You are on page 1of 12

Composites: Part B 42 (2011) 1739–1750

Contents lists available at ScienceDirect

Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb

Fire resistance of concrete slabs reinforced with FRP bars. Part I: Experimental
investigations on the mechanical behavior
Emidio Nigro ⇑, Giuseppe Cefarelli, Antonio Bilotta, Gaetano Manfredi, Edoardo Cosenza
Department of Structural Engineering, University of Naples ‘‘Federico II’’, Via Claudio 21, Naples 80125, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Several building codes are now available for the design of concrete structures reinforced with FRP even if
Received 2 July 2010 no calculation model taking account of fire conditions has been suggested. Only the Canadian code (CAN/
Received in revised form 22 December 2010 CSA 806-02) provides a design procedure in fire situations based on critical temperature. Furthermore,
Accepted 27 February 2011
the literature provides some significant experimental results of failure tests performed on FRP–reinforced
Available online 23 April 2011
concrete members working in flexure that were exposed to conventional fire conditions.
Tests recently performed by the authors allowed the behavior of six concrete slabs reinforced with
Keywords:
GFRP bars exposed to fire action to be evaluated: four slabs were tested under typical design loads in fire
A. Glass fibres
B. Adhesion
situations (40% and 60% of design bending moment resistance at normal temperature) and two unloaded
B. High-temperature properties slabs were tested after the cooling phase in order to evaluate their residual resistance.
D. Mechanical testing In the present paper the experimental programme is extensively reported, giving detailed information
on tests to highlight the practical significance of the experimental research. The results of investigations
are discussed with particular reference to the structural behavior of concrete members once the glass
transition temperature in the bars is attained and resin softening reduces adhesion at the FRP-to-concrete
interface. Results are shown with the aim of providing suggestions for updating design codes. In a com-
panion paper the pattern of temperatures recorded during the tests by means of thermocouples applied
both on the surface of bars and in the concrete are presented in depth and compared with results of
numerical simulations.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction found that the fire resistance of FRP–reinforced concrete slabs


mainly depends on (a) the critical temperature of FRP reinforce-
Fiber–reinforced polymer (FRP) materials have several impor- ment, (b) the thickness of the concrete cover, and (c) the type of
tant characteristics, such as high strength-to-weight ratios and aggregate in the concrete mix. They then provided different design
resistance to corrosion, which are advantageous in the construc- abacuses which allow evaluation of the minimum concrete cover
tion field. Recent progress in research and technology of FRPs have required to maintain the bars’ temperatures within acceptable lim-
led to reduced material costs and increased confidence in the use its depending on the critical temperatures of the FRP reinforce-
of polymers for a variety of civil engineering applications, as testi- ment. Nevertheless, the critical temperature of FRP bars remains
fied by many examples worldwide. unknown for most currently available FRP-reinforcing products.
Recent studies carried out at by Keller et al. [1,2] and Correia Consequently, employment of FRP–reinforced concrete (FRP–RC)
et al. [3] on the fire response of GFRP pultruded profiles showed is limited mainly to applications where fire resistance aspects are
that FRP profiles can be used also in fire situation. On the other not particularly relevant. Thus performance of these materials in
hand, several building codes (CAN/CSA 806-02 [4]; ACI 440.1R-04 fire situations must be evaluated in order to improve confidence
[5]; CNR-DT203 [6]) are now available for the design of concrete in the use of FRP–RC members in multi-story buildings, parking
structures reinforced with fiber reinforced polymer (FRP) bars in garages, and industrial structures.
place of traditional steel reinforcement, even if few provisions On this point, the literature provides a broad state of the art on
and no calculation model have been suggested that take account the behavior in fire events of concrete structures reinforced or
of fire conditions. Only the Canadian code (CAN/CSA 806-02 [1]) strengthened with FRPs (Bisby et al. [8]) and some results of failure
provides a design procedure in fire situations based on the results tests performed on FRP–reinforced concrete members working in
of a parametric study conducted by Kodur and Baingo [7]. They flexure that were exposed to conventional fire conditions (Blontrock
et al. [9]; Sakashita [10]; Saafi [11]; Tanano et al. [12]; Weber [13]).
⇑ Corresponding author. Kodur et al. [14] showed that higher fire resistance for FRP–RC
E-mail address: emidio.nigro@unina.it (E. Nigro). slabs can be obtained by using larger concrete cover thickness and

