Cement & Concrete Composites 26 (2004) 765–775

www.elsevier.com/locate/cemconcomp

Flexural strengthening of RC continuous beams using

CFRP laminates
A.F. Ashour *, S.A. El-Refaie, S.W. Garrity
Department of Civil and Environmental Engineering, University of Bradford, West Yorkshire BD7 1DP, UK
Received 24 July 2002; accepted 14 July 2003

Abstract
This paper presents test results at failure of 16 reinforced concrete (RC) continuous beams with different arrangements of internal
steel bars and external carbon fibre reinforced polymer (CFRP) laminates. All test specimens had the same geometrical dimensions
and were classified into three groups according to the amount of internal steel reinforcement. Each group included one un-
strengthened control beam designed to fail in flexure. Different parameters including the length, thickness, position and form of the
CFRP laminates were investigated. Three failure modes of beams with external CFRP laminates were observed, namely laminate
rupture, laminate separation and peeling failure of the concrete cover attached to the laminate. The ductility of all strengthened
beams was reduced compared with that of the respective unstrengthened control beam.
Simplified methods for estimating the flexural load capacity and the interface shear stresses between the adhesive and concrete at
failure of beams tested are presented. Comparisons between results from experiments and those obtained from the simplified
methods show that most beams were close to achieving their full flexural capacity and the longitudinal elastic shear stresses at the
adhesive/concrete interface calculated at beam failure conformed to the limiting value recommended in the Concrete Society
Technical Report 55.

2003 Published by Elsevier Ltd.

1. Introduction was prevented but peeling failure of the concrete cover

still occurred. Other researchers [2,4–6] studied the use
Using carbon fibre reinforced polymer (CFRP) lam- of end anchorage techniques such as U-shape and L-
inates has proved to be an effective means of upgrading shape jackets and steel bolts. Those end anchorage
and strengthening reinforced concrete (RC) beams. techniques increased the ultimate failure load, but did
However, premature failures such as peeling failure and not prevent peeling failure.
laminate separation can significantly limit the capacity Although many in situ RC beams are of continuous
enhancement and prevent the full ultimate flexural ca- construction, there has been very little research into the
pacity of the retrofitted beams from being attained. behaviour of such beams with external reinforcement
Laminate peeling or separation may occur due to the [4,7–9]. In addition, most design guidelines [10,11] were
high longitudinal shear and transverse normal stresses at developed for simply supported beams with external fi-
the plate end resulting from the abrupt curtailment of bre reinforced polymer laminates. The principal aims of
the plate [1,2]. this paper may be summarised as follows:
Several studies were conducted to identify ways of
preventing peeling failure with a view to improving the • to present test results at failure of 16 RC continuous
load capacity and ductility of strengthened concrete beams strengthened with different arrangements of
beams [2–6]. Garden and Hollaway [3] extended the external CFRP laminates, including the analysis
plate ends under the end supports in order to obtain full and discussion of different modes of failure, beam
plate end anchorage. It was found that plate separation ductility and enhancement of load and moment ca-
pacities;
*
Corresponding author. Tel.: +44-1274-233-870; fax: +44-1274-234-
• to develop a simplified technique for estimating the
111. flexural capacity of RC continuous beams strength-
E-mail address: [email protected] (A.F. Ashour). ened with external CFRP laminates;

0958-9465/$ - see front matter
2003 Published by Elsevier Ltd.

doi:10.1016/j.cemconcomp.2003.07.002
766 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775

Nomenclature

Afl area of CFRP at concrete crushing and ten- Mhe theoretical negative flexural capacity of the
sile rupture of CFRP laminates unstrengthened section at the CFRP sheet
Af area of CFRP laminates end
As internal longitudinal tension steel reinforce- Mse theoretical positive flexural capacity of the
ment area unstrengthened section at the CFRP sheet
A0s internal longitudinal compression steel rein- end
forcement area nf external plate modular ratio
b section width Pt theoretical ultimate mid-span point load ob-
C concrete compressive force tained from flexural mechanism
Cs compressive force in steel reinforcement Pth 2Pt , total theoretical ultimate load on both
above the neutral axis spans obtained from flexural mechanism
d effective depth of internal longitudinal ten- Pu experimental failure load of test specimens
sion steel bars Q summation of the internal forces
d0 depth of internal longitudinal compression Ts tensile force in steel reinforcement below the
steel bars from the top face of the section neutral axis
df depth of CFRP laminates Tf tensile force in external CFRP laminates
Es Young’s modulus of steel bars tf thickness of CFRP laminates
Ef elastic modulus of CFRP laminates V shear force calculated at beam failure
Ec elastic modulus of concrete x neutral axis depth
fcu concrete compressive strength yf depth of the CFRP plate measured from the
ff tensile stresses of CFRP laminate neutral axis to the centroid of the plate
ffu ultimate strength of CFRP laminates Ic transformed second moment of area of the
fs stress in tensile steel bars cracked plated reinforced concrete cross-sec-
fs0 stress in compressive steel bars tion in terms of concrete
fy yield strength of steel bars n ultimate load enhancement ratio
fu ultimate strength of steel bars g ultimate moment enhancement ratio
k1 ratio of the average compressive stress to the s longitudinal shear stress at the adhesive level
concrete cube strength fcu lD deflection ductility index
k2 ratio of the depth of the idealised rectangular Du mid-span deflection at beam ultimate load
stress block to the neutral axis depth Dy mid-span deflection at the lower yielding load
L beam span of the tensile reinforcement over the central
L1 length of the CFRP sheets over the central support or the beam mid-span
support ec concrete strain at the extreme compression
L2 length of the CFRP sheets at the beam soffit fibre
Mh experimental negative flexural failure mo- ef CFRP laminates strain at the extreme tension
ment of test specimens fibre
Ms experimental positive flexural failure moment ecu concrete crushing strain
of test specimens efu tensile rupture strain
Mu theoretical flexural strength of RC section es tensile steel bar strain
with CFRP laminates e0s compressive steel bar strain
Mus theoretical flexural capacity of mid-span sec- kl ratio between experimental failure load and
tion flexural capacity of test specimens
Muh theoretical flexural capacity of central sup-
port section

