Composites: Part B 40 (2009) 85–93

Contents lists available at ScienceDirect

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

Effect of elevated service temperature on bond between FRP EBR systems

and concrete
Marianovella Leone a,*, Stijn Matthys b,1, Maria Antonietta Aiello a,2
a
Department of Engineering for Innovation, University of Salento, Via per Arnesano, 73100 Lecce, Italy
b
Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University, Technologiepark-Zwijnaarde 904, B-9052 Gent, Belgium

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

Article history: The method to increase the ultimate capacity of reinforced concrete elements, by means of Fiber Rein-
Received 18 February 2008 forced Polymers bonded on their substrate with thermosetting resins, represents a technique useable
Received in revised form 12 June 2008 worldwide. The bonding materials used, generally epoxy resins, are sensitive to a temperature increase.
Accepted 22 June 2008
In fact, the curing process of epoxies leads to low glass transition temperature (Tg) values, that may cause
Available online 30 August 2008
a relevant decay of the mechanical properties of the adhesive, even under service thermal conditions. The
reduction can influence the bond performance, compromising the effectiveness of the reinforcing tech-
Keywords:
nique. In this work the interface behaviour FRP–concrete at elevated service temperatures is analyzed.
A. Laminate
B. Debonding
The experimental results show a relevant influence of the temperature on bond performance, in terms
B. Thermal properties of type of failure of the interface, effective bond length and bond strength.
Concrete Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction tures, mostly in relation to the bond performance. In fact, the

effectiveness of the externally reinforced FRP sheets significantly
The interest in fibre reinforced polymer (FRP) in construction depends on the properties of the FRP–adhesive and adhesive–con-
has considerably increased, and especially the application of FRP crete interfaces. In consideration of this, the fib Bulletin 14 [4]
as externally bonded reinforcement (FRP EBR) has become more suggests that the glass transition temperature of the adhesive used
and more established. The use of FRP EBR has been adopted for bonding the FRP to the concrete should equal 20 °C in excess of
world-wide as a very attractive technique for structural strength- the maximum ambient temperature at normal service conditions,
ening and rehabilitation [1]. The static behaviour of FRP EBR con- and should be at least 45 °C.
crete structures is extensively studied in the literature, while the Knowledge about the decrease of the properties of the FRP–con-
influence of the service conditions (temperature, water, humidity, crete interface at elevated temperatures is currently very limited.
etc) on the structural performance represents an open topic that In [5] the influence of elevated service temperature (20, 30, 60,
needs to be analysed in more detail. 90 and 120 °C) on the bond between externally bonded steel rein-
As is well-known the adhesive used to apply the FRP reinforce- forcement and concrete was studied. The authors, performing dou-
ment to concrete is generally a thermosetting polymer, like epoxy ble face shear tests, observed a relevant decrease of the failure load
or polyester resin. Based on the available literature [2,3], the epoxy increasing the temperature. Referring to the specimens tested at
resin used to glue the FRP composite on the concrete and which room temperature the reduction of the bond strength was in the
also forms the matrix of hand layup FRP sheets is largely influenced range 0.9–31.8% at 30 °C, 48.6–55.3% at 60%; 72.2–70.9% at 90 °C
by the service temperature, in relation to its glass transition tem- and 91.1–92.7% for the specimens tested at 120 °C.
perature (Tg). Above the Tg value the mechanical properties of A different trend of the failure load was obtained in [6], per-
the resin will significantly drop and the material will change from forming double face shear tests on concrete elements reinforced
a hard and glass-like material to a more rubber-like material. The with externally bonded CFRP laminates at different temperature
reduction of the mechanical properties of the adhesive at elevated values (20, 40, 55 and 70 °C). The authors, considering the test at
temperatures is of relevant importance for the strengthened struc- 20 °C as a reference, noted that the ultimate load increased for
the test at 40 °C and 55 °C (41% and 24%, respectively) while it
decreased at 70 °C (19%). Klamert et al. [7] carried out double face
* Corresponding author. Tel.: +39 0832 297266. shear tests on concrete specimens reinforced with CFRP laminates
E-mail addresses: [email protected] (M. Leone), Stijn.Matthys@
UGent.be (S. Matthys), [email protected] (M.A. Aiello).
varying the service temperature (20, 50, 65 and 75 °C). The trend of
1
Tel.: +32 09 2645520. the recorded failure load is in agreement with the experimental re-
2
Tel.: +39 0832 297248. sults reported in [6]. In fact, passing from 20 to 50 °C, the failure

1359-8368/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesb.2008.06.004
86 M. Leone et al. / Composites: Part B 40 (2009) 85–93