1359-8368/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesb.2011.02.025
1740 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

through the use of carbonate aggregate concrete. Moreover, they S6 were 4000 mm long, 1250 mm wide and 180 mm thick; the
pointed out that it is necessary to consider the effects of two concrete cover values were 51 mm (see Fig. 2). The test
important factors on the fire endurance of RC slabs, namely the ap- programme, namely the main geometrical characteristics and
plied load and the reinforcement bond degradation. the spacing of the reinforcement of specimens are synthesized in
Weber [13] pointed out the importance of distinguishing be- Table 1.
tween two different limit temperatures: one related to the deteri- The concrete was identical for all slabs and characterized by cal-
oration of the bond strength and one related to the decrease of the careous aggregate (C35/45 according to EC2 [17]). In particular, for
tensile strength in the bars. Indeed, bond test results showed that 1 m3 of concrete the following components were used: 810 kg
bond strength between FRP and concrete decreases substantially (306.2 l) of sand 0/6, 525 kg (196.3 l) of gravel 6/10, 600 kg
when the glass transition temperature (Tg ffi 180 °C) was attained. (224.4 l) of gravel 10/16, 315 kg (104.3 l) of cement Cem I 52.5,
Moreover, the results of tensile tests at 400–500 °C showed a bar 155 kg (155 l) of water, 10 l of air entrainment and 3.5 l of super-
strength reduction ranging between 30% and 80%. Finally, a full- plasticiser. For reinforcement, E glass fibers and orthophthalic
scale test, performed according to DIN EN 1363 on a concrete slab polyester resin were used by the manufacturer providing the FRPs.
reinforced with the same GFRP bars, highlighted that failure after The surface of the bar is characterized by wraps of helical braid of
90 min of fire exposure was not attained due to the rupture of fibers and additional fine sand particles embedded evenly.
the bars. Indeed, the longitudinal bars were all lap spliced in the Compressive tests were performed on six specimens at 28 days
middle of the slab and failure was attained due to the loss of bond- after casting: average cube strength, Rcm, was 46.45 MPa; average
ing between bars and concrete in the midspan of the slab. cylindrical strength, fcm, was determined as fcm = 0.83 Rcm =
Abassi and Hogg [15] performed fire tests on concrete beams 38.55 MPa, according to the Italian code [19]). Mean tensile
reinforced with GFRP bars with a concrete cover of about 70 mm; strength and modulus of GFRP bars were provided by means of tests
the tests showed that beams can attain fire endurance for longer at room temperature. The glass transition temperature, Tg, was
than 90 min. Therefore a minimum concrete cover of 70 mm was evaluated by means of the DMA test on a specimen of 50 mm
recommended for design of GFRP RC beams under fire conditions. long  12.5 mm wide  2.92 mm thick. DMA measurements were
The value suggested is non-standard and particularly high with re- performed with a DMA Q800 manufactured by TA Instruments in
spect to both those required for corrosion control of steel–RC three point bending geometry operating at a frequency of 1 Hz
beams and those adopted for FRP–RC members. Such a suggestion and 15 lm strain amplitude. The temperature was varied from
does not seem to be suitable from a practical and economic point of 50 °C to 205 °C at a heating rate of 2 °C/min. The storage modulus,
view, even if encouraging results seem to depend on the large va- loss modulus and tan d curves, obtained from DMA test, are re-
lue of the concrete cover. Therefore, further experimental tests are ported in Fig. 3. The Tg value was assessed taking into account the
required to improve knowledge of the structural response of FRP– tan d maximum with reference to the first heating cycle. The
reinforced concrete members in fire conditions. mechanical properties of materials are reported in Table 3.
Based on these considerations, six concrete slabs reinforced The slabs were designed according to CNR-DT203 [6] that pro-
with glass fiber reinforced polymer (GFRP) bars characterized by vides design relationships to evaluate the bending moment resis-
different values of concrete cover and anchorage length were tance, MRd, at normal temperature. The design bending moment
tested in fire conditions by exposing them to heat in a furnace resistance MRd,1 ffi 65 kN m for slabs S1, S2, S4, S5 and
according to the time–temperature curve of ISO834 provided in MRd,2 ffi 46 kN m for slabs S3, S6 are reported in Table 2; as better
EN 1363-1 [16]. Tests were performed to evaluate their resistance discussed below, Lunexp represents the length of the ends of each
and deformability in fire situations by varying: (a) external loads in slab not directly exposed to fire action during the test. To avoid
the range of the service loads, (b) concrete cover in the range of forming bar splice anchorages in the span of the slab, a single GFRP
usual values and (c) bar anchorage lengths at the end of the con- bar was employed whose length was that of the slab minus 20 mm
crete members, namely in the zone not directly exposed to fire. (i.e. twice the 10 mm of concrete cover at each end of the slab).
It is worth noting that that in concrete structures zones of this Each slab, before casting, was instrumented with thermocou-
type, not directly exposed to fire, are often represented by mutual ples (insulated k type twisted cables with diameter equal to
connections between members. The experimental program, the 0.8 mm) and strain gauges in order to measure temperatures and
test results, and the analysis of the structural behavior of the slabs strains during the fire tests in the slabs: thermocouples and strain
during the fire tests are shown in the following sections. As will be gauge locations are shown in Fig. 1 for slabs S1, S2, S3 and Fig. 2 for
shown, the mechanical response of the slabs is often intercon- slabs S4, S5, S6; further details are reported in the companion pa-
nected to the temperatures attained by the GFRP bars during the per [20].
fire exposure. In order to avoid redundancy in the presentation of Six cubic specimens, 150 mm  150 mm  150 mm, were kept
the experimental results the temperatures recorded within the under the same ambient conditions as the slabs; the weight of such
slab, both in the concrete at different depths and near the bars, specimens was recorded the day of slab casting, a month after cast-
are widely reported in the companion paper [20]. ing and once a week in the second month after casting. After two
months the subsequent weight reduction, due to water evapora-
tion, became negligible. Nevertheless, the ratio between volume
2. Experimental program
and surfaces exposed to air for the slabs is lower than those related
to cubic specimens; thus in order to reduce concrete spalling due
The experimental program involved the design and fabrication
to evaporation of residual water in the slabs, fire tests were carried
of six full-scale concrete slabs (S1–S6) reinforced with GFRP bars.
out about 6 months after concrete casting. In the same period as
Simply-supported slabs were tested in a four-point bending
the fire tests, compressive tests were performed on the six speci-
scheme in fire conditions.
mens: the average cube strength was Rcm = 55.15 MPa (fcm = 0.83
Rcm = 45.77 MPa), about 20% greater than that measured at 28 days
2.1. Geometrical and mechanical characteristics after casting. Moreover, at room temperature a moisture content of
concrete, equal to 4% by weight, was directly measured by drying
Three slabs (S1, S2 and S3) were 3500 mm long, 1250 mm wide at 105° C for 24 h on concrete three cubic specimens, 150 mm 
and 180 mm thick. The concrete cover was 32 mm, as estimated by 150 mm  150 mm, that were kept under same ambient condi-
reference to the centroid of the GFRP bars (see Fig. 1). Slabs S4, S5, tions as the slabs until fire tests.
E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750 1741

