Composite Structures 92 (2010) 2738–2746

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Investigation into bond behaviour of a new CFRP anchorage system for concrete
utilising a mechanically strengthened substrate
R. Kalfat 1, R. Al-Mahaidi *
Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, P.O. Box 218, Hawthorn, Vic. 3122, Australia

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

Article history: Failure of CFRP strengthening systems when applied to concrete structures is usually typified by de-lam-
Available online 22 April 2010 ination of the CFRP from the concrete substrate. To prevent this type of failure, current standards and
design guidelines impose strict limitations on the maximum strain level of the composite material which
Keywords: may be utilised in design. Emerging research has shown that anchoring the ends of the CFRP plates or
CFRP sheets can result in a significantly higher load/stress being reached before de-bonding occurs. This paper
Substrate strengthening investigates a method of anchoring CFRP laminates by utilisation of a mechanical chase cut into the con-
Bond
crete over the anchorage length. Experiments have shown this to be an effective way to improve the
Anchorage
Mechanical chase
strength of the concrete substrate, resulting in higher CFRP bond strength and improvements in maxi-
mum elongations, bond stress, slip and load carrying capacity of the CFRP to concrete joint.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction This is often the result of using higher modulus fibres and increas-
ing plate thicknesses or where several plies are required. In such
Due to the need of structural rehabilitation of buildings, bridges cases, design guidelines can limit the CFRP material strain to levels
as well as other infrastructure, many new rehabilitation technolo- as low as 10–25% of the ultimate CFRP material strain at rupture. In
gies have been introduced in recent years. Using carbon fibre practice these limitations result in severe under utilisation of the
reinforced polymers (CFRP) has become a popular method to CFRP material properties reducing economy.
strengthen existing buildings and bridge structures which are Current literature and design guidelines recognise the benefits
being subjected to higher bending, shear and torsional forces. A of CFRP anchorage systems to increase the bond strength of the
good example of this is the Westgate Bridge in Melbourne, Australia, CFRP to concrete connection. Research has shown that anchoring
which is one of the largest CFRP strengthening projects in the the ends of the CFRP plates or sheets results in a significantly high-
world [9]. er bond stress being reached before de-lamination occurs. When
CFRP material is also being used to repair older deteriorated sufficiently anchored, the CFRP material strain at failure can ap-
structures and possesses many advantages over existing strength- proach its ultimate strain at rupture. Design guidelines such as
ening methods. This is primarily due to the materials high tensile [1] recognise the benefits of anchorage systems and permit design-
strength and excellent durability properties. CFRP systems are ers to utilise a higher CFRP strain provided that the anchorage
characterised by their ease of installation and minimum mainte- device is backed up by sufficient experimental testing (Refer
nance requirements. Section 11.4.1.2).
It has been demonstrated that failure of concrete structures ret- There have only been limited attempts to investigate CFRP
rofitted with CFRP usually occurs by de-lamination of the CFRP anchorage measures and many remain to be quantified. This paper
from the concrete substrate. To prevent this type of failure, na- investigates a new type of CFRP anchorage solution which relies on
tional standards and design guidelines impose strict limitations increasing the mechanical properties of the concrete substrate to
on the allowable strain level in the composite material which which the FRP is bonded to over a nominal anchorage length.
may be safely utilised in design. In order to achieve acceptable lev-
els of concrete–FRP contact bond stress, allowable strains become 2. Development of a new FRP–concrete anchorage solution
lower in cases where a higher degree of strengthening is required.
2.1. Bond behaviour between FRP and concrete in the anchorage zone
* Corresponding author. Tel.: +61 (0) 3 9214 8429, mobile: +61 (0) 421 910 593.
E-mail addresses: [email protected] (R. Kalfat), [email protected] (R.
In an attempt to understand the bond behaviour in the anchor-
Al-Mahaidi). age zone, several bond strength models have been proposed by
1
Tel.: +61 (03) 9214 4610 (office), mobile: +61 (0) 43 441 3471. researchers. Experiments backed up by analytical and FE analysis

