Professional Documents
Culture Documents
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Temperature
Controller
Specimen
3. Test results
typical stress levels in practice. The furnace was then heated in
accordance with the experimental fire curve, as shown in Fig. 3. The 3.1. Temperature and time
fire curve attempted to follow the widely used ISO [13] fire curve
(also known as a standard temperature–time curve) which can be
The variation of the surface temperature on a typical specimen
used to assess the fire resistance capacity of structures. Due to the
(i.e. TS-1000-1) against time is shown in Fig. 5. The horizontal
limitation of the furnace used in this test program, the maximum
axis of this figure represents the test time (in minutes) and the
heating rate could not exceed 100 °C/min. Therefore, the initial
vertical axis represents the temperature (in °C). Fig. 5 shows that
part of the ISO fire curve could not be achieved, as can be observed
there are two peaks in each of the three thermocouple curves. The
in Fig. 3. In order to simulate the ISO fire curve as closely as possible,
first peak occurs when the specimen was heated to over 350 °C
though, the following heating procedure was adopted.
while the second is at around 640 °C. Shortly after these two peaks
• During the first 5 min of heating, the maximum heating power were reached, the surface temperature on the specimen was higher
of the furnace was used, namely 100 °C/min, in order to raise than the temperature of the furnace. The following comments are
the temperature inside the furnace from 20 °C up to 520 °C. therefore offered as a means of explanation.
• During the following period, various heating rates in the range At the first peak, the epoxy ignited, although it had already
5 –25 °C/min were adopted in order to make the furnace
started to decompose upon the glass transition temperature Tg
temperature match the ISO fire curve from minute to minute.
(around 60 °C) being reached. Upon ignition, gas was emitted and
The air temperature in the furnace was sometimes higher than the FRP plate caught fire, as observed in Fig. 6. When the epoxy
expected because of the inevitable overshooting of the heaters in had completely burned off, the temperature on the specimen then
the furnace. This was due to the adoption of the proposed heating dropped. The second peak, at around 600 °C, was due to heat
procedure and use of the maximum heating rate capability. being released from oxidation of the carbon fibres. In this case, CO
K. Wang et al. / Engineering Structures 33 (2011) 2154–2161 2157
Table 1
Ultimate stress, failure mode, elastic modulus, and strain at break of specimens at room temperature.
Group Specimen Temperature Holding time Ultimate stress Failure Elastic modulusc Strain at break
identificationa T (°C) (min) fu,normal (MPa) modeb Enormal (GPa) εu,normal (%)
G1 (room temperature) SS-22-0-1 22 0 3239.3 I 188.0 1.70
SS-22-0-2 22 0 3169.8 I 193.4 1.65
SS-22-0-3 22 0 3052.1 I 193.7 1.61
SS-22-0-4 22 0 3072.5 I 191.4 1.57
a
The specimen identification is described in Table 2.
b
The failure modes are described in the text.
c
For calculation of the ultimate stress and elastic modulus, the composite area and the measured width and thickness of the specimen were used to determine the
cross-sectional area of the specimen. The range of strain for calculating the elastic modulus is 1000–3000 microstrain [23].
Table 2
Temperature, ultimate stress, and failure mode of tests at elevated temperatures.
fu,T
Group Specimen identificationa Temperature T(°C) Holding time (min) Ultimate stressb fu,T (MPa) fu,normal
Failure modec
Table 3
Properties of Sika⃝
r
CarboDur⃝
r
1214 plates at room temperature [24].
Property Thickness (mm) Fibre content by volume (%) Tensile strength Tensile modulus Strain at rupture (%)
Mean value (MPa) Min value (MPa) Mean value (GPa) Min value (GPa)
and CO2 were produced [4], and the maximum loss of fibre mass The results of the Group G2 tests show that the ultimate
occurred. Oxidation of carbon fibres was also observed in the tests strength and failure modes for the specimens were almost the
of Foster and Bisby [4]. same as compared to the corresponding specimens in Group G1.
2158 K. Wang et al. / Engineering Structures 33 (2011) 2154–2161
T = 22°C
T = 308°C
T = 50°C T = 420°C
T = 103°C T = 520°C
T = 625°C
T = 155°C
T = 211°C T = 706°C
T = 599°C
T = 152°C 2000 MPa
T = 507°C
T = 180°C 1500 MPa
T = 135°C
T = 163°C 1061 MPa
(b) Group G2 time series, holding time =30 min. (c) Group G3 stress series.