Fig. 1. Geometrical details for slabs S1 and S2 (sizes in millimeters).

2.2. Test set-up ure; indeed after a deflection of about 500 mm, slabs rest on the
brick columns.
Tests were performed by exposing the slabs to heat in a furnace.
The furnace, designed specifically for fire testing, has a top open-
ing, 3000 mm wide by 3860 mm long, in which the specimens 2.3. Fire and structural loading
were placed for the tests. About one week prior to each fire test,
the slabs were taken to the furnace room for instrumentation. The slabs were tested concurrently by exposing them to heat
Three specimens (each specimen consisting of one slab, see from below in such a way that the average temperature of the fur-
Fig. 4) were placed centrally at the roof level of the furnace side nace followed automatically, as closely as possible, the standard
by side in order to test them concurrently. The gap between two time–temperature curve of ISO 834 (EN 1363-1 [16]) given by
adjacent slabs and between the slabs and the furnace walls were TISO = 20 + 345 log10(8t + 1)where TISO is the temperature in degrees
carefully packed with insulation (rock wool) to reduce heat loss Celsius and t is the time in minutes. The furnace temperature was
and allow the beam to deflect freely under load. In particular, in or- controlled during the tests by taking the average temperature of
der to avoid slabs were exposed to fire on the lateral sides after eight shielded thermocouples located 300 mm below the exposed
their inflection, the gap of 3 cm between slabs a panel coated with surface of the concrete slabs.
rock wool was inserted (Fig. 5a). The panel is shaped so that it can The load values applied during the fire tests were evaluated
rest on the top edge of the perimeter wall of the oven (Fig. 5b) dur- based on the bending moment resistance values at normal temper-
ing the test. ature. The European code EN 1991-1-2 [21] suggests, for analysis in
As the span between supports was 3200 mm, the slabs fire conditions, that the effects of actions, Ed,fi, may be obtained
3500 mm long (S1, S2, S3) were external to furnace for a length from a structural analysis for normal temperature design as
of 150 mm, whereas the slabs 4000 mm long (S4, S5, S6) were Ed,fi = gfiEd. Ed is the design value of the corresponding force or
external to furnace for a length of 400 mm. A strip of about moment for normal temperature design for a fundamental combi-
100 mm of rock wool was used to protect the steel supports. There- nation of actions (see EN 1990 [18]). gfi is the reduction factor for
fore the ends of each slab were not directly exposed to fire action the design load level in fire situations. gfi may be considered
for a length, Lunexp, of about 250 mm for slabs S1, S2, S3, and mainly dependent on the ratio between the characteristic values
500 mm for slabs S4 S5, S6 (see Fig. 6). Inside the furnace, of the principal variable load and the permanent loads, namely
500 mm below each slab, refractory brick columns were finally Qk/Gk, as well as on the combination factors wfi (see EN1991-1-2
placed to prevent the slabs being completely destroyed after fail- [21]); its value generally varies between 0.3 and 0.7.
1742 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

Fig. 2. Geometrical details for slabs S4 and S5 (sizes in millimeters).