0263-8223/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2010.04.004
 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746 2739

have shown six possible distinct failure modes for CFRP bonded to 3. Experimental program
concrete: concrete failure, CFRP rupture, adhesive failure, CFRP de-
lamination, concrete to adhesive failure and plate to adhesive fail- 3.1. Specimen design
ure [6,12]. However, most experimental data has shown that fail-
ure typically occurs in the concrete a few millimetres beneath A full scale test set-up was designed using materials properties
the concrete/adhesive interface [10,14]. The strength of the con- prevalent on site for a comparative study. The factors which were
crete substrate is therefore of paramount importance and usually considered in the current test program included the thickness of
governs the bond strength between the CFRP and the concrete the concrete prism tc, CFRP bond length Lf and the width ratio be-
which can be achieved in practice. Substrate strength is typically tween the FRP strip and the concrete prism bp/bc. Bond strength
influenced by surface preparation and the strength/quality of the models developed by Täljsten and Yuan and Wu [13,16] have
concrete [3,5]. An adequately prepared substrate will ensure that shown that the thickness of the concrete prism can significantly af-
the forces from the fibres are adequately transferred to the con- fect the stress distribution within the specimen (refer Eq. (1)).
crete resulting in the expected composite action. Preparation usu- Using the model proposed by Yuan and Wu [16], it was found that
ally consists of removal of the cement laitance and all surface a concrete thickness, tc, of 250 mm exhibited a 1.0% reduction in
contaminants by either wet open grit blasting, mechanical grinding bond strength when compared to a 500 mm thickness. This illus-
or sandblasting. This is usually followed by water jetting and appli- trated that an adopted prism depth of 250 mm was sufficient to
cation of a suitable primer. not adversely influence the overall bond strength of the specimen.
The strength/quality of the concrete determines the tensile and It was expected that the member thickness prevalent in site appli-
shear strength of the substrate to resist the imposed peeling forces cations was to be greater than 250 mm.
at the plate ends combined with interface shear stresses that con- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tribute to de-bond [3]. These forces are associated with the 2Gf Ep t p
P u ¼ bp ð1Þ
concrete tensile strength and flexural rigidity of any cracked con- 1 þ ðEp tp bp =Ec tc bc Þ
crete. Design guidelines specify minimum tensile strength require-
ments of the substrate which must be tested and verified prior to Yuan and Wu [16]
qffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffi
application of CFRP to the concrete surface. The strength of the
substrate is therefore of paramount importance in the develop- Pu ¼ 0:427bp bL fc0 Le ; Le ¼ Ep t p = fc0 ð2Þ
ment of any new anchorage system and will directly influence
the bond strength capacity of CFRP/concrete joints. where,
  ( )
2  ðbp =bc Þ 1 L  Le
2.2. Proposed anchorage solution bp ¼ ; bL ¼   ð3Þ
pL
sin 2L L < Le
1 þ ðbp =bc Þ e