(1 + a)
f 1−a
= tanh (−b (T − c )) + , (1)
f0 2 2
where f is the mechanical property in question (i.e. tensile
strength, elastic modulus, bond strength) at temperature T (in
degrees Celsius, °C), f0 is the room-temperature value of the
mechanical property in question, and a is an assumed constant,
with the coefficients b and c obtained by least-squares regression Fig. 8. Stress–displacement relationship of Group G1.
of test databases. The proposed values of a, b, and c are provided
in [7]. While there is a considerable scatter of the results in [7] hyperbolic tangent function model is presented as follows [15]:
(i.e. low R2 values), the analytical curves presented in [7] captured
the overall trends in the data gathered. Evaluation of Eq. (1) with Pinitial + PR Pinitial − PR
P (T ) = Rn · −
the test results reported in this paper shows the predictions to 2 2
deviate somewhat. Such discrepancy may be due to Eq. (1) being
originally calibrated from scattered data from a variety of FRP
products. × tanh (km · (T − Tcentral )) , (2)
More recently, Gibson et al. [25] proposed a model which was
derived from composite laminate theory for elevated temperatures where P (T ) is a particular mechanical property (i.e. the elas-
up to 200 °C. More specifically, the model takes the form of tic modulus, strength) at temperature T , Pinitial is the room-
a hyperbolic tangent function which predicts the temperature temperature value of that property, PR is the average value of
evolution through the thickness and the profile of the residual the property at high temperature, km is a constant describing the
resin content, which in turn reflects the extent of thermal damage breadth of the distribution which is determined by a nonlinear
of the FRP [25]. Subsequently, Chowdhury et al. [15] calibrated multi-parameter least squares regression analysis, and Tcentral is the
Gibson’s [25] model with glass FRP (GFRP) coupon tests and FRP- temperature around which the curve is nearly symmetrical (note
to-FRP bond strength tests also for temperatures up to 200 °C. The that Tcentral is not necessarily equal to Tg ). In addition, Rn accounts
2160 K. Wang et al. / Engineering Structures 33 (2011) 2154–2161
(a) Stress–displacement relationship of Group G2. (b) Stress–displacement relationship of Group G3.
Table 4 fibres (Mode IV). At 300 °C, the ultimate strength of the plates
Coefficients of proposed ultimate strength equation. reduced to approximately 50% of that at room temperature. In
Temperature (°C) Coefficients addition, when the temperature rose to around 700 °C, the test
A B C n specimens experienced a significant drop in strength compared
22 ≤ T < 150 1.00 22 200 0.9 to the room-temperature strengths. Finally, an equation for the
150 ≤ T < 420 0.59 150 490 0.7 ultimate strength of pultruded CFRP plates for temperatures
420 ≤ T < 706 0.48 420 76,000 1.8 ranging from room temperature to about 700 °C has been proposed
in this study.
the strength properties at 395 °C. In addition, Pinitial is also equal to
1.0, which is the average of the observed values of the property at References
room temperature. Owing to a lack of relative test results, Tcentral
[1] Hollaway LC, Teng JG. Strengthening and rehabilitation of civil infrastructures
is estimated to be 100 °C, since, at such temperature, the trend
using fibre-reinforced polymer (FRP) composites. Cambridge (UK): Woodhead
of test data below 200 °C in Fig. 12 is nearly symmetrical. Using Publishing Limited; 2008.
a nonlinear single-parameter least-squares regression analysis, km [2] Smith ST, Hu S, Kim SJ, Seracino R. FRP-strengthened RC slabs anchored with
becomes 0.0139. The resulting calibrated curve shown in Fig. 12 FRP anchors. Eng Struct 2011;33(4):1075–87.
[3] Leone M, Matthys S, Aiello MA. Effect of elevated service temperature on bond
matches the test results very well (i.e. the R2 value is 0.93). The between FRP EBR systems and concrete. Composites: Part B 2009;40:85–93.
model is, however, limited to temperatures up to about 400 °C. A [4] Foster SK, Bisby LA. Fire survivability of externally bonded FRP strengthening
new modelling approach is therefore required in order to describe systems. J Compos Construct, ASCE 2008;12(5):553–61.
[5] Kodur VKR, Bisby LA, Foo SHC. Thermal behavior of fire-exposed concrete
the complete tested temperature range. slabs reinforced with fiber-reinforced polymer bars. ACI Struct J 2005;102(6):
In order to determine a new model, inspiration is drawn 799–808.
from metals subjected to high temperatures. More specifically, [6] Stratford TJ, Gillie M, Chen JF, Usmani AS. Bonded fibre reinforced polymer
strengthening in a real fire. Adv Struct Eng 2009;12(6):867–78.
the stress–temperature relationship for the CFRP pultruded [7] Bisby LA. Fire behavior of fiber-reinforced polymer (FRP) reinforced or
plates reported herein exhibited the same shape as that of confined concrete, Doctor of Philosophy Dissertation, Kingston (Canada):
elevated temperature tests on stainless steel and cold-formed steel Department of Civil Engineering, Queen’s University; 2003.
[8] Hawileh RA, Naser M, Zaidan W, Rasheed HA. Modeling of insulated CFRP-
materials conducted by Chen and Young [26,27]. As a result, the
strengthened reinforced concrete T-beam exposed to fire. Eng Struct 2009;
stress–temperature model presented in [26,27], which is given in 31(12):3072–9.
