers) are under
compression, and therefore at the cross section level, the already damaged/wrinkled lamina due
to the induced heat tend to fold apart from each other, leading to excessive deformation and
reduction in stiffness. This is clearly depicted in the sketch shown in Figure 9. In addition, matrix
cracks (intralaminar) are areas of local stiffness reductions and can eventually propagate by local
structural instability of the fibers. Under compression load (Figure 10), the sides of the truss tend
to elongate and set the fibers under tension (unfolding mechanism-stiffening effect), which in
turns are not that heavily damaged as the matrix. This behavior is clearly sketched in Figure 10.
top Cross section behavior
view
Figure 9. Tension failure mode (30 kW/m2), Fibers in cross-section under compression (solid
lines represent anticipated deformation- dashed line intial position)
Side
view Cross section behavior
Figure 10. Compression failure mode (30 kW/m2), Fibers in cross-section under tension (solid
lines represent anticipated deformation-dashed line intial position)
Reinforcing the previous findings related to the structural performance of fire-damaged [+45/-
45] angle ply laminates, Figure 11 shows the precent reduction in compression and tensile
moduli for the unexposed and F/T exposed samples. For all burnt samples (unexposed and F/T
exposed) a higher reduction in tensile moduli are observed with respect to the baseline. This
ranged between 60 % and 80 %. For the compression modulus, this range spans from 40%-60%.
This clearly confirms that even if the compression and tensile strengths are relatively comparable
for all burnt samples in a [+45/-45] fabric architecture, the mechanics of failure due to the
induced heat shifts the failure modes to unpredicted outcome. It is also interesting to highlight
that for the F/T exposed samples (Figure 11, fourth exposure group) the reduction in
compression modulus is higher than the tensile modulus. This behavior is valid to all burnt
samples. This was anticipated due to the interlaminar cracks and delamination under extreme F/T
environment. Additional microstructure examination and DMA results on the baseline, exposed
and fire damaged laminates are presented in the next section.
ea
80.00%
70.00%
% Reduction
60.00%
50.00%
40.00%
30.00%
20.00%
us us us us us u s us us us u s us us us us
du l dul dul d ul dul dul d ul d ul dul dul d ul d ul dul dul
o o o o o o o o o o o o o o
M M M M M M M M M M M M M M
li e on i l e o n i l e on i l e on i l e
on i l e on i l e
on
n s ssi ns ssi n s ssi ns ssi n s ssi ns ssi n s ssi
Te p re Te pr e Te p re Te pr e Te pr e Te pr e Te pr e
m m m m m m m
Co Co Co Co Co Co Co
2 2 2 T 2 2 2
Exposure /m /m /m F/ /m /m /m
kW kW kW kW k W kW
10 20 30 10 20 30
d d d
an an an
T T T
F/ F/ F/
Figure 11. Percent Modulus reduction with respect to baseline laminate for unexposed, F/T
exposed, and Fire damaged samples.
Figure 12. Optical image showing microstructure of the laminate a) baseline, b) interafacial
failure after F/T exposure, c) matrix cracking at 10 kW/m2 and c) F/T exposure and burnt at 10
kW/m2- Intralaminar and Interlaminar cracks shown
Figure 13 displays the F/T exposed and burnt samples at 20 kW/m2 flux. In general,
delamination reduces the overall stiffness of the laminate. Figure 13b clearly illustrates such
delamination. The frequency, location and size of the cracks are easily observed.
Figure 14 shows significant matrix crazing and damage in all plies after 30 kW/m2 flux. The
“zigzag” crack shape shown in Figure 14b propagated along the entire width of the sample and is
believed to cause severe out-of-plane bending or buckling under in-plane axial loads. Ultimately,
this size of a damage is a source of stress concentration and eventually reduced the strength of
the burnt materials. In addition, the width of the crack shown in figure 14b creates a zone of local
stiffness reduction, and hence can propagate by local structural instability.
Figure 13. Optical image showing cracks and thermal degradation of the laminate a) burnt at 20
kW/m2, b) F/T exposed and then burnt at 20 kW/m2
Figure 14. Optical image showing microstructure of the laminate a) burnt at 30 kW/m2, b) F/T
exposed and then burnt at 30 kW/m2-noticeable matrix crazing and sizeable cracks
The Dynamic mechanical analysis (DMA) was used as a tool to assess the morphology and
viscoelastic transition response of the laminate. The bending mode was used in the DMA test,
and the char layer was located on the compression side (top side) of the three-point bend fixture.