The efficiency of any anchorage system can be enhanced

through an improvement of the mechanical properties of the con- Chen and Teng [6].
crete substrate over the length of the anchorage resulting in a With reference to Eqs. (1)–(3) shown above: bc is the concrete
higher FRP-to-concrete bond strength. In practice, where the pull section width (mm), bp is the width of the bonded FRP plate
off strength of the concrete substrate is less that the minimum (mm), Ep is the modulus of elasticity of the bonded FRP plate
requirements, it is sometimes possible to improve the pull off (MPa), Gf is the fracture energy (MPa), fc0 is the compressive
bond by impregnating the concrete with a very low viscosity re- strength of the concrete (MPa), L is the bonded length (mm), Le is
sin. However improvement by this method will only occur when the effective bond length (mm), Pu is the bond strength of a joint
the substrate is porous [2]. This method also remains to be (N), tp is thickness of the bonded FRP plate (mm), tc is thickness
quantified. of the concrete prism (mm), bL is a geometric bond length coeffi-
The proposed solution relies on the introduction of a mechani- cient, bP is a geometric width coefficient.
cal chase cut into the concrete over the length of the end anchorage Current bond slip models proposed by researchers stipulate that
zone. The chase is to be filled with epoxy resin, prior to bonding the the width ratio’s of the CFRP to concrete prism (bp/bc) can have
FRP laminate over the prepared surface (refer Fig. 2). The purpose some influence on the ultimate strength of a specimen
of the chase is to prevent the critical mode of de-bond which nat- [12,16,17]. Based on proposed models by Chen and Teng [6], it is
urally occurs a few millimetres beneath the concrete/adhesive anticipated that a 10% reduction in bond strength would be in-
interface. It utilises the superior mechanical properties of the curred as a result of adopting a prism width of 300 mm as opposed
epoxy to distribute the stresses over a larger area and depth within to 500 mm.
the concrete prism.
The proposed concept has wide application but was developed 3.2. Test preparation and material properties
specifically combined shear/torsional strengthening of box girder
bridge webs to be utilised at the web/flange connection interface. Reinforced concrete blocks of dimension 250 mm  300 mm 
In addition to the proposed concrete chase a N24 reinforcement 600 mm (Fig. 2) were used for the construction of three test spec-
bar was installed within the chase to be embedded into the under- imens. A control specimen formed the basis for comparison to-
side of the bridge deck to anchor the forces from the ends of the gether with two anchorage specimens which formed the first
vertical laminates on the outside webs into the underside of the stages of the experimental program. Blocks were reinforced nomi-
bridge deck. The purpose of the bar was to augment the amount nally with N16-200 each face to replicate the existing reinforce-
of steel reinforcement in the web-flange joint. The latter was found ment present in the webs of the bridge.
to be inadequate for the increased loading on the bridge (refer The control test consisted of one No. 120 mm  2 mm 
Fig. 2). Due to the method of CFRP de-lamination observed later 1000 mm laminate strip bonded to the surface of the concrete
in the testing is it stipulated that this reinforced bar does not ac- block with a bond length of 500 mm. The face of the concrete block
tively contribute to the enhancement of the CFRP laminate anchor- was cleaned, sandblasted to achieve a profile similar to 60 grit sand
age. It is expected that the omission of this bar in future testing will paper prior to surface preparation and application of the laminate
not degrade the improvement in FRP-to-concrete bond strength strip. This was followed by curing which occurred in a temperature
observed. controlled chamber of 50 °C for a period of 48 h, then further
2740 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746

curing at room temperature (22 °C) for a further 72 h prior to test- Table 2
ing. Adhesion tests were carried out on additional adhesion sam- Adhesives and saturant properties data.