load increased around 5% while the comparison with the speci- non-instrumented zones, extra clamp anchorages are used (left
mens tested at 75 °C showed a decrease of the bond strength of hand side in Fig. 1).
around 38%. For the tests at elevated temperatures an electrical hollow fur-
Finally, similar results in terms of the type of failure were ob- nace was used. The oven was placed around the specimen in the
served in [5] and [7]: a cohesion failure in the concrete occurred zone without clamps. All gaps between the furnace and the speci-
at moderate temperatures while at high temperatures an adhesive men were filled with mineral wool. The temperature in the furnace
failure took place at the adhesive–concrete interface. was controlled by a thermocouple (type K) measuring the air tem-
The experimental program presented in this paper mainly aims to perature inside the furnace. The specimens aligned in a universal
determine what effect elevated service temperatures have on the testing machine (2500 kN servo – hydraulic testing machine) were
bond performance between FRP reinforcement and concrete mem- subjected to a tensile force with a displacement rate of 0,1 mm/min.
bers. To that extent specimens have been tested under double-face In the Fig. 2 a view of the test set-up with and without the oven
pure shear test and at different test temperatures: 20, 50, 65 and is given.
80 °C. A comparison between un-conditioned (20 °C) and condi-
tioned specimens was performed to evaluate the degradation caused 2.2. Test parameters
by the thermal exposure. Three types of FRP reinforcement were
used in this study: CFRP (Carbon Fiber Reinforce Polymer) sheet The influence of the type of reinforcements (factory made ver-
and laminate and GFRP (Glass Fiber Reinforced Polymer) sheet. sus wet lay-up sheet), as well as the type of fibers was evaluated.
The experimental results are presented and discussed in terms of The externally bonded reinforcement used comprised CFRP (Car-
strains within the reinforcement during the load history, bond bon Fiber Reinforced Polymer) sheet and laminate, and GFRP (Glass
stress–slip law, transfer length, bond strength and kind of failure. Fiber Reinforced Polymer) sheet. The material properties are given
in the materials section. Three different test temperatures were
2. Outline of the experiments used: 50, 65 and 80 °C. The temperature level was chosen in rela-
tion to the glass transition temperature (Tg) of the utilized epoxy
The performed experimental program aimed to determine what resin, experimentally evaluated as reported in the materials sec-
effect elevated service temperatures have on the bond perfor- tion. Given a Tg value of 55 °C, the test temperatures of 50, 65
mance of FRP reinforced concrete members. The investigation and 80 °C represent 10%, +18% and +45% of the glass transition
comprised 18 double face shear tests, performed varying the tem- temperature, respectively. The tests conducted at elevated temper-
perature value. A comparison between un-conditioned (20 °C) and atures were compared with those at room temperature (20 °C).
conditioned specimens was made to evaluate the degradation An overview of the experimental program is given in Table 1.
caused from the thermal exposure. The specimens are listed using the following code: the first letter
C or G indicates carbon fibers or glass fibers, respectively; the sec-
2.1. Test set-up ond letter, either L or S, indicates the type of reinforcement: factory
made or wet lay-up sheet; the third notation indicated the test
The test set-up is given in Fig. 1. The concrete specimens temperature, 20, 50, 65 and 80 °C, respectively. The last letter indi-
(150 150 800 mm) consist of two concrete prisms (length cates the test sequence of the two similar specimens tested for
400 mm). A thin metal plate separates the two concrete blocks, each analyzed parameter (Table 1).
the height of this plate is 15 mm less than the height of the prisms.
Two steel bars, with a diameter of 16 mm, are embedded in each 2.3. Material properties
prism. As is also shown in Fig. 1 these internal steel bars do not
connect the two concrete parts, which means that the two blocks The concrete composition and material properties of the FRP
are only joined through the externally bonded FRP reinforcement. reinforcements, fresh and hardened concrete, are reported in the
Special care was taken during the installation of the formwork to following.
position the internal steel reinforcement perfectly aligned, so that Three different concrete batches, with the same concrete com-
their working lines in the two separated blocks coincide. The position, were used. The batches are indicated as T_I, T_II and
length of the protruding part of the steel rebars was chosen in or-
der to allow clamping in the tensile testing machine. The speci-
mens are characterized by four FRP to concrete bond zones with
a length of 300 mm, of which two zones are instrumented (right
hand site in Fig. 1). An un-bonded zone was also left in the central
part of the specimens as is better specified in the specimens prep-
aration and measurements section. To prevent bond failure in the

STEEL CLAMPS
FRP SHEET
100 300
STEEL REBARS

80
460

STEEL CLAMPS

UN-BONDED ZONE
FRP SHEET
STEEL REBARS
150
100

400 400

Fig. 1. Test set-up. Fig. 2. Specimens in the tensile machine. (A) without oven, (B) with oven.
M. Leone et al. / Composites: Part B 40 (2009) 85–93 87