• to calculate the reserve capacity of beams tested if 2. Test programme

premature failure was prevented;
• to compare the longitudinal elastic shear stresses The test specimens consisted of 16 RC continuous
along the adhesive/concrete interface at the beam fail- beams classified into three groups (H, S and E)
ure with the limiting values suggested by the Concrete according to the arrangement of the internal steel rein-
Society Technical Report 55 [11]. forcement as given in Table 1. Beams in group H
 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775 767

Table 1
Details of the test specimens
Group no. Beam no. Main longitudinal steel Form of CFRP Top CFRP over the Bottom CFRP at mid- fcu (N/mm2 )
laminates central support span
Top Bottom No. L1 (m) No. L2 (m)
H H1 2T8 2T20 – – – – – 24.0
H2 2T8 2T20 CFRP sheets 2 2.0 – – 43.5
H3 2T8 2T20 CFRP sheets 6 2.0 – – 33.0
H4 2T8 2T20 CFRP sheets 10 2.0 – – 33.2
H5 2T8 2T20 CFRP sheets 6 1.0 – – 46.0
H6 2T8 2T20 CFRP sheets 2 3.0 2 1.0 44.0
S S1 2T20 2T8 – – – – – 26.0
S2 2T20 2T8 CFRP sheets – – 2 2.0 42.9
S3 2T20 2T8 CFRP sheets – – 6 2.0 33.3
S4 2T20 2T8 CFRP sheets – – 6 3.5 42.8
S5 2T20 2T8 CFRP sheets – – 10 3.5 24.4
E E1 2T16 2T16 – – – – – 24.0
E2 2T16 2T16 CFRP plate 1 2.5 – – 43.6
E3 2T16 2T16 CFRP plate – – 1 3.5 47.8
E4 2T16 2T16 CFRP plate 1 2.5 1 3.5 46.1
E5 2T16 2T16 CFRP sheets 6 2.5 – – 44.7

(H1–H6) were reinforced with two bars of 8 mm diam- The position, length, thickness and form (sheets or
eter on the top side of the beam and two bars of 20 mm plates) of the CFRP laminates were the main parameters
diameter on the bottom side, whereas beams in group S investigated as summarised in Table 1. Beams H1, S1
(S1–S5) were reinforced with an opposite arrangement and E1 had no external reinforcement and were used as
of the internal longitudinal steel reinforcement as given control specimens. Beams E2, E3 and E4 were
Table 1. The top steel reinforcement of beams in group strengthened with CFRP plates of 1.2 mm thickness and
E (E1–E5) was the same as the bottom steel reinforce- 100 mm width; whereas the rest of the beams were
ment, each consisting of two 16 mm diameter bars. strengthened with CFRP sheets, each layer of 0.117 mm
Vertical links of 6 mm bar diameter at 100 mm centres thickness and 110 mm width. The top of the central
were provided throughout each beam length in order to region of beam E5 was strengthened with CFRP sheets
prevent shear failure. All beams had the same geomet- (six layers of 0.702 mm total thickness · 110 mm wide-
rical dimensions: 150 mm wide · 250 mm deep · 8500 · 2500 mm long) of the same tensile strength and length
mm long. Beam geometry and reinforcement as well as as that of the CFRP plate used for beam E2. The CFRP
the loading and support arrangement are illustrated in laminates applied to the top face of the beams were
Fig. 1. placed symmetrically about the central support and

Fig. 1. Test set-up and strengthened RC beam details.