Eq. (3), is proposed herein. [9] Kodur VKR, Bisby LA, Green MF. Preliminary Guidance for the design of FRP-
strengthened concrete members exposed to fire. J Fire Prot Eng 2007;17(1):
fu,T (T − B)n 5–26.
=A− . (3) [10] Bisby LA, Green MF, Kodur VKR. Fire endurance of FRP-confined concrete
fu,normal C columns. ACI Struct J 2005;10(6):883–91.
[11] Gao WY, Hu KX. Fire resistance research of CFRP-strengthened concrete
In Eq. (3), the coefficients A, B, C , and n are calibrated for structures and prospect. Fiber Rein Plast/Compos 2008;2:47–52 [in Chinese].
different temperature ranges with the steady-state and transient- [12] Williams B, Kodur VKR, Green MF, Bisby L. Fire endurance of fiber-reinforced
polymer strengthened concrete T-beams. ACI Struct J 2008;105(1):60–7.
state test results presented herein. The calibrated coefficients are
[13] ISO 834-1. Fire-resistance tests — elements of building construction — Part 1:
presented in Table 4. Despite the small scatter in the data in the General requirements. International Standard ISO 834. Geneva; 1999.
three zones (i.e. the R2 of three zones are 0.90, 0.82 and 0.96, [14] Eedson RT, Bisby LA, Green MF. Performance of FRP strengthening systems
respectively), the model is robust, and is capable of describing the for concrete during exposure to elevated temperatures. In: Proceedings (CD-
Rom), 9th International symposium on fiber reinforced polymer reinforce-
entire temperature range. ment for concrete structures. 2009.
[15] Chowdhury EU, Eedson R, Green MF, Bisby LA, Benichou N. Mechanical
characterization of fibre reinforced polymers materials at high temperature.
4. Conclusions Fire Technol 2009;45(4):1–18.
[16] Cao S, Wu Z, Wang X. Tensile properties of CFRP and hybrid FRP composites at
An experimental investigation of the mechanical properties of elevated temperatures. J Compos Mater 2009;43(4):315–30.
[17] Han LH, Wang WH, Yu HX. Experimental behaviour of reinforced concrete (RC)
pultruded CFRP plates at elevated temperatures for both steady-
beam to concrete-filled steel tubular (CFST) column frames subjected to ISO-
state and transient-state tests has been presented. It should be 834 standard fire. Eng Struct 2010;32(10):3130–44.
noted that the results are specific to the type of resin and fibre [18] Lim L, Buchanan AH, Moss PJ. Numerical modelling of two-way reinforced
as well as to the fibre volume fraction of the plate tested. In the concrete slabs in fire. Eng Struct 2004;26(8):1081–91.
[19] Lim JBP, Young B. Effects of elevated temperatures on bolted moment-
steady-state tests, the temperatures ranged from approximately connections between cold-formed steel members. Eng Struct 2007;29(10):
20 to 700 °C, whereas in the transient-state tests the stress ratios 2419–27.
ranged from 0.34 to 0.64. The failure modes and ultimate strengths [20] Chen J, Young B. Cold-formed steel lipped channel columns at elevated
temperatures. Eng Struct 2007;29(10):2445–56.
have been discussed and the stress–displacement relationships [21] To ECY, Young B. Performance of cold-formed stainless steel tubular columns
have been plotted. The FRP plate was found to ignite when at elevated temperatures. Eng Struct 2008;30(7):2012–21.
the temperature reached around 350 and 600 °C. These two [22] Young B, Ellobody E. Performance of axially restrained concrete encased steel
composite columns at elevated temperatures. Eng Struct 2011;33(1):245–54.
self-ignitions were caused by the epoxy matrix transforming [23] ACI440 3R-04. Guide test methods for fiber-reinforced polymers (FRPs) for
into a gaseous state and then by oxidization of the carbon reinforcing or strengthening concrete structures. Farmington Hills (USA):
fibres, respectively. Generally, the stress–displacement relation American Concrete Institute; 2004.
[24] Sika. Technical Data Sheet for Sika⃝ r
CarboDur⃝r
Plates. Sika Hong Kong Pty
was essentially linear from room temperature to temperatures as
Ltd.; 2004.
high as 520 °C. Four different failure modes were then identified [25] Gibson AG, Wu YS, Evans JT, Mouritz AP. Laminate theory analysis of
and described, namely (i) rupture of the FRP plate (Mode I), composites under load in fire. J Compos Mater 2006;40(7):639–57.
(ii) partial loss of epoxy matrix followed by plate rupture (Mode [26] Chen J, Young B. Stress–strain curves for stainless steel at elevated
temperatures. Eng Struct 2006;28(2):229–39.
II), (iii) complete loss of epoxy matrix (Mode III), and (iv) complete [27] Chen J, Young B. Experimental investigation of cold-formed steel material at
loss of epoxy matrix as well as considerable oxidation of the carbon elevated temperatures. Thin-Walled Struct 2007;45(1):96–110.