Table 1 Table 3
Experimental program Mechanical properties of FRP EBR systems

Specimens FRP reinforcement Test temperature ( °C) FRP reinforcement ffu (MPa) efu (%) Ef (GPa) t Ef (GPa mm)
C_S_20_a CFRP sheet 20 CFRP sheet 2600 1.19 226 26.4
C_S_20_b CFRP sheet 20 GFRP sheet 780 – 73 21.9
C_S_50_a CFRP sheet 50 CFRP laminate 2450 1.41 176 176
C_S_50_b CFRP sheet 50
C_S_65_a CFRP sheet 65
C_S_65_b CFRP sheet 65
C_S_80_a CFRP sheet 80 gives effective estimation of the strength area of the reinforcement
C_S_80_b CFRP sheet 80 in the field of strengthening and/or upgrading of reinforced con-
C_L_20_a CFRP laminate 20
crete elements. In fact, according with the available guide line
C_L_20_b CFRP laminate 20
C_L_80_a CFRP laminate 80 [4,14] that area is calculated using the known area of the fiber,
C_L_80_b CFRP laminate 80 neglecting the total width and thickness of the cured system.
C_L_50_a CFRP laminate 50 The thermal characteristics of the two utilized epoxy resins
C_L_50_b CFRP laminate 50 were analysed using a Differential Scanning Calorimetry (DSC) by
G_S_20_a GFRP sheet 20
G_S_20_b GFRP sheet 20
means of a thermoanalyzer (DSC-7 Perkin Elmer). The analysis
G_S_80_a GFRP sheet 80 has been performed on three samples of each epoxy resin, after
G_S_80_b GFRP sheet 80 the complete hardening was reached (curing of test coupons under
ambient conditions). Each sample was held for 1.0 min at 5 °C and
subsequently heated in a nitrogen atmosphere from 5 °C to 200 °C
at 10 °C/min. The glass transition temperature, Tg, of each speci-
T_III. The properties of the fresh concrete were evaluated perform- men has been calculated and the results averaged. For both resins
ing slump tests and flow tests, according to CEN [8,9]. The obtained the average Tg value equals 55 °C. In the Fig. 3 the plot of the scan-
results are reported in the Table 2, as well as the measured density ning performed to evaluate the glass transition temperature is
of the fresh concrete. The mechanical properties of the hardened reported. The curve is referred on a sample of PC5800 CARBO resin
concrete were experimentally evaluated at the same age of the but similar results were found for the other analyzed specimens.
bond shear test [10,11] e [12]. From the Fig. 3 it is possible to note the presence of a large endo-
In particular, compressive strength, fc,cub, was evaluate by thermic peak, that very probably is due to the physical aging phe-
means three single cubes with side dimensions of 150 mm, while nomenon [15]. As reported in experimental work on cold-cured
compressive cylindrical strength, fc, and splitting strength, fct,S1, epoxy resin [15–17] the physical aging superimposed on Tg making
were estimate testing three single cylinders with a diameter of the measure of it difficult. However, the obtained values of glass
150 mm and height of 300 mm, respectively. In addition flexural transition temperature are in agreement with other experimental
strength, fc, fl, was measured by means of 3-point bending tests observations on the same kind of epoxy trade system.
on three single prisms 150 150 600 mm with a clear span of For the higher viscosity resin, an additional DSC analysis was
500 mm; and splitting strength, fct,S2, were evaluated on the performed involving a pre-heating cycle between 50 °C and
remaining halves of the prisms. Finally, secant modulus of elastic- 200 °C (rate 10 °C/min); the obtained Tg was equal to 81 °C.
ity, Ec, was determined by axial compression test on concrete cyl-
inders with diameter of 150 mm and height of 300 mm. The
average results can be found in Table 2.
The FRP EBR system is based on continuous unidirectional car-
0.30
bon and glass fibres for CFRP and GFRP hand lay-up sheets, respec-
tively, and an angle ply (±45) laminate the for factory made CFRP.
The hand lay-up sheets and the factory made laminate were glued 0.25
ΔHrelaxation = 0.41(W˚C/g)
on the substrate by two different epoxy resins, (PC5800 CARBO for
hand lay-up sheets and PC5800/BL for factory made laminate. Both 0.20
Heat Flux (W/g)

adhesive are supplied by TRADECC). The resin utilized to glue the

laminate on the substrate presents higher viscosity in order to 0.15
facilitate the application procedure.
The tensile properties of the FRP reinforcement were experi- 0.10
mentally evaluated by means of tensile tests according to standard
[13]. The average experimental results, on three tested specimens,
0.05 Τ g= 53ºC
in terms of tensile strength (ffu), strain at failure (efu), elastic mod-
ulus (Ef) are given in Table 3. In this table, the effective stiffness of
0.00
the reinforcement (t Ef), equal to the nominal thickness (dry fibers
0 20 40 60 80 100 120 140 160 180
thickness for the hand layup sheet and whole laminate thickness Temperature (ºC)
for the factory made reinforcement, supplied by the manufactures)
multiplied by the elastic modulus, is also given. This parameters Fig. 3. Heat flux versus temperature curves.

Table 2
Quality control tests on fresh and hardened concrete – Experimental results

Batch Age of test (days) fc (MPa) fc,cub (MPa) fct,S1 (MPa) fct,fl (MPa) fct,S2(MPa) Ec (MPa) Slump (mm) Flow test Density (kg/m3)
T_I 31 – 50.3 3.3 3.9 3.3 31227 160 2.3 1905
T_II 30 32.1 50.2 – 3.8 3.2 27529 160 2.2 1813
T_III 30 32.1 41.3 – 3.7 3.1 32804 180 2.5 1915
88 M. Leone et al. / Composites: Part B 40 (2009) 85–93

measurements are located on the x axis, with the origin set at

FRP REINFORCEMENT x
the loaded end (Fig. 3). Referring to two consecutive strain
UN-BONDED ZONE
300 50 gauges, ranging Dxi, the equilibrium equation, assuming uniform
distribution of the bond stress in the analysed discrete interval,
39 50 50 50 35 35 35 SIDE A
gives:
150

100

9 8 7 6 5 4 2 1

25
Dei

50
3
sðxÞ ¼ Ef t ð1Þ
Dxi
400 with s(x) the bond stress in the FRP reinforcement between two
consecutive strain gauges, Ef and t the elastic modulus and the nom-
FRP REINFORCEMENT inal thickness of the FRP reinforcement, respectively, and Dei the
x
UN-BONDED ZONE measured strain difference between the two considered strain
300 50
gauges.
The slip between the FRP reinforcement and concrete is calcu-
17 16 15 14 13 12 11 10
SIDE B lated through integration of the strain along the bonded length.
150