768 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775

those applied to the bottom face of the beam were truded CFRP plates, and epoxy bonding and structural
positioned symmetrically about the centres of both adhesives were provided by Weber and Broutin (UK)
spans. Ltd. The epoxy bonding and structural adhesives were
used to bond the CFRP embrace sheets and pultruded
CFRP plates, respectively in according with the manu-
3. Material properties facturer’s recommendations. Details of the mechanical
properties of these strengthening materials, taken from
The compressive strength fcu of the concrete obtained the manufacturer’s data sheets, are summarised in Table
for each beam from compression tests carried out on 3.
three 100 mm cubes is given in Table 1. The mechanical
properties of the steel reinforcing bars used in the test
beams were obtained by carrying out uniaxial tensile 4. Test results
tests on three steel bar specimens. Table 2 shows the
yield strength fy , ultimate strength fu and Young’s Each test beam, which comprised two equal spans of
modulus Es obtained for different types of steel bars 3830 mm, was loaded as shown in Fig. 1. Only test re-
used. The unidirectional CFRP embrace sheets, pul- sults at failure are presented here and full test results are

Table 2
Properties of internal steel reinforcing bars
Bar diameter (mm) Yield strength, fy (N/mm2 ) Ultimate strength, fu (N/mm2 ) Young’s modulus, Es (kN/mm2 )
6 (plain round mild steel stirrups) 308 355 200
8 (high yield deformed steel bars) 505 605 200
16 (high yield deformed steel bars) 520 630 201
20 (high yield deformed steel bars) 510 615 200

Table 3
Properties of strengthening materials
Material property Bonding adhesive Structural adhesive CFRP sheet CFRP plate
Compressive strength (N/mm2 ) 80 85 N/A N/A
Tensile strength (N/mm2 ) 17 19 3900 2500
Young’s modulus (kN/mm2 ) 5.0 9.8 240 150
Flexural strength (N/mm2 ) 28 35 N/A N/A
Bond to concrete (N/mm2 ) 4.0 20.0 N/A N/A

Table 4
Experimental results at failure of test specimens
Beam no. Failure modes Pu (kN) n
H1 1 138.0 1.00
H2 2 followed by 1 152.3 (121.5a ) 1.10 (0.88a )
H3 3 172.9 1.25
H4 3 162.6 1.18
H5 3 162.6 1.18
H6 2 followed by 3 172.9 (121.5a ) 1.25 (0.88a )
S1 1 83.6 1.00
S2 4 121.8 1.46
S3 3 121.8 1.46
S4 3 170.5 2.04
S5 4 111.7 1.34
E1 1 149.7 1.00
E2 3 178.6 1.19
E3 3 207.0 1.38
E4 3 231.4 1.55
E5 3 174.6 1.17
Pu ¼ total ultimate load (¼2P where P is the mid-span point load); n ¼ ultimate load enhancement ratios.
a
Tensile rupture of the CFRP sheets occurred over the central support.
 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775 769

given elsewhere [12–14]. Failure modes of test specimens

are summarised in Table 4 and described below.

4.1. Failure Mode 1

Conventional ductile flexural failure due to yielding

of the internal tensile steel reinforcement followed by
concrete crushing at both the central support and mid-
span sections occurred for the three control beams H1,
S1 and E1 as shown in Fig. 2 for beam E1.

4.2. Failure Mode 2 Fig. 3. Tensile rupture of the CFRP sheets (beam H2).

Beams H2 (see Fig. 3) and H6 exhibited tensile rup-

4.3. Failure Mode 3
ture of the CFRP sheets over the central support at 80%
and 70% of each beam failure load, respectively. Tensile
Peeling failure of the concrete cover along the steel
rupture of CFRP laminates did not cause complete
reinforcement level adjacent to the external CFRP
collapse of beams H2 and H6. As the load increased
laminates (see Fig. 4 for beam H5) occurred for a large
beyond tensile rupture of the CFRP sheets, beam H2
number of strengthened beams in the test programme as
failed in a conventional ductile flexural failure similar to
given in Table 4. It was brittle, explosive and was fol-
that of the control beam H1 and peeling failure of the
lowed immediately by beam failure. Extending the
concrete cover attached to the soffit sheets in beam H6
CFRP sheet to cover the entire negative moment zone
occurred. Rupture of the CFRP sheets was sudden and
such as in beams H3, E2, E4 and E5 or the entire po-
accompanied by a loud noise indicating a rapid release
sitive moment zone such as in beams S4 and E3 did not
of energy.
prevent peeling failure of the cover concrete. However,
beams E5 and E2 strengthened with different form of
CFRP laminates, they exhibited similar peeling failure
of the concrete cover adjacent to the CFRP laminates.

4.4. Failure Mode 4

CFRP sheet separation without concrete attached

was noticed in beams S2 (Fig. 5) and S5. It was ac-
companied by a little noise before beam failure indi-
cating impending failure of the adhesive. This may be as
a result of the low compressive strength of concrete,
especially in case of beam S5, which affects the bond
strength at the adhesive/concrete interface.