Table 1
Fire test main parameters for FRP–reinforced concrete slabs.

Slab Length (mm) Width (mm) Thickness (mm) Cover Thickness (mm) Bottom rebars (diameter/spacing) (mm) Top rebars (diameter/spacing) (mm)
Longitudinal Transverse Longitudinal Transverse
S1 3500 1250 180 32 U12/150 U12/200 U12/200 U12/200
S2
S3 U12/225
S4 4000 1250 180 51 U12/125 U12/200 U12/200 U12/200
S5
S6 U12/200

Thus slabs S2, S5 and S3, S6 were loaded with a predefined ser- tively for slab S4, S5, S6 and for slabs S1, S2, S3. If the slab did not
vice load providing a bending moment that corresponds to about fail in this range of time the load would be increased up to slab
40% and 60%, respectively, of the ultimate bending moment resis- failure. For these reason the test consisted of two stages:
tance of the slab in normal temperature design. For each slab such
values were obtained by adding to the slab weight two external – Stage 1: constant load until 180 min and 120 min of fire expo-
forces of 17.5 kN applied in two sections at a distance of 1.25 m sure time respectively for slab S4, S5, S6 and for slabs S1, S2, S3.
from the support sections and symmetric relative to the midspan – Stage 2: load increasing until slab failure.
section (see Fig. 5). The external total load (35 kN) was applied
using a load transmission steel beam (with a weight of 5 kN) Slabs S1 and S4 were not loaded during the fire exposure: in or-
loaded by means of two hydraulic jacks (each loading 13.5 kN). der to evaluate their residual resistance after fire exposure, the
The load transmission beam, placed over the tested slab, trans- load was applied about 24 h after the fire exposure had stopped,
ferred the load to the extrados of the member through two steel when the temperature of the bars in the slab attained an average
profiles, 10 cm width  125 cm long (each one weighing 1.5 kN). of about 45 °C. The purpose is to assess the permanent damage that
Slabs S2, S3 and S5, S6 were loaded independently of each other GFRP bars undergo due to high temperatures, assuming bar dam-
about 30 min prior to fire exposure in order to ensure stabilization age does not stem from their stress level during fire exposure.
of deflections and deformations. The load was expected to be kept The deflection was measured on the extrados of the slabs at
constant up to a fire exposure time of 180 min and 120 min respec- mid-span. A draw wire displacement sensor was used, to avoid
E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750 1743

Fig. 3. DMA results (first scan).

Table 2
3. Experimental results
FRP–RC slabs – geometrical properties.