100

39 50 50 50 35 35 35
On the basis of the strain compatibility hypothesis in the infinites-
imal range, dx, of the FRP–concrete interface and neglecting the
400
concrete strain, the following equation can be written:
Z Lb

Fig. 4. Strain gauges set-up. sðxÞ ¼ sð0Þ þ ef dx ð2Þ

o

with, s(x) the slip along the bond length, s(0) the slip at the loaded
The increase of the Tg value demonstrates the beneficial effect of end, ef the strain in the FRP reinforcement and Lb the bond length.
the resin post-cure. Assuming that the slip at the free end s(Lb) can be considered neg-
ligible before debonding (given that the provided bond length is
2.4. Specimen preparation and measurements longer than the minimum transfer bond length defined in fib Bulle-
tin 14 [4]), the total slip s can be obtained by integrating the strain
After casting the concrete specimens, the formwork was re- along the FRP sheet. In the discrete field one obtains:
moved after 1 day and the specimens were kept covered under a
X
n
plastic foil during 7 days and further stored in the laboratory envi- s¼ ei Dxi ð3Þ
ronment until the day of the test. The FRP reinforcement i¼1
(100 700 mm) was bonded at a concrete age of 15 days, taking
with n the number of strain measurements along the bond length
into account proper specimen preparation as specified by the man-
and Dxi the distance between two consecutive measurement points.
ufacturer. This comprised concrete roughening by means of grind-
The slip between FRP reinforcement and concrete can be also
ing. The FRP reinforcement was left un-bonded for approximately
directly measured with a displacement transducer during the test
50 mm in the central zone (Fig. 1). This was done to ensure that
(measured slip). As observed by other researchers performing bond
first cracking (separation of the two blocks) happened in the un-
tests, the experimental measurement of the slip between the
bonded area, allowing the proper evaluation of tensile force within
external FRP reinforcement and concrete often resulted not very
the reinforcement. The bond length equalled 300 mm for each
accurate. In this test programme, the experimental slip measure-
bond zone. A bond length of 300 mm is in agreement with the min-
ments are rather used to evaluate the relative behaviour of differ-
imum transfer length defined in fib Bullettin 14 [4] to establish the
ent specimens or possible eccentricities in loading. However, in
full anchorage capacity that an externally bonded FRP reinforce-
absolute terms, the slip values obtained from the experimental
ment can develop according to fracture mechanics modelling. At
strain measurements (Eq. 3) are utilized.
one prism end, the FRP was additionally fixed with clamps
During the data elaboration stage, a high scatter has been ob-
(Fig. 1), allowing the occurrence of the delamination of the rein-
served between similar specimens. Based on a critical review of
forcement only on the opposite prism end, where the strain gauge
the measurement data, results believed to be significantly influ-
measurements were provided.
enced by secondary effects (such as the presence of eccentricity
Electrical strain gauges were glued on the composite reinforce-
in the specimens or the inadequacy of some experimental mea-
ment on both sides according to Fig. 3. The relative displacement
surements), were omitted.
between the FRP reinforcement and concrete was recorded with
The averaged experimental results in terms of maximum bond
two displacement transducers (one per side of the monitored prism),
stress (Eq. 1, smax), measured and calculated slip (Eq. 3) are re-
fixed to the concrete and directly connected to the reinforcement.
ported in Table 4.
To make a more correct comparison of the maximum bond
stress between the specimens of different concrete batches the fol-
3. Experimental results
lowing multiplier coefficient was used:
1=2
3.1. Evaluation of the measurement fc;cub
ð4Þ
fc;cub;REF
In the following, reference will be made (unless stated other-
wise) to the average test results obtained by the two equivalent where fc,cub is the cubic compressive strength of a given specimen
specimens tested for each parameter combination. The experimen- and fc,cub,REF is the cubic compressive strength of specimens chosen
tal results are elaborated in terms of maximum force, maximum as reference between those analysed. This way to make the bond
bond stress and corresponding slip between the concrete and stress independent on the variability of concrete strength is consis-
reinforcement. tent with the indications provided by the relationship between
Bond stresses have been evaluated by utilizing experimentally bond strength and compressive concrete strength given in Model
recorded strains along the FRP. The positions of the strain Code [18].
M. Leone et al. / Composites: Part B 40 (2009) 85–93 89

Table 4 0.0010
Averaged experimental results
0.0009 C_L_80_a
Specimens Maximum bond stress Calculated slip Measured slip
0.0008 C_S_80_b
(MPa) (mm) (mm)
G_S_80_a
C_S_20 4.22 0.181 0.661
0.0007
C_S_50 5.62 0.197 0.508 0.0006

Strain (ε)
C_S_65 3.68 0.271 0.205
C_S_80 1.96 0.116 0.078
0.0005
C_L_20 5.65 0.087 0.270 0.0004
C_L_50 3.28 0.183 0.169
C_L_80 4.21 0.237 0.866
0.0003
G_S_20 3.41 0.186 0.645 0.0002
G_S_80 0.96 0.182 0.221
0.0001
0
0 100 200 300
Position (mm)
3.2. Influence of type of FRP reinforcement
Fig. 5. Strain versus position curves – F = 4 kN – T = 80 °C.