Fig. 2. Conventional ductile failure of control beam E1. (a) Over

central support; (b) at mid-span. Fig. 4. Peeling failure of the cover concrete (beam H5).
770 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775

surface at the central support and the beam soffit was

found to be the most effective arrangement of the CFRP
laminates to enhance the beam load capacity such as in
beams H6 and E4.
The failure moments and the ultimate moment en-
hancement ratios, g (the ratio between the ultimate
moment of a strengthened section in a strengthened
beam and that of the corresponding section in the re-
spective unstrengthened control beam) for all tested
beams are presented in Table 5. The negative and po-
sitive bending moments given in Table 5 were calculated
on the basis of satisfying equilibrium using the measured
Fig. 5. CFRP sheet separation without concrete cover (beam S2). end support reaction and mid-span applied load,
P ð¼ Pu =2Þ. Table 5 indicates that increasing the number
of CFRP layers increased the moment enhancement
5. Enhancement of load and moment capacities ratio of a strengthened section up to a certain amount
above which no further enhancement in the moment
The failure loads and ultimate load enhancement capacity could be achieved. For example, beams H2, H3
ratios, n (the ratio between the ultimate load of a and H4 exhibited an increase in the negative moment
strengthened beam and that of the corresponding un- capacity of the strengthened section at the central sup-
strengthened control beam) for all tested beams are port due to increasing the number of CFRP layers,
summarised in Table 4. It should be noted that the whereas increasing the number of CFRP layers between
concrete compressive strengths of the control beams beams S4 and S5 decreased the positive moment ca-
(24–26 MPa) are much lower than those of strengthened pacity.
beams (33–47.8 MPa) as given in Table 1. Therefore, the Tables 4 and 5 show that using CFRP laminates for
enhancement ratios of the strengthened beams are likely the strengthening of continuous beams is an effective
to be higher than would be the case if the same com- technique; the load and moment capacities can be in-
pressive strength of concrete was used for control and creased by factors of up to 2.04 and 3.34, respectively, as
strengthened beams in each group. The quality of the was the case for beam S4. By comparing the ultimate
concrete also influences the performance of strengthened load enhancement ratio of a strengthened beam and the
beams, especially with regard to the bond strength at the moment enhancement ratio of a strengthened section in
adhesive/concrete interface. This can be confirmed by the same beam, it can be concluded that the latter was
comparing the notable difference of the failure loads of significantly higher than the former. Such a conclusion is
test specimens S4 and S5 (with concrete compressive not valid for simply supported beams strengthened with
strength of 42.8 and 24.4 MPa, respectively). external reinforcement where the moment and load en-
For a specified length of CFRP sheets, there was an hancement ratios are always the same.
optimum number of CFRP layers above which the beam
load capacity was not improved (beams H3 and H4 in
group H and beams S2 and S3 in group S), as indicated 6. Beam ductility
in Table 4. Increasing the length of the CFRP sheets was
found to increase the load capacity of the strengthened The deflection ductility index, lD , used for simply
beams as in the case of beams H3 and H5 in group H supported beams by Mukhopadhyaya et al. [15] is
and beams S3 and S4 in group S. However, when tensile adopted to measure the ductility of the continuous
rupture is likely to occur, increasing the CFRP sheet beams tested by the authors as given in Table 5. The
length was not effective. This can be noticed by com- deflection ductility index, lD , is defined as:
paring the load at which tensile rupture of the CFRP Du
sheets occurred over the central support for beams H2 lD ¼ ð1Þ
Dy
and H6 (121.5 kN for both beams as indicated in Table
4). After tensile rupture, beam H2 behaved as an un- where Du is the mid-span deflection at beam ultimate
strengthened beam similar to the control beam H1. load and Dy is the mid-span deflection at the lower
However the failure load of beam H2 (152.3 kN) is yielding load of the tensile reinforcement over the cen-
higher than that of the control beam H1 (138.0 kN) due tral support or the beam mid-span. As can be seen from
to the difference in the concrete compressive strength. Table 5, all strengthened beams exhibited less ductility
Using CFRP sheets of equivalent total tensile load ca- than the corresponding unstrengthened control beams.
pacity to the CFRP plates produced nearly the same Only beam H2 showed nearly similar ductility at failure
load capacity (beams E2 and E5). Strengthening the top as that of the control beam H1 because both beams
 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775 771

Table 5
The moment, moment enhancement ratio, g and ductility index at failure of test specimens
Beam no. Mh (kN m) Ms (kN m) g lD
H1 21.21 56.78 1.00 4.10
H2 31.60 61.00 1.49 (h) 4.34 (1.83a )
H3 46.48 59.56 2.19 (h) 1.35
H4 53.07 51.32 2.50 (h) 1.74
H5 35.00 64.27 1.65 (h) 2.64
H6 28.26a 70.24 1.33 (h); 1.24 (s) 3.46 (1.71a )
S1 57.77 11.13 1.00 12.85
S2 71.28 22.67 2.04 (s) 6.21
S3 66.90 24.72 2.22 (s) 2.93
S4 88.97 37.15 3.34 (s) 2.52
S5 50.18 28.36 2.55 (s) 1.00b
E1 54.49 44.41 1.00 6.12
E2 79.78 45.64 1.46 (h) 3.58
E3 53.56 72.35 1.63 (h) 3.21
E4 77.00 72.29 1.41 (h); 1.63 (s) 2.10
E5 77.42 44.87 1.42 (h) 2.60
Mh and Ms ¼ ultimate negative and positive bending moments; g ¼ ultimate moment enhancement ratio; (h) ¼ over the central support section and
(s) ¼ at mid-span section.
a
Tensile rupture of the CFRP sheets occurred over the central support.
b
There was no ductility.