Slabs Distance between Unexposed length Design bending 3.1. Deflection behavior
supports L (mm) Lunexp (mm) resistance MRd (kN m)
S1 3200 250 65 The displacements recorded at the midspan of all slabs are plot-
S2 ted against time in Fig. 7a and b, respectively. After about 60 min of
S3 46
fire exposure, slab S3, stressed to about 60% of MRd,2, once the
S4 3200 500 65 deflection attained a value fS3,60  70 mm, suddenly failed (the
S5
S6 46
abbreviation fX,Y stands for the deflection of slab X at time Y). Slab
S2, stressed to about 40% of MRd,1, attained a deflection value
fS2,120  90 mm at about 120 min of fire exposure. Therefore the
loads, at such instant, on the slab S2 were increased: slab S2 failed
when the load increased up to 50% of MRd,1. Unloaded slab S1 after
Table 3
120 min attained a deflection fS1,120  65 mm due to its own
Concrete and reinforcing bars properties.
weight and thermal strains; its residual resistance evaluated 24 h
Parameter Values after fire exposure was about 55% of MRd,1.
Concrete (35/45) Mean compressive cube strength 46.45 Slabs S5 and S6, stressed to about 40% of MRd,1 and 60% of MRd,2,
(28 days) (MPa) respectively, exceeded 180 min of fire exposure and attained
Mean compressive cube strength 55.15
deflection values fS5,190  80 mm and fS6,190  100 mm after fire
(6 months) (MPa)
exposure of t  190 min. Therefore, at such instant, the loads on
GFRP bars Mean tensile strength (MPa) 1000
the slabs were increased and the tests ended when the bending
Mean tensile modulus (GPa) 50
Glass transition temperature (°C) 100 moment attained 85% of MRd,1 and 100% of MRd,2 for slabs S5 and
Diameter (mm) 12* S6 respectively. Unloaded slab S4 at 190 min attained a deflection
Glass fiber mass fraction (%) 70* fS4,190  70 mm; its residual resistance was about 100% of MRd,1.
Density (g/cm3) 1.9* Figs. 8 and 9 show slabs S5, S6 and S4, respectively after failure.
*
Provided by manufacturer. Qualitative observations for all slabs were made from the fur-
nace after the tests were completed. Only in a little zone very lim-
ited spalling was observed for any of the slabs. In particular, the
the fire affect the sensor measurements. According to EN1363-1 spalling phenomena were characterized by a limited concrete peel-
(2001) it was decided that the test should be terminated when ing of about 3.0–4.0 mm.
both following conditions would occur: In Table 4 for the first stage of the test (Stage 1, with constant
load level), load level during fire tests, gfi, and fire resistance time
– The deformation rate of the specimen exceeded L2/(9000d) for all slabs are summarized. Moreover, the failure mode and the
equals to 7.7 mm/min for slabs S1, S2, S3 and 8.8 mm/min for load level at failure gfail for the Stage 2 of the tests are reported.
slabs S4, S5, S6. Clearly, referring to four slab tests under service load, those
– The maximum deflection value of L2/(400d), equals to 173 mm characterized by lower concrete cover and lengths of zones not di-
for slabs S1, S2, S3 and 198 mm for slabs S4, S5, S6, was rectly exposed to fire (i.e. c = 32 mm and Lunexp = 250 mm, respec-
attained. tively) showed a fire endurance of 60 and 120 min depending on
the load level gfi. By contrast, two slabs with larger values of both
where L is the distance between the supports and d is the distance c and Lunexp (i.e. 51 mm and 500 mm, respectively) attained a fire
between the center of the FRP bar from the top edge of the slab. endurance exceeding 180 min.
1744 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

Fig. 4. Test set-up – slabs on furnace (sizes in millimeters).

Fig. 5. Test set-up – photos.

Moreover, comparison between the load levels attained at similar result is shown by the values of failure loads attained on
failure (gfail) on slabs S1 and S2 clearly shows that the residual slabs S4 and S5 exposed to fire for about 180 min. Therefore the
resistance of slab S1, loaded after cooling phase, is substantially damage that GFRP bars undergo due to high temperatures seem
equal to that of slabs S2, loaded after 120 min of fire exposure, to be almost completely permanent and, cooling does not seem
regardless of the load level under fire conditions (see Table 4). A to allow resistance recovery.
E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750 1745

Fig. 6. Test set-up (sizes in millimeters).

120 120
Deflection (mm)

Deflection (mm)

Load
110 Load 110 increasing
100 increasing 100
90 90
80 80
70 70
60 60
50 50
S3 S6
40 S2 40 S5
30 S1 30 S4
20 20
10 10
Time (min) Time (min)
0 0
0 30 60 90 120 150 180 210 240 0 30 60 90 120 150 180 210 240

(a) Slabs S1,S2,S3 (L=3500mm, c =32mm) (b) Slabs S4,S5,S6 (L = 4000mm, c = 51mm)
Fig. 7. Displacements vs. time relationships during the fire test.

All time–displacement curves of Fig. 7 present a significant depth analysis of temperatures recorded in the slabs (see compan-
change of slope at about 30 min that corresponds to the change ion paper [20]) has shown that during fire tests in the zone ex-
of slope characterizing the standard time–temperature curve of posed to fire the average temperature of FRP bars attained the
ISO 834 (EN 1363-1 [16]). This suggests that the trend of time–dis- glass transition temperature value (i.e. 100 °C) after not much time
placement curves depends mainly on the temperatures. As well, in- (about 15 min and 20 min, respectively for slabs S1, S2, S3 and for
1746 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

lower than Tg = 100 °C. Thus, during fire exposure the structural re-
sponse depended mainly on the reinforcement anchorage at the
fire-unexposed end of the member.