The experimental results reported in Table 4 show that the

maximum bond stress at room temperature (20 °C) varies with reinforcement stiffness and that the slope of the CFRP and GFRP
the type of reinforcement. In particular, the highest value of sheet (with almost the same effective stiffness) are very close.
the maximum bond stress was registered for specimens rein-
forced with CFRP laminates. The factory production of the lami- 3.3. Influence of service temperature
nates guarantees a high quality of the reinforcement, which
seems to have a positive effect on the maximum bond stress. Analysing the experimental results obtained at T = 20 °C,
Different maximum bond stresses are also obtained comparing T = 65 °C and T = 80 °C for the different reinforcement types, in all
specimens reinforced with CFRP and GFRP hand lay-up sheets, cases a decrease of the maximum bond stress can be observed
which seems again linked to the quality of the reinforcement, for elevated temperatures, beyond the glass transition tempera-
in terms of reduced imperfections involved during the applica- ture. In particular, smax decreases of 13% at 65 °C and 54% at
tion phase. In fact, in the available literature, as in [4,19,20] 80 °C in the case of specimens reinforced with CFRP sheets; the de-
and [21], is generally reported that mechanical properties of crease at 80 °C is of 25% and 72% as regard CFRP laminates and
reinforcement slightly influence the maximum bond stress, GFRP sheets, respectively.
which is mainly governed by the mechanical properties of the The Tg values of the two utilized epoxy resins are both around
substrate. Therefore, a possible reason for the difference in max- 55 °C, even if referring to the factory made laminate the above
imum bond stress when using CFRP or GFRP sheets may be property (Tg = 55 °C) characterizes only the adhesive used to glue
related to the different surface texture (CFRP sheets with fibres the composite on the substrate. In fact, the reinforcement itself is
in the main direction kept together by woven at almost ±45° obtained by the pultrusion process, where the matrix is cured at
with respect to the longitudinal direction; GFRP fabric with the a high temperature, significantly higher than the ambient temper-
longitudinal fibres kept by a small amount of woven at only ature, involving higher Tg values, certainly over 55 °C. Conse-
90°) that aids a better fiber alignment in the case of CFRP rein- quently, the lower bond stress decay for the specimens
forcement, guaranteeing a more correct stress transfer. reinforced with CFRP laminates could be linked to the different
Similar observations can be made for the experimental results chemical process of the resin, at least as concerns the matrix. How-
obtained for the bond tests at 80 °C. Conversely, an opposite trend ever it is believed, by the authors, that the obtained quality of the
was observed for the experimental results of tests performed at FRP–concrete interface is also of influence. In particular, referring
50 °C: the maximum bond stress for the CFRP laminate is lower to the registered scatters between CFRP and GFRP sheets, the dif-
than the maximum bond stress for the CFRP hand lay-up sheet. ferent surface textures (before mentioned) of the utilized products
This result will be further analysed in the next section. may affect the bond behaviour, in terms of fiber alignment, uni-
The bond behaviour along the bond length is reflected by the form resin distribution and thickness.
strain readings at different load levels, from which the bond
stress–slip relationship can also be derived. A comparison between
the strain along the reinforcement for the different FRP types used 6
is reported in Fig. 5 at a load level of 4 kN and a test temperature of C_L_20_a
80 °C. At this stage the specimen G_S_80_a is at 50% of its ultimate 5
load, Fu, while C_L_80_a and C_S_80_b are at 9% Fu and 32% Fu, C_S_20_a
respectively.
Bond Stress (MPa)

4
The strain curves along the bond length show the influence of
the reinforcement stiffness on the strain distribution. A significant
3
difference is noted between specimens C_L_80_a and the speci-
mens with the hand lay-up reinforcement, while almost the same
2
results are obtained between G_S_80_a and C_S_80_b specimens.
G_S_20_a
The observed difference is related to the difference between the
1
effective stiffness of the FRPs (Table 3). Similar results were found
for the other specimens tested at different temperatures.
0
The influence of the type of reinforcement has also been ana-
0 0.1 0.2 0.3 0.4 0.5
lysed referring to the bond stress (s) versus calculated slip (s), as
Slip (mm)
reported in Fig. 6. From Fig. 6 it can be noted that the slope of
the ascendant branch of the s–s curves increases with the effective Fig. 6. Comparison between types of reinforcement – T = 20 °C.
90 M. Leone et al. / Composites: Part B 40 (2009) 85–93