failed in flexure at the ultimate load level. However, if point load, Pt , for beams H4 and H5 was calculated
the ductility was considered at tensile rupture of the from (see Fig. 6(b)):
CFRP sheets, beam H2 would have less ductility than  
2 Mus L1 þ Mhe L
that of the control beam H1. Increasing the thickness Pt ¼ þ 2Mus ð3Þ
and length of CFRP laminates was found to decrease L L  L1
the beam ductility. and that for beams S2 and S3 was determined from (see
Fig. 6(c)):
 
2 2L
Pt ¼ Muh þ Mse ð4Þ
7. Predictions of capacity of tested beams L L  L2

Simplified methods for estimating the flexural load where Mhe and Mse are the negative and positive moment
capacity and the elastic shear stresses at the adhesive/ capacities of the unstrengthened sections at the CFRP
concrete interface at failure of beams tested are given in sheet end, respectively, and L1 and L2 are the lengths of
this section. the CFRP sheets over the central support and at the
beam soffit, respectively as shown in Fig. 6(b) and (c).
The calculation of flexural capacities of different critical
7.1. Flexural capacity of beams tested sections as required by Eqs. (2)–(4) is based on the nu-
merical method presented in Appendix A.
The prediction of the flexural capacity of the test The flexural mechanism of failure presented above is
specimens was based on satisfying equilibrium condi- only suitable for test specimens in which ductile be-
tions and assuming that both the mid-span and central haviour was dominant, such as the control beams H1,
support sections have high ductility and reached their S1 and E1. It may also be used for test specimens partly
flexural capacities Mus and Muh , respectively as shown in strengthened over the central support or at soffit mid-
Fig. 6(a). Accordingly, from equilibrium considerations, spans where ductile failure occurred at the unstrength-
the applied point load, Pt , at the beam mid-span is cal- ened section and then brittle flexural failure took place
culated from (beam self weight is neglected): at the strengthened section due to either concrete
2 crushing or CFRP laminate rupture. However, the au-
Pt ¼ ðMuh þ 2Mus Þ ð2Þ thors applied the flexural mechanism to the beams with
L
CFRP laminates bonded over the central support or at
When Eq. (2) is applied to beams H4, H5, S2 and S3, it mid-span soffit to calculate the reserve capacity of test
was found that the bending moment at the unstrength- specimens if brittle peeling or separation failure of
ened section next to the sheet end (the end closer to the CFRP laminates was prevented. For beams fully
end support for soffit sheets) is higher than the moment strengthened with CFRP bonded over the central sup-
capacity of that section. In such cases, the mid-span port and at the mid-span soffit where beam failure was
772 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775

different critical sections using Eq. (A.8) are also given in

Table 6. Good agreement between flexural load capacity
obtained from current theoretical method and that
measured during experiments for beams failed in flex-
ural mode such as the three control beams (H1, S1 and
E1) and beam H2 failed due to tensile rupture of the
CFRP laminates. This comparison also shows that,
apart from beams H4, S4 and S5, the strengthened
beams tested were close to developing their full flexural
capacity.

7.2. Interface shear stresses of beams tested

(a) Several analytical methods have been developed for

predicting peeling failure of RC beams strengthened
with external plates [16,17]. These methods, however,
were found to be derived for simply supported RC
beams strengthened with external reinforcement and,
therefore, are not directly applicable to continuous RC
beams. As elastic theory has been widely used by a large
number of researchers [1,11,18–20], it is applied to the
beams tested in order to estimate interface shear stresses
between the adhesive and concrete at failure.
According to the elastic theory, the longitudinal shear
stress at the adhesive level, s, is calculated from Eq. (5)
below:
Vnf tf yf
s¼ ð5Þ
(b) Ic