S6
3.2. Failure modes

S5 Based on this observation as regards the behavior of slabs S1, S2


and S3, in Fig. 7a it may be noted that the time–displacement
curves, between 30 min and 60 min, show an equal slope. Between
60 min and 120 min the slope of the time–displacement curve re-
S4 lated to the loaded slab S2 is greater than that related to the un-
loaded slab S1. Since the external load is constant, this difference
is probably due to a bond reduction of the anchorage. By contrast,
with reference to slabs S4, S5 and S6 (see Fig. 7b) between 30 min
and 180 min, i.e. during the test phase in which the external load
did not increase, three time–displacement curves show the same
Fig. 8. Slabs S5 and S6 after failure. slope. This result suggests that the bond in the anchorage zone
was not particularly affected by the low increase of temperature.
In Fig. 10 slab S5 is viewed from the furnace after the test. The spac-
ing between cracks is constant and equal to about 200 mm, corre-
sponding to the spacing between the transverse reinforcement
bars. Near the midspan, through the larger crack, the GFRP bars
may be examined. A detailed photo (Fig. 11) clearly shows the glass
fibers of the longitudinal bars broken and completely without resin.
Further investigations were performed by breaking the ends of the
slabs with a pick after tests in order to examine the bond behavior
of bars at the unexposed zone of slabs in which temperatures did
not attain the glass transition temperature value.
As regards slabs S1, S2 and S3, the position of the longitudinal
S4 bars in the slabs could easily be identified at vertical cracks on
the lateral faces of the slabs, as shown by black lines in Fig. 12.
Breaking with a pick the concrete under such lines allowed the
presence of bars to be checked (see Fig. 13). Almost every bar
pulled out at one end. The bars that did not show slip signs failed
at midspan due to fiber tensile rupture. In some cases the covering
of the external bar remained bonded to the concrete and fibers
slipped in the slab. The number and position of GFRP bars that
Fig. 9. Slab S4 after failure. pulled out due to bond loss are reported in Fig. 14.
A more extensive investigation, performed by overturning slabs
and breaking the whole ends of slabs with a pick in the fire-unex-
slabs S4, S5, S6). Therefore, already in this initial test phase, the re- posed zone (see Fig. 15), confirmed the results of preliminary
sin softening, due to temperatures attained in the bars, consider- investigations and evidenced the presence of the polymer resin
ably reduced adhesion at the FRP–concrete interface in the fire- matrix around the glass fibers, even if the resin appeared to have
exposed zone of the slab. By contrast, the temperature recorded partly deteriorated along the bar anchorage length.
at the unexposed zone of slabs remained significantly low during Also for slabs S4, S5, S6 it was observed that part of the FRP bar
the whole test and achieved a temperature almost equal to the anchorage underwent the temperature action. Nevertheless, a
glass transition temperature just after 180 min of fire exposure. length of anchorage in the not directly exposed zone of the slab
It is worth note that the thermocouples in this unexposed zone (namely Lunexp ffi 500 mm larger than Lunexp ffi 250 mm related to
were at 400 mm and 300 mm from the end of the slabs (see com- slabs S1, S2, S3) allowed bar pull-out to be avoided.
panion paper [20]). Hence, up to 180 min in the last 400 mm of the Thus slabs S1, S2 and S3 failed when bars pulled out in the
unexposed zone of the slabs S4, S5 and S6, the temperature was anchorage zone without the resistance of glass fibers in the fire-

Table 4
Load level during fire tests and fire resistance time (Stage 1); load level increasing at failure (Stage 2); failure mode.

Slab Cover thickness c Un-exposed length Lunexp Stage 1 Stage 2


(mm) (mm)
Load level during fire test gfi Fire resistance time Te Load level at failure gfail Failure
(%) (min) (%) mode
S1 32 250 10 180 55 Pull out
S2 40 120 50
S3 60 60 –
S4 51 500 10 180 100 Bar rupture
S5 40 180 85
S6 60 180 100
E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750 1747

Fig. 10. Slab S5 viewed from the furnace after testing.

Fig. 11. Longitudinal and transverse reinforcement after failure.