This layout was systematically consistent for all samples considered. Figure 15 shows the
variation of the tangent δ with temperature, as a function of exposure. It is shown that the glass
transition temperature subsequently increased with the induced heat flux on the laminate. This
was significant from the 10 to 20 kW/m2 samples. For the baseline sample, the Tg was estimated
around 98 0C, assumed at the peak of the curve, and for the 20 kW/m2 it was estimated at 106 0C.
This is probably due to residual curing which out-paces any decrease in Tg from thermal
degradation of polymer or combustion decomposition. There was little or no significant change
from 20-30 kW/m2. It seems that the induced heat in the unburnt region increased the
crosslinking of the matrix.
Figure 15. Tan δ DMA Bending mode testing (5 0 C/ min, 1 Hz, coupon size 50 mm x 12 mm x
3 mm)
To further understand the post-fire failure mechanisms using the same heat induced fluxes,
additional mechanical and thermomechanical analysis (TGA) are scheduled for a different
laminate construction (cross-ply [0/90]). Further post-fire testing on laminates and sandwich
structures is of critical need for the infrastructure and composite construction sectors.
4. CONCLUSIONS
The growing use of E-glass-Vinyl ester composites in shear wall and structural applications
demands a better understanding of their durability and relative post-fire structural performance.
Relevant to this application, the fiber architecture chosen in this experimental work constituted
of double Bias E-glass (+45/-45//mat) reinforced Vinyl ester resin. Composite laminates were
fabricated using the VARTM process. The effect of fire-induced damage under different fluxes
on strength and modulus before and after 120 days of freeze-thaw (F/T) exposure in saline
solution is investigated. Results on exposed and burnt specimens are presented for compression,
shear and tension. The degradation and change in microstructure of the materials were
characterized using several destructive/nondestructive techniques; the following results could be
drawn:
6. REFERENCES
1. Mouritz AP, Mathys Z. Post-fire mechanical properties of marine polymer composites.
Comput Struct 1999;47:643–53.
2. Mouritz AP, Mathys Z. Post-fire mechanical properties of glass-reinforced polyester
composites’. Comput Sci Technol 2001;61:475–90.
3. Mathys Z, Gardiner CP, Mouritz AP. Tensile and compressive properties of GRP
composites with localised heat damage. Appl Compos Mater 2002;9:353–67.
4. Mouritz AP. Post-fire flexural properties of polyester-, epoxy- and phenolic based
composites. J Mater Sci 2002;37:1377–86.
5. Mathys Z, Mouritz AP, Gardiner CP, Townsend CR. Mechanical properties of GRP
composites with localised heat damage’. Int J Mater Prod Technol2002;17:134–42.
6. Mouritz AP. Simple models for determining the mechanical properties of burnt FRP
composites. Mater Sci Eng A 2003;359:237–46.
7. Gibson AG, Wright PHN, Wu YS, Mouritz AP, Mathys Z, Gardiner CP. Modelling
residual mechanical properties of composites after fire. Plast Rubber
Compos2003;32:81–90.
8. Mouritz AP. Fire resistance of aircraft composite laminates. J Mater Sci Lett
2003;22:1507–9.
9. Gardiner CP, Mathys Z, Mouritz AP. Post-fire structural properties of burnt GRP plates.
Mar Struct 2004;17:53–73.
10. Mouritz AP, Mathys Z, Gardiner CP. Thermomechanical modelling the fire properties of
fibre–polymer composites. Composites 2004;35B:467–74.
11. Bai Y, Keller T. Modeling of post-fire stiffness of E-glass fiber-reinforced polyester
composites. Compos A 2007;38:2142–253.
12. Henderson JB, Wiebelt JA, Tant MR. A model for the thermal response of polymer
composite materials with experimental verification. J Comput Mater1985;19:579–95.
13. Lattimer BY, Ouellette J, Trelles J. Thermal response of composite materials to elevated
temperatures. Fire Technol 2011;47:823–50.
14. Correia, João R., Yu Bai, and Thomas Keller. "A review of the fire behaviour of
pultruded GFRP structural profiles for civil engineering applications." Composite
Structures 127 (2015): 267-287.
15. Anjang, A., V. S. Chevali, B. Y. Lattimer, S. W. Case, S. Feih, and A. P. Mouritz. "Post-
fire mechanical properties of sandwich composite structures." Composite Structures 132
(2015): 1019-1028.