where V is the shear force calculated at beam failure; nf

is the CFRP laminate modular ratio (¼ Ef =Ec where Ef
and Ec are the elastic moduli of the CFRP laminate and
concrete, respectively); tf is the thickness of the CFRP
laminate; yf is the depth of the CFRP laminate measured
from the neutral axis to the centroid of the CFRP
laminate and Ic is the transformed second moment of
area of the cracked reinforced concrete cross-section
with external CFRP laminates in terms of concrete.
Table 6 presents the peak shear stresses, s, along the
CFRP laminate length calculated at the experimental
failure load of the CFRP laminates i.e. at rupture, sep-
aration or peeling of the concrete cover attached to the
(c) CFRP laminates. The peak shear stresses along the ad-
hesive level ranged from 0.80 to 1.55 N/mm2 for beams
Fig. 6. Ultimate positive and negative moments at critical sections of failed due to peeling failure apart from the soffit CFRP
continuous beams tested. Critical sections over the central support and
sheets in beam H6. Conversely, the peak shear stresses
at mid-span (a), at the plate end and mid-span (b), over the central
support and the plate end (c). calculated according to the elastic theory for beams that
exhibited tensile rupture of the CFRP sheets such as H2
and H6 (top sheets) were less than 0.80 N/mm2 . This
brittle and sudden, the flexural mechanism is not ap- listed data conformed to the 0.80 N/mm2 upper limit for
plicable. Table 6 presents the total load capacity, interface shear stresses proposed in the Concrete Society
Pth ð¼ 2Pt Þ, obtained from Eqs. (2)–(4) for all test speci- Technical Report 55 [11]. In other words, when the
mens except H6 and E4 and the ratio, kl ð¼ Pu =Pth Þ, be- elastic shear stresses along the adhesive level are found
tween experimental and theoretical load capacities. The to exceed 0.80 N/mm2 , peeling failure would be expected
calculated negative and positive flexural capacities of to occur.
 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775 773

Table 6
Theoretical negative and positive flexural moment capacities, load capacity and elastic shear stresses at concrete/adhesive interface of test specimens
Beam no. Muh (kN m) Mus (kN m) Pth (kN) kl s (MPa)
H1 12.52a 50.58 118.73 1.16 N/A
H2 33.83 60.33 161.35 0.94 0.48
H3 69.96 57.83 193.86 0.89 1.32
H4 87.08 (11.20b ) 57.83 210.19 0.77 1.55
H5 75.30 (11.67b ) 60.51 165.22 0.98 1.05
H6 33.96 70.24 –c –c 0.47d /0.16e
S1 49.78 10.68 74.30 1.13 N/A
S2 60.23 33.81 (11.61f ) 113.66 1.07 0.81
S3 57.98 70.21 (11.22f ) 109.60 1.11 1.12
S4 60.23 74.63 218.77 0.78 1.54
S5 54.25 81.06 225.97 0.49 1.15
E1 41.20 41.20 129.09 1.16 N/A
E2 82.97 42.44 175.30 1.02 0.80
E3 42.64 85.24 222.58 0.93 0.82
E4 84.17 84.17 –c –c 0.97d /0.75e
E5 83.35 42.48 175.78 0.99 0.79
a
Moment capacity calculated based on ultimate strength of steel bars as this beam exhibited very high rotational capacity before failure.
b
Moment capacity of negative unstrengthened section at the central support laminate end.
c
Beam is fully strengthened over the central support and mid-span soffit, therefore flexural mechanism is not applicable.
d
Interface shear stresses calculated at the top CFRP laminates.
e
Interface shear stresses calculated at the soffit CFRP laminates.
f
Moment capacity of positive unstrengthened section at the soffit laminate end.

8. Conclusions 5. The shear stresses calculated along the adhesive level

using elastic theory should not exceed 0.80 N/mm2 to
Test results at failure of 16 RC beams with external avoid brittle peeling failure of FRP laminates.
CFRP laminates have been presented. A simple method
to predict the flexural load capacity of beams tested has
been developed. The flexural failure of continuous Acknowledgements
beams is assumed when two critical sections attained
their flexural capacities that are estimated based on The second author acknowledges the grant provided
satisfying the compatibility and equilibrium conditions. from the Egyptian government to fulfil this research.
The elastic shear stresses at the adhesive/concrete in- The experimental work described in this paper was
terface is calculated at beam failure and compared with conducted in the Heavy Structures Laboratory in the
the limiting value suggested by the Concrete Society University of Bradford; the assistance of the laboratory
Technical Report 55 [11]. Based on the work described staff is acknowledged. The authors are grateful to Weber
in this paper, the following conclusions are drawn: and Broutin (UK) Ltd. for providing the CFRP rein-
forcement and the associated priming and bonding
1. All strengthened beams exhibited a higher beam load materials for the research.
capacity but lower ductility compared with their re-
spective unstrengthened control beams.
2. Brittle peeling failure of the concrete cover adjacent Appendix A. Bending capacity of strengthened sections
to the CFRP sheets was the dominant failure mode
of the strengthened beams tested. However, most In this appendix, the flexural capacity of critical sec-
beams were close to achieving their flexural capacity. tions as required by Eqs. (2)–(4) is based on the nu-
3. Increasing the CFRP sheet length to cover the entire merical method presented below [21,22]. The following
negative or positive moment zones did not prevent assumptions are considered:
peeling failure of the CFRP laminate and was found
to be ineffective when tensile rupture of the CFRP 1. Plane sections before bending remain plane after
sheets was the failure mode. bending.
4. Unlike simply supported beams, the enhancement of 2. The stress–strain relationships of concrete in com-
the bending moment capacity of a continuous beam pression and steel reinforcement in both tension and
due to external strengthening was found to be higher compression proposed by BS8110 [23] is adopted.
than that of the load capacity of the continuous 3. A linear stress–strain relationship of the CFRP lami-
beam. nate up to rupture is assumed.
774 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775