Comparing the maximum bond stress at T = 50 °C (which is 3.4. Failures

close to the glass transition temperature) of CFRP laminate and
CFRP hand lay-up sheets, with respect to that evaluated at room Two different types of failure were observed. In most cases a
temperature, a different trend of the bond degradation was ob- bond failure occurred, however for a majority of specimens rein-
served with respect to the previously discussed results obtained forced with CFRP sheet, fracture of the FRP reinforcement took
at 80 and 65 °C, as also mentioned previously (see section ‘‘Influ- place. The latter occurred at an ultimate load lower than the FRP
ence of type of FRP reinforcement”). While, an increase in maxi- tensile capacity, indicating local stress concentrations within the
mum bond stress of around 34% was found for the CFRP hand FRP sheet. Even though FRP fracture has also been observed by
lay-up sheet, a decrease of around 42% was obtained for the CFRP other authors [23] it is difficult to clearly identify the cause. Prob-
laminate. Similar contradicting results are reported in the litera- ably, the presence of geometrical imperfections, due to the mis-
ture [6,7,22]. Considering that the effective Tg value generally falls alignment of the two concrete prisms or the internal steel bars,
within a certain range of variation, the related chemical process introduced an additional bending moment that could be the reason
could happen at different temperature levels (within the men- for that FRP fracture. In addition the FRP fracture at high tempera-
tioned range) for the two FRP types. However, the precise influence tures could also be caused by the local delamination (matrix-fi-
of the chemical process of the resin within the Tg range on the bond bers), due to the adhesive degradation and the consequent non
performance of the FRP-concrete system, needs further study aim- uniform stress distribution within the fibers. For specimens rein-
ing also to have a more insight about the observed increase of the forced with CFRP laminates a bond failure was generally observed,
maximum bond stress in some cases (specimens with CFRP sheets indicating that the thicker and prefabricated CFRP laminate ap-
at 50 °C). pears less vulnerable to possible imperfections with respect to
Furthermore, it is observed that higher temperatures than those specimens with hand lay-up CFRP sheets.
used in this test program are needed to have a complete degrada- The bond failure of the specimens was typically characterized
tion of the bond performance. by a thin layer of a concrete attached to the FRP reinforcement. It
A comparison between the bond stress–slip curves at different was found that the bond failure aspect was influenced to a large
test temperature is given in Fig. 7. From this figure it can be ob- extent by the temperature, as illustrated in Fig. 8. At T = 50 °C a
served that by increasing the temperature, the stiffness at the cohesion failure [4] was observed, with a thin layer of concrete at-
interface FRP – concrete decreases besides the decay of the maxi- tached to the CFRP sheet. Cohesion failure happens if the tensile
mum bond stress, as discussed before. and shear strength of the adhesive (epoxy resin) are higher than
Results reported in this section indicate that the bond perfor- the tensile and shear strength of the concrete (which is generally
mance is significantly influenced by the service temperature; as a the case). Increasing the temperature (T = 80 °C) a debonding at
consequence, temperature effects must be properly taken into ac- the interface between adhesive and FRP reinforcement occurred
count from a design point of view. First of all a temperature value (adhesion failure), [4]. Indeed, by increasing the temperature the
(limit service temperature), below which any variation in bond mechanical properties of the glue dropped below that of the con-
performance is expected, must be given on the basis of the Tg val- crete. Failure at T = 65 °C was intermediate between cohesion
ues of the utilized resins, experimentally evaluated. Considering and adhesion failure. An analogous behaviour was observed for
the possible variation of the Tg value, a reliable limit service tem- the GFRP sheets, Fig. 8. Similar considerations about the influence
perature must be kept sufficiently below the evaluated Tg, namely of the service temperature on the bond failure are reported in [7].
10 °C 15 °C. For higher service temperatures the maximum bond In the case of the CFRP laminates and for all test temperatures a
stress should be reduced consistently, however more accurate mixed failure occurred, characterized by cohesion failure, adhesion
indications could be given on the basis of wider experimental data. failure at the adhesive–concrete interface and adhesion failure at
With regard to the initial stiffness of the bond stress–slip curves, it the adhesive–FRP interface, though adhesion failure tends to be
is related, as is well known, to the stiffness of the adhesive; there- predominant for high temperatures. This different behaviour may
fore the variation of the initial slope of the curves could be taken related to the different type of epoxy resin used to glue the lami-
into account reducing the shear modulus of the adopted adhesive nates on the concrete, namely a more thixothropic resin was used
at the design stage; that reduction must be carefully evaluated in this case. That property may influence the impregnation quality
by determining the mechanical properties of the adhesive at given of the substrate, that could be lowered by increasing the adhesive
temperatures. viscosity. Therefore the kind of failure (mixed failure) observed for
these specimens at room temperature could be explained by an
insufficient impregnation of some zones of the substrate. On the

T=20ºC
5
Bond Stress (MPa)

3 T=50 ºC

1 T=80ºC

0
0 0.1 0.2 0.3 0.4 0.5
Slip (mm)

Fig. 7. Bond stress–slip curves for varying test temperatures – CFRP laminate. Fig. 8. Reinforcement after failure.
M. Leone et al. / Composites: Part B 40 (2009) 85–93 91