exposed zone was attained. On the contrary, slabs S4, S5 and S6 4. Remarks and discussions
failed when glass fibers attained their tensile strength at about
500 °C, lower than tensile strength at normal temperature. It is well known that the variability in the main thermal and
mechanical properties of FRP and concrete materials with temper-
3.3. Fire resistance ature (i.e. thermal conductivity, specific heat, stiffness and
strength) has to be taken into account in order to simulate the
Based on such observations it is possible to look again at the test mechanical behavior of GFRP–reinforced concrete slabs in fire
results reported in Table 4 in terms of fire endurance and resis- situations.
tance. The fire exposure time of slab S6 is significantly higher than In addition, experiments have shown that failure of the concrete
S3 due to both the larger concrete cover (51 mm vs. 32 mm) and slabs can be attained by the rupture of fibers in the middle of the
larger length of the FRP bar anchorage in the unexposed zone of member if continuous reinforcement from side to side of the con-
the slab (500 mm vs. 250 mm). Larger cover allowed the decrease crete element is used and a large zone not directly exposed to fire
in bar strength to be mitigated and larger anchorage lengths is guaranteed (i.e. 500 mm). In this case, failure is due to very high
avoided bar pull-out. temperatures achieved in GFRP bars (i.e. average temperature of
Furthermore, the fire resistance time of S2 slab is higher than about 460 °C, much higher than Tg) that particularly affect the
that of S3, due to lower stresses in the bars which require a lower strength of glass fibers in the exposed zone while the anchoring
anchorage length under fire conditions. Indeed, as previously sta- zone allows pull-out of bars to be avoided. Fire endurance depends
ted, slab S2 was stressed to about 40% of MRd,1 and slab S3 stressed on fiber strength at high temperatures and hence on the concrete
to about 60% of MRd,2. Finally, it is worth noting that slab S2 failed cover that affects the temperatures in the bars during fire as well.
when the load increased to about 50% of MRd,1 that is a percentage Therefore numerical simulation can neglect the modeling of the
lower than slab S3, but failure occurred at 120 min, i.e. when the anchorage area not directly exposed to fire and can mainly be
anchorage had undergone fire action for a further 60 min with re- based on defining the mechanical properties of bars at different
spect to slab S3. temperatures. It’s important to note that the FRP reinforcement
1748 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

Fig. 12. Lateral face of Slab S2 before picking.

Fig. 13. Lateral face of Slab S1 after picking.

will become unbonded even if the cold anchorage is maintained, up to the top edge of the slab that is barely warm when the fire
and this will lead to larger and fewer cracks, which may expose comes from below. In such a way the width of the beam required
the bar directly to the fire and initiate failure. to protect the end of the slab can be noticeably reduced. The
On the other hand, if the zone not directly exposed to fire is production process allowing the bar to be bent is easily imple-
shorter (i.e. 250 mm) tests showed that fire endurance of members mented by FRP bars manufacturers owing technologically
chiefly depends on the slippage of the bars due to a reduction in advanced systems.
bonding at the FRP-to-concrete interface that is also attained at
not particularly high temperatures (i.e. about Tg). Therefore numer-
ical simulation has to take account of the bond behavior of bars 5. Conclusions
embedded in the concrete versus temperature. For this purpose a
bond theoretical model calibrated in normal conditions could be Recently performed tests allowed to assess the resistance and
refined in the presence of high temperatures and used to assess deformability of six GFRP bar reinforced slabs exposed to fire ac-
the safety against the pull-out of bars. Note that in both cases tion according to the ISO 834 standard time–temperature curve.
(anchorage zone of 500 mm and 250 mm) the construction detail Three slabs with concrete cover values of 51 mm and anchorage
with the overlap of bars in the area directly exposed to fire (namely length values in the slab unexposed zone of about 500 mm,
in the midspan of the member) cannot be considered because the showed a better structural behavior in the event of fire than three
temperature attains such high values that the loss of adhesion and slabs with concrete cover values of 31 mm and anchorage lengths
then structural failure would occur in a short time. of about 250 mm.
Moreover it is worth noting that the failure for a reinforced Thus experimental test results confirmed that the concrete cov-
concrete element depends on the combination of stress, tempera- er and anchorage length of the FRP bars in the zone not directly ex-
ture, and anchorage length. Such parameters must be evaluated posed to fire are fundamental parameters for the fire resistance of
for defensible design of FRP reinforced concrete members in fire concrete members:
situation.
Finally, referring to the specific case of concrete slabs, it should – the concrete cover is particularly important for the protection
be noted that they are generally linked to concrete beams that pro- afforded to FRP bars, due to its low thermal conductivity;
tect a portion of slab for lengths equal to the width of the beams – the anchorage length at the end of the members ensures slab
themselves (usually in the range between 250 and 500 mm). More- resistance once in the fire-exposed zone of slab the glass transi-
over, it seems worth considering the possibility of anchoring the tion temperature is achieved and resin softening reduces the
bottom reinforcement at the end of the slab by bending the bars adhesion at the FRP–concrete interface.
E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750 1749

Bonded Pulled out

S1
100mm
90mm
30mm

60mm

90mm
85mm

S2

45mm
55mm
60mm
55mm
55mm

60mm

S3

60mm

60mm

55mm

55mm

30mm

Fig. 14. Pulled-out GFRP bars: slip of each bar at the ends of the slabs (in mm).

Fig. 15. Slab S2: loss of adhesion of bar anchorage.