ples to verify laminate adhesion prior to testing. Properties MBRACE laminate Units
A 40 mm  40 mm  500 mm chase was cut into the 300 mm Tensile strength 3300 MPa
wide side of the concrete block for the anchorage specimens. A pri- Tensile modulus 210 GPa
mer coating was applied to entire surface of concrete block (includ- Ult. elongation 0.014 Strain
ing chase) prior to any bonding. An N24 deformed reinforcement Density 1.56 gm/cm3
Thickness 2 mm
bar was then bonded into the chase using laminate adhesive. In Width 120 mm
addition, glass fibre fabric (120 mm  400 mm) was applied cen-
trally over reinforcement bar (to prevent galvanic corrosion of
the reinforcement).
Laminate adhesive was then applied to the underside of the plate. This prevented any forward movement of the concrete spec-
laminate strip and centrally on the prepared concrete surface. imens that may occur during loading.
Adhesive was applied to both surfaces using a profiled template Once the specimens were centrally located within the testing
to ensure accurate application (peak thickness of 1.5 mm in centre rig, a rigorous cross checking program was implemented to ensure
of adhesive strip). A 120 mm  2 mm  1000 mm (210 GPa) lami- the verticality of the test laminate strip. A spirit level was used to
nate strip was applied centrally to the concrete block. check the verticality of the laminate strip, with shims being used to
The laminate strip was pressed down onto the concrete block create a vertical test specimen. This procedure was cross checked
using a special profiled tool to ensure accurate thickness of adhe- by two independent people.
sive between concrete surface and laminate strip and central
placement of laminate. Excess adhesive was cleaned up from the 3.4. Instrumentation and loading procedure
concrete and laminate surface. Specimens cured in a manner sim-
ilar to the control specimen. All Samples were tested in a Baldwin Strain and load results were obtained from surface mounted
universal testing machine. Tables 1 and 2 below show the relevant strain gauges and a 3D non-contact measuring technique based
adhesive and CFRP material used in the experimental program. on image correlation photogrammetry [8].
A series of seven strain gauges, from G1 to G7, were attached to
3.3. Experimental set-up the surfaces of CFRP plates (refer Fig. 2). G1 and G2 were installed
to monitor any bending in the CFRP plate during testing indicating
Many alternative experimental set-ups have been used by the presence of tilting. G1 was placed at the back of the laminate
researchers for determining the FRP-to-concrete bond strength. and G2 at the front at the same location. The specimens were
Of these far end supported double shear tests and near end sup- tested under displacement control of 0.0167 mm/s until beyond
ported single shear tests are most popular due to their simplicity de-bonding of the CFRP from the concrete specimen.
[4,15]. In crack-induced de-bonding failures, the stress state in The 3D photogrammetry measurements were taken using a pair
the critical region of a beam is closely similar to that of a concrete of high resolution, digital CCD cameras. A measuring step of 10 s
prism in a near end supported (NES) single shear pull test and the was used between recording intervals. 3D image correlation
latter serves as a promising candidate for a standard set-up for software analyses the deformation of a random or regular pattern
determining the FRP-to-concrete bond strength [4,15]. On this ba- pixels with good contrast which is applied to the surface of the
sis the experimental design used in this study was based on the specimen and recorded by the CCD cameras for processing.
NES single pull test configuration. In tables and figures which follow reference is made to AR (pho-
A test rig was constructed to ensure each specimen was able to togrammetry) and SG (strain gauge). These refer to the two data
be securely fixed to a Baldwin Universal testing machine (refer acquisition techniques used in the experimental programme.
Fig. 1). The test rig was bolted down to the moving lower platform
of the testing machine which clamped the specimen into place. The 3.5. Experimental results
test rig was constructed using 30 mm thick steel plates. A back
plate, 600 mm high was welded at the rear of the test rig with nine 3.5.1. QA tests
No. M12 bolts placed across the face of the rear plate to prevent A total of six concrete cylinders were tested to assess the con-
any movement of the concrete specimens during loading. After crete compressive strength. After 53 days curing at room tempera-
the initial series of tests an additional steel plate was place at the ture, the average compressive strength of the concrete was 62 MPa.
font the specimens (at the base) and bolted to the rear vertical Pull off tests conducted prior to testing of the specimens indi-
cated that laminate bond failure also occurred within the concrete
at a bond pressure of 3.6 MPa. This suggested full bonding between
Table 1 laminate strip and concrete surface.
CFRP properties data.

Properties MBRACE laminate MBRACE Units

3.5.2. Failure modes
adhesive primer 3.5.2.1. Control specimen. This specimen failed by separation of the
composite plate from the concrete block with a layer of concrete
Resin type Epoxy Epoxy –
Specific gravity 1.8 1.08 – still bonded to the plate, as shown in Fig. 3. This is the typical mode
Glass transition temperature >65 – °C of failure observed by Maeda et al. and Van Gemert [10,14] where
Modulus of elasticity 10 0.7 GPa failure occurs a few millimetres beneath the concrete surface.
Lap shear strength to CFRP >17 – MPa
Bond (to concrete) >3.5 >3.5 MPa
Tensile strength 32 >12 MPa
3.5.2.2. Anchorage specimen. As Fig. 4 shows, the majority of the
Compressive strength >60 – MPa CFRP plate was left exposed, with no concrete or epoxy bonded
Flexural strength >35 >24 MPa to it.
Full cure at However, traces of adhesive did remain in the surface of the
25 °C 7 0.208 Days
composite plate in localised areas. The adhesive mostly remained
40 °C 3 0.125 Days
attached to the concrete during the plate de-lamination failure.
 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746 2741

Fig. 1. Specimen testing rig details (a) configuration of test rig (front view); (b) configuration of test rig (side view).