4. The tensile strength of concrete is ignored. compression as shown in Fig. 7(c), b is the section
5. Perfect bond exists between CFRP laminates and width and fcu is the concrete cube strength. k1 is the
concrete surface. ratio of the average compressive stress to the concrete
cube strength fcu and k2 is the ratio of the depth of the
A.1. Size of CFRP laminates at concrete crushing and idealised rectangular stress block to the neutral axis
CFRP rupture depth. They can be calculated according to the as-
sumed strain ec at the extreme compression fibre
At the instant of failure, either the concrete strain ec [21,23]; k1 ¼ 0:67 and k2 ¼ 0:9 for ec ¼ ecu ¼ 0:0035.
at the extreme compression fibre or the CFRP laminates • Considering equilibrium of the internal forces, the
strain ef at the extreme tension fibre reaches the re- area Afl of CFRP, where both concrete crushing
spective ultimate strain, that is ec ¼ ecu ¼ 0:0035 (con- and tensile rupture of CFRP laminates simulta-
crete crushing) or ef ¼ efu (tensile CFRP rupture), where neously occur, is calculated from:
ecu and efu are the ultimate strains of concrete and CFRP 0:603fcu bx þ A0s fs0  As fs
laminates, respectively. The area of the CFRP laminates Afl ¼ ðA:5Þ
ffu
that distinguishes between concrete crushing and tensile
rupture of the CFRP laminates is first estimated as where ffu is the ultimate tensile strength of CFRP
follows. laminates, and As and A0s are the internal longitudinal
tension and compression steel reinforcement areas,
• Calculate the neutral axis depth, x, the tensile steel respectively.
bar strain, es , and the compressive steel bar strain,
e0s , from the strain profile shown in Fig. 7(b) where
ec ¼ ecu ¼ 0:0035 and ef ¼ efu as given below: A.2. Estimation of bending capacity
0:0035
x¼ df ðA:1Þ By comparing the actual CFRP area Af used with Afl
0:0035 þ efu obtained from Eq. (A.5), strains in either the concrete
x  d0 extreme compression fibre or CFRP laminates may be
e0s ¼ 0:0035 ðA:2Þ
x predicted; that is:
d x
es ¼ 0:0035 ðA:3Þ ef ¼ efu Af 6 Afl
x ðA:6Þ
ec ¼ ecu Af > Afl
where df , d 0 and d are the depths of CFRP laminates,
compression steel bars and tension steel bars from the For complete determination of the strain distribution,
top face of the section as shown in Fig. 7(a). the neutral axis depth, x, is initially assumed and the
• Consequently, stresses in the tensile steel bars, fs , and correct value is iteratively determined when the equi-
compressive steel bars, fs0 , are determined from the bi- librium of internal forces is satisfied. Strains and
linear stress–strain relationship for the steel reinforce- stresses in different materials can then be calculated.
ment [23]. The summation of the internal forces, Q, can be deter-
• The concrete compressive force C is calculated from: mined:
C ¼ k1 k2 fcu bx ðA:4Þ
Q ¼ C þ Cs  Ts  Tf
where k1 and k2 are two parameters, defining the
idealised rectangular stress block of concrete in ¼ k1 k2 fcu bx þ A0s fs0  As fs  Af ff ðA:7Þ

(a) (b) (c)

Fig. 7. Strains, stresses and forces on RC section with externally bonded CFRP laminates at failure. (a) RC section with FRP; (b) section distri-
bution; (c) stresses and forces.
 A.F. Ashour et al. / Cement & Concrete Composites 26 (2004) 765–775 775

where Cs ¼ A0s fs0 ¼ compressive force in steel reinforce- mer Reinforcement for Reinforced Concrete Structures. American
ment above the neutral axis, Ts ¼ As fs ¼ tensile force in Concrete Institute; 1999. p. 647–57.
[9] Tann DB, Delpak R. Shear strengthening of continuous rein-
steel reinforcement below the neutral axis and forced concrete beams using externally bonded carbon fibre
Tf ¼ Af ff ¼ tensile force in external CFRP laminates as sheets. Concrete Communication Conference 2000, The 10th BCA
shown in Fig. 7. The equilibrium of the internal forces is Annual Conference on Higher Education and the Concrete
satisfied when the summation of forces, Q, is zero or lies Industry, 29–30 June, Birmingham, UK, 2000. p. 325–38.
within an accepted tolerance (say j CþA Q 6 [10] ACI Committee 440. Starte-of-the-art report on fiber reinforced
0 f 0 j < 10 ). If
s s plastic reinforcement for concrete structures, Report ACI 440R-
not, the neutral axis depth, x, is iteratively adjusted until 96. Detroit, USA: American Concrete Institute; 1996.
equilibrium of forces is satisfied. Consequently, the [11] Concrete Society. Design guidance for strengthening concrete
flexural strength, Mu , is calculated by taking moments of structures using fibre composite materials. Concrete Society
internal forces about the level of tensile steel as follows: Technical Report No. 55, 2000. 71p.
[12] El-Refaie SA, Ashour AF, Garrity SW. Tests of reinforced
 