other hand, increasing the temperature values, the kind of failure the load transfer mechanism is observed as well. In fact, from this
becomes similar to that observed for other kinds of reinforcement, strain distribution profiles two distinct stages depending on the le-
as in this condition the adhesive is always the critical component. vel of load, may be identified. The first stage is characterized by an
almost exponential trend of strain at low load level. The distance
3.5. Transfer length required for the strain to reach almost zero in the stage of linear
elastic material behaviour is the so-called initial transfer length
Fig. 9 shows the change in bond stress along the bond length, (Lt). Once bond interface failure starts at the loaded end, a further
comparing three consecutive regions, as a function of the relative increase in load gradually displaces the transfer region towards the
load level F/Fu. In particular the figure refers to specimens rein- unloaded end (beyond 70% Fu) and the curves becomes more irreg-
forced with GFRP hand-made sheets at room temperature ular. The limit of this second stage coincides with the bond failure
(Fig. 9A) and specimens reinforced with CFRP hand-made sheets (100% Fu). Analysing Fig. 10A and B, it seems to be evident as the
at 80 °C (Fig. 9B); similar trends were observed for all tested shape of the curve remains almost similar increasing the
specimens. temperature.
The curves are obtained from the stress measurements accord- In Fig. 11 the comparison between the strains along the rein-
ing to Eq. (1) for any given load level. With increasing the load, the forcement varying the service temperature is reported. Fig. 11A re-
graphs show that the bond stress in the region near the loaded end fers to specimens reinforced with CFRP sheets while Fig. 11B refers
(6 mm < x < 14 mm in the Fig. 4 and curve 6–41 mm in Fig. 9) reach to specimens reinforced with GFRP sheets. The curves within the
a peak and then start to decrease abruptly. At the same time, the same graph are plotted at a fixed load, that belongs to the elastic
bond stress in the adjacent region (41 mm < x < 76 mm in the range of the specimen tested at room temperature. It has to be no-
Fig. 4 and curve 41–76 mm in the Fig. 9) starts to increase. The ticed that the chosen load value could not be in the elastic stage of
decrease of the bond stress indicates bond failure in that region, the specimens tested at the high service temperature. Analyzing
while the build-up of the bond stress in the adjacent region indi- the figure it is possible to note as the strains along the reinforce-
cates that the load is transferred further along the bond length. ment increase with the service temperature. In particular the strain
This phenomenon is observed progressively from one region to value at the loaded end at 80 °C is double the strain value at 20 °C,
another until the total failure of the joint occurs [24]. registered in the same position. Similar results are reported in [7].
In Fig. 10 the strain versus position (starting from the loaded The initial transfer length was evaluated considering the dis-
end) is reported for different load levels with reference to speci- tance from the loaded end (x = 0 in the Fig. 3) to the point where
men G_S_20 (Fig. 10A) and G_S_80 (Fig. 10B) From this figure, the strain reaches almost zero: the strain value was considered

3.50 2.50
6-41 mm 6-41 mm
3.00 41-76 mm
41-76 mm 2.00 76-111 mm
Bond Stress (MPa)

Bond Stress (MPa)

2.50 76-111 mm

1.50
2.00

1.50
1.00
1.00
0.50
0.50

0.00 0.00
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
F/Fu F/Fu

Fig. 9. Bond stress as function of relative load level – (A) G_S_20, (B) C_S_80.

0.007 0.004
10% Fu
0.006 30% Fu
50% Fu
0.003 70% Fu
0.005
100% Fu
10% Fu
Strain (ε)

0.004
Strain (ε)

30% Fu 0.002
50% Fu
0.003
70% Fu
100% Fu
0.002
0.001
0.001

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Position (mm) Position (mm)

Fig. 10. Strain versus position curves – (A) G_S_20, (B) G_S_80.
92 M. Leone et al. / Composites: Part B 40 (2009) 85–93

0.0025 0.0015
T=20ºC T=20ºC
T=80ºC T=80ºC
0.0020

0.0010

Strain (ε )
Strain (ε )

0.0015

0.0010
0.0005

0.0005

0.0000 0.0000
0 100 200 300 0 100 200 300
Position (mm) Position (mm)

Fig. 11. Strain along the reinforcement with increasing service temperature – (A) C_S, (B) G_S.

Table 5 The bond tests performed in the range of the glass transition
Evaluated transfer length temperature of the matrix and the adhesive of the FRP gave
Specimens Initial transfer length (mm) Maximum load (kN)
anomalous results. Probably, the chemical processes occurring
at the Tg temperature range could be the cause of the scattered
C_S_20 40.42 11.99
C_S_50 40.74 14.85
results. How this chemical processes influences the bond perfor-
C_S_65 108.16 12.82 mance of FRP reinforced concrete elements needs further
C_S_80 104.42 10.74 investigation.
C_L_20 99.95 40.47 The type of failure changes with increasing test temperature.
C_L_50 152.87 34.33
Specimens tested at T = 50 °C show cohesion failure within the
C_L_80 251.32 44.02
G_S_20 68.37 14.63 concrete. Increasing the temperature (T = 80 °C), an adhesion
G_S_80 214.81 11.64 failure at the interface was observed. At temperature similar
to or higher than the Tg the adhesion strength of the adhesive
drops below that of the concrete, causing the bond failure at
the FRP reinforcement–adhesive interface.
negligible (almost zero) at 1% of the strain measured at the loaded The strain values along the reinforcement significantly increases
end for each load level. The obtained Lt values are reported in Table with temperature.
5. In this table also the maximum load of each specimens (Fu) is The initial transfer length increases with the test temperature
reported. while the maximum force remains almost constant. In particular
From the Table 5 it is possible to note that for all types of tested the initial transfer length increases of 2.5–3 times with respect
FRP reinforcement, the initial transfer length increases with the to that at 20 °C. This result appears to be particularly interesting
test temperature while the maximum force remains almost con- because the apparent independence of the maximum load on
stant. In particular Lt increases with a factor 2.5–3 comparing re- the service temperature, in the tested range, appears related to
sults obtained at 80 °C with results obtained at 20 °C. This result the considerable increase of the initial transfer length.
appears to be particularly interesting since it clarifies the apparent Temperature effects must be properly taken into account from a
independence of the maximum load on the service temperature, design point of view. First of all a temperature value (limit ser-
for the tested temperature range. In fact, the higher ultimate load vice temperature) below which any variation in bond perfor-
values, registered also with the temperature increase, is basically mance is expected must be given, on the basis of the Tg values
linked to the considerable increase of the initial transfer length. of utilized resins. For higher service temperatures the bond
stress slip curves must be consistently evaluated, however more
4. Conclusions accurate indication could be given on the basis of a wider exper-
imental data.
On the basis of the performed experimental investigation and
the related results obtained the following considerations can be
remarked: References