In particular, for the specific FRP–RC slabs tested herein, it can – the use of continuous reinforcements from side to side of the
be stated that: concrete element is advised;
– an anchorage area for FRP bars at the end of the slab not directly
– usual values of concrete cover (i.e. between 30 and 50 mm) can exposed to fire for a length of 500 mm avoids pull-out of bars and
be adopted; allows failure due to the rupture of the fibers to be obtained.
1750 E. Nigro et al. / Composites: Part B 42 (2011) 1739–1750

Acknowledgments [9] Blontrock H, Taerwe L, Matthys S. Properties of fiber reinforced plastics at


elevated temperatures with regard to fire resistance of reinforced concrete
members. In: Proceedings of IV international symposium on fiber reinforced
The present paper was developed within the Research Project concrete structures, Baltimore; 1999.
MIUR-PRIN 2006 (‘‘Effects of high temperatures on the performances [10] Sakashita M. Deflection of continuous fiber reinforced concrete beams
subjected to loaded heating. In: Proceedings of non-metallic (FRP)
of concrete slabs reinforced with FRP bars or grids‘‘). The authors
reinforcement for concrete structures, vol. 58. Japan Concrete Institute;
appreciated the contributions of COOPSETTE Soc. Coop – Castel- 1997. p. 51–8.
nuovo (RE) and SIREG SpA – Arcore (Mi). They would also like to [11] Saafi M. Design of FRP reinforced concrete beams under fire conditions. FRP
composites in civil engineering, vol. II. Alabama, USA: Elsevier; 2001. p. 1235–
thank the CSI SpA – Bollate (MI) laboratory for assistance in carry-
44.
ing out the fire tests described briefly herein. [12] Tanano H, Masuda Y, Tomosawa F. Characteristics and performances
evaluation methods of continuous fiber bars – state-of-the-art studies on
fire properties and durability of continuous fiber reinforced concrete in Japan.
References In: Proceedings of IV international symposium on fiber reinforced concrete
structures, Baltimore; 1999.
[13] Weber A. Fire-resistance tests on composite rebars. In: Proceedings of
[1] Keller T, Zhou A, Tracy C, Hugi E, Schnewlin P. Experimental study on the CICE2008, Zurich, Switzerland; 2008.
concept of liquid cooling for improving fire resistance of FRP structures for [14] Kodur VKR, Bisby LA, Foo S. Thermal behaviour of fire-exposed concrete slabs
construction. Compos Part A: Appl Sci Manuf 2005;36(11):1569–80. reinforced with fibre reinforced polymer bars. ACI Struct J
[2] Keller T, Tracy C, Hugi E. Fire endurance of loaded and liquid-cooled GFRP slabs 2005;102(6):799–808.
for construction. Composites: Part A 2006;37(7):1055–67. [15] Abbasi A, Hogg PJ. Fire testing of concrete beams with fibre reinforced plastic
[3] Correia JR, Branco FA, Ferreira JG, Bai Y, Keller T. Fire protection systems for rebar. Composites Part A 2006;37:1142–50. Elsevier.
floors of buildings made of GFRP pultruded profiles. Part 1: experimental [16] European committee for standardization. EN 1363-1:2001. Fire resistance
investigations. Composites Part B 2010;41(8):617–29. tests – general requirements; 2001.
[4] CSA. Design and construction of building components with fiber reinforced [17] European committee for standardization. EN 1992-1-2. Eurocode 2. Design of
polymers. CAN/CSA S806-02, Canadian Standards Association, Ottawa, ON; concrete structures – Part 1–2: general rules – structural fire design; March
2002. 210pp. 2004.
[5] ACI. Guide for the design and construction of concrete reinforced with FRP [18] European committee for standardization. EN1990. Eurocode 0. Basis of
bars. ACI 440.1R-04, American Concrete Institute, Farmington Hills, MI; 2003. structural design; April 2002.
42pp. [19] Italian code. Norme Tecniche per le Costruzioni (in Italian). Supplemento
[6] CNR. Guide for the design and constructions of concrete structures reinforced Ordinario della Gazzetta Ufficiale della Repubblica Italiana, 4th February 2008,
with Fiber Reinforced Polymer bars. CNR-DT203/2006, Italian National no. 29.
Research Council; 2006. [20] Nigro E, Cefarelli G, Bilotta A, Manfredi G, Cosenza E. Fire resistance of concrete
[7] Kodur VKR, Baingo D. Fire resistance of FRP–reinforced concrete slabs. Internal slabs reinforced with FRP bars. Part II: experimental results and numerical
Report No. 178. Institute for Research in Construction, National Research simulations on the thermal field. Composites – Part B 2011;42:1751–63.
Council Canada; 1998. [21] European committee for standardization. EN 1991-1-2. Eurocode 1. Actions on
[8] Bisby LA, Green MF, Kodur VKR. Response to fire of concrete structures that structures – Part 1–2: general actions – actions on structures exposed to fire;
incorporate FRP. Prog Struct Eng Mater 2005;7(3):136–49. John Wiley & Sons November 2002.
Ltd.

You might also like