k2 x concrete continuous beams strengthened with carbon fibre sheets.
Mu ¼ k1 k2 fcu bx d  þ A0s fs0 ðd  d 0 Þ þ Af ff ðdf  dÞ The 10th BCA Annual Conference on Higher Education and the
2 Concrete Industry, 29–30 June, Birmingham, UK, 2000. p. 187–
ðA:8Þ 98.
[13] El-Refaie SA, Ashour AF, Garrity SW. Strengthening of rein-
forced concrete continuous beams with CFRP composites. The
International Conference on Structural Engineering, Mechanics
References and Computation, Cape Town, South Africa, 2–4 April, 2001. p.
1591–8.
[1] Jones R, Swamy RN, Charif A. Plate separation and anchorage of [14] El-Refaie SA, Ashour AF, Garrity SW. Sagging strengthening of
reinforced concrete beams strengthened by epoxy-bonded steel continuous reinforced concrete beams using carbon fibre sheets.
plates. Struct Eng, Inst Struct Engrs, London 1988;66(5/1):85–94. The 11th BCA Annual Conference on Higher Education and the
[2] Swamy RN, Mukhopadhyaya P. Debonding of carbon-fibre- Concrete Industry, Manchester, UK, 3–4 July, 2001. p. 281–92.
reinforced polymer plate from concrete beams. Proc Inst Civil [15] Mukhopadhyaya P, Swamy RN, Lynsdale C. Optimizing struc-
Engrs Struct Buildings 1999;134:301–17. tural response of beams strengthened with GFRP plates. J
[3] Garden HN, Hollaway LC. An experimental study of the Compos Construct, ASCE 1998:87–95.
influence of plate end anchorage of carbon fibre composite plates [16] Smith ST, Teng JG. FRP-strengthened RC beams. I: Review of
used to strengthen reinforced concrete beams. Compos Struct debonding strength models. Eng Struct 2002;24(4):385–95.
1998;42:175–88. [17] Smith ST, Teng JG. FRP-strengthened RC beams. II: Assessment
[4] Arduini M, Nann A, Tommaso AD, Focacci F. Shear response of of debonding strength models. Eng Struct 2002;24(4):397–
continuous RC beams strengthened with carbon FRP sheets. 417.
Non-metallic (FRP) reinforcement for concrete structures. In: [18] Mukhopadhyaya P, Swamy N. Interface shear stress: A new
Proceedings of the Third International Symposium (FRPRCS-3), design criterion for plate debonding. J Compos Construct
Sapporo, Japan, October, vol. 1, 1997. p. 459–66. 2001;5(1):35–43.
[5] Spadea G, Bencardino F, Swamy RN. Structural behaviour of [19] El-Mihilmy MT, Tedesco JW. Prediction of anchorage failure for
composite RC beams with externally bonded CFRP. J Compos reinforced concrete beams strengthened with fiber-reinforced
Construct, ASCE 1998:132–7. polymer plates. ACI Struct J 2001:301–14.
[6] Lamanna AJ, Bank LC, Scott DW. Flexural strengthening of [20] Roberts TM. Approximate analysis of shear and normal stress
reinforced concrete beams using fasteners and fiber-reinforced concentrations in the adhesive layer of plated RC beams. Struct
polymer strips. ACI Struct J 2001;98(3):368–76. Eng, Inst Struct Engrs, London 1989;67(12/20):229–33.
[7] Khalifa A, Tumialan G, Nanni A, Belarbi A. Shear strengthening [21] Ashour AF. Size of FRP laminates to strengthen reinforced
of continuous reinforced beams using externally bonded carbon concrete sections in flexure. Struct Buildings J, Inst Civil Engrs
fiber reinforced polymer sheets. In: Fourth International Sympo- 2002;152(3):209–17.
sium on Fiber Reinforced Polymer Reinforcement for Reinforced [22] El-Mihilmy MT, Tedesco JW. Analysis of reinforced concrete
Concrete Structures. American Concrete Institute; 1999. p. 995– beams strengthened with FRP laminates. J Struct Eng, ASCE
1008. 2000;126(6):684–91.
[8] Grace NF, Soliman AK, Abdel-Sayed G, Saleh KR. Strengthen- [23] British Standards Institution. Structural use of concrete: code of
ing of continuous beams using fiber reinforced polymer laminates. practice for design and construction. BS8110, Part 1, Milton
In: Fourth International Symposium on Fiber Reinforced Poly- Keynes, UK, 1997.