The bond stress–slip curves with increasing service tempera- [1] Matthys S, Taerwe L. FRP reinforcement: developments in Europe. Concrete,
The Concrete Society 2001;35(6):20–1.
tures show a similar qualitative shape even if a significant vari- [2] Aiello MA, Frigione M, Leone M, Aniskevich AN, Plushchik OA. Effect of
ation of some relevant parameters were observed. The initial temperature on mechanical performances of CFRP rebars. In: Proceeding of the
slope of the ascending branch of the curves drops when the tem- 24th international SAMPE Europe conference of the society for the
advancement of material and process engineering, Paris, France, 2003. p.
perature increases. The maximum bond stress decreases with 239–49.
service temperatures above the glass transition temperature of [3] Frigione M. The transport of water and its effects on properties od cold-curing
the adhesive. In particular, at 80 °C smax decreases by 54% in epoxy adhesive, in: Zaikov GE, et al., editors. Homolytic and heterolytic
reaction: problem and solution, 2004, p. 283–305.
the case of CFRP sheet, 72% for GFRP sheet and 25% for CFRP lam- [4] CEB-FIP Task Group 9.3. Externally bonded FRP reinforcement for RC
inate with respect to the room temperature. structures. Comité Euro-International du Béton Bulletin 14, 2001.
M. Leone et al. / Composites: Part B 40 (2009) 85–93 93

[5] Tadeu AJB, Branco FJFF. Shear tests of steel plates epoxy bonded to concrete [17] Frigione M, Aiello MA, Naddeo C. Water effects on the bond strength of
under temperature. Journal of Materials in Civil Engineering 2000;12(1):74–80. concrete/concrete adhesive joints. Construction and Building Materials
[6] Blontrock H, Taerwe L, Vanwalleghem H. Bond testing of externally glued FRP 2006;20:957–70.
laminates at elevated temperature. In: Proceeding of the international [18] CEB-FIP. CEB-FIP model code 1990. Comité Euro-International du Béton Bull.
conference ‘‘bond in concrete- from research to standard”, Budapest, d’Information, Lusanne, Switzerland, 1991. p.203–5.
Hungary, 2002, p.648–54. [19] Dai JG, Ueda T, Sato Y. Development of the non-linear bond stress–slip model
[7] Klamert EL, Hordijk DA, Janssen JM. The influence if temperature on the of fiber reinforced plastic sheet – concrete interface with a simple model.
debonding of externally bonded CFRP. Special issue of ACI 2005;230:1551–70. Journal of Composites for Construction 2005;9:52–62.
[8] UNI EN 12350-2. Testing fresh concrete – Slump test, 2001. [20] Lu XZ, Teng JG, Ye LP, Jiang JJ. Bond-slip model for FRP sheet/plate-to-
[9] UNI EN 12390-5. Testing hardened concrete – Flexural strength of test concrete interfaces. In: Proceeding of the second international
specimens, 2002. conference advanced polymer composites for structural applications in
[10] UNI 6556. Testing of concrete – Determination of static modulus of elasticity construction. Guildford, UK: University of Surrey; 2004. p. 152–61.
in compression, 1976. April.
[11] UNI EN 12390-3. Testing hardened concrete – Compression strength of test [21] Nakaba K, Kanakubo T, Furuta T, Yoshizawa H. Bond behavior between fiber
specimens, 2003. reinforced polymer laminates and concrete. ACI Structural Journal
[12] UNI EN 12390-6. Testing hardened concrete – Tensile splitting strength of test 2001;98(3):359–67.
specimens, 2002. [22] Galati N, Nanni A, Dharani LR, Focacci F, Aiello M. Thermal effects on bond
[13] ASTM D3039. Standard Test method for tensile properties of Polymer between FRP rebars and concrete. Construction and Building Material
Composite Materials, 2000. 2006;37(8):1223–30.
[14] ACI 440.2R-02. Design and construction of externally bonded FRP system for [23] Brosens K. Anchorage of externally bonded steel plates and CFRP laminates for
strengthening concrete structures, 2002. the strengthening of concrete structures. Ph.D Thesis, Katholieke Universiteit
[15] Frigione M, Naddeo C, Acierno D. Cold-curing epoxy resins: aging and Leuven, Facolteit Toegepaste Wetenschappen, Department Burgerlijke
environmental effects. I: Thermal properties. Journal of Polymer Engineering Bouwkunde, Heverlee, Belgium, 2001.
2001;21:23–51. [24] Bizindavyi L, Neale KW. Transfer length and bond strength for composites
[16] Frigione M, Lettieri M, Mecch AM. Environmental effects on epoxy adhesives bonded to concrete. Journal of Composites for Construction
employed for restoration of historical buildings. Journal of Materials in Civil 1999;3:153–60.
Engineering 2006;18(5):715–22.