3.6. Tilt 3.6.1. CFRP strain distributions

CFRP elongation along the length of the laminate are reported in
In practical pull tests, there may be a small unintended offset d Fig. 6a and b for both control and anchored specimens under differ-
in the position of the load [15]. The result of any eccentricity in ent loading conditions.
load application can result in a localised bending effect at the top Photogrammetry data indicate a drop in strain level at approx-
of the specimen and the likely hood of premature de-lamination. imately 50 mm from the loaded face for both specimens. This can
Detection and monitoring of any eccentricity has been considered only be observed from the photogrammetry results as no strain
in test measuring and instrumentation through the installation of gauge was positioned at this location for comparison. The control
strain gauges G1 and G2 at the front and back of the laminate. specimen shows a lower drop in strain which could be due to
The degree of tilting can be determined from the variation in the observed tilt and localised bending effect at the edge of the
strains between these two gauges. As shown in Fig. 5a the control loaded face. The strain of the control specimen shows a gradual
specimen has shown some deviation in strain between gauges G1 flattening towards the onset of de-bonding. This indicates a grad-
and G2 indicating the presence of tilting. Since G2 shows a higher ual and progressive de-bonding failure from the edge of the loaded
strain than G1 the bending is expected to produce push/pull (com- face to furthest end from the free edge of the specimen.
pressive/tensile peeling stresses) along the length of the concrete Figs. 7–9 all show good correlation between photogrammetry
block. and strain gauge measurements. Progressive de-bonding is evident
The effects of eccentricity for future testing has been mitigated in both specimens. Observed strain values at gauge locations are
by using key clamping devices in the test set-up (refer Fig. 1). Each seen to converge after partial de-bonding has occurred. This is evi-
specimen was tensioned to 25 kN to verify accuracy of specimen dent in Fig. 7.
mounting. Gauges G1 and G2 were compared to ensure variation Load vs. strain graph (Fig. 8) shows an initially linear strain load
between the gauges were within an acceptable tolerance (±10% response for gauges 5 and 6 with a gradual reduction in stiffness.
of each other), thus ensuring uniform tensioning of laminate dur- This is shown by a gradual flattening of the curves and sudden
ing testing. If readings were not within tolerance, the specimen strain increases which are indicative of partial de-bonding. The
was unloaded and re-aligned. same trend is also seen in Fig. 9.
2742 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746

Fig. 2. Anchorage type 1 specimen geometry (WG1 and WG2) (a) configuration of strain gauges; (b) chase details and installation of N24 reinforcement bar; (c) section
through chase.

Fig. 3. Failed control sample (WGB9) (a) complete de-bonding of laminate from concrete surface; (b) concrete surface post de-bonding of laminate; (c) de-bonded laminate
strip; (d) real time load, strain and ARAMIS recordings during testing phase.

Strain readings on gauges G8 and G9 (Fig. 2) were located on of mirco cracking during testing is believed to have a marked
either side of the steel reinforcement bar suspended under the test reduction of stiffness of the unstrengthened concrete
rig. It was observed that a negative strain was induced on one side substrate.
of the bar and a positive strain on the other. It was established that  Strain readings taken 150–250 mm from the loaded face
during testing, a rotational effect about the in plane axis was in- (gauges G5–G7) for the anchorage specimen indicate better load
duced on the specimen. As the steel bar was anchored to the test transfer and a higher strain being reached prior to partial de-
rig, the steel bar restrained the specimen from movement. (See bonding occurring. The control sample shows a sudden increase
Figs. 1 and 2). in strain which is evidence of de-bonding, with very little partial
A comparison of the curves for the control specimen and de-bond for gauges G5–G7.
anchorage specimen yields the following observations:  The anchorage specimen has shown significant improvement in
both the maximum load and strain reached prior to failure
 Load–strain relations for the anchored sample show a steeper (Refer Table 3).
slope which is indicative of a stiffer substrate compared to  The effect of the chase is a 95–100% increase in ultimate capac-
the control sample. Material properties indicate concrete to ity and an 83–93% increase in the maximum strain level
have a higher modulus than epoxy. However the presence reached prior to failure.
 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746 2743

Fig. 4. Testing of WGB1 (a) specimen ready for testing; (b) concrete rupture at adhesive–concrete interface; (c) de-bonded laminate strip.

Fig. 5. Load vs. strain distribution (gauges G1 and G2); (a) control specimen (WG9); (b) type 1 – anchorage specimen (WG1).

3.6.2. Experimental bond–slip curves (3) Linear variation of strains in FRP plate between two subse-
Strain measures along FRP plate/sheet at different loading levels quent strain gauges.
were used to calculate shear stress–slip data. Considering an elas-
tic behaviour for the composite, average shear stress values be- Ferracuti et al. [7].
tween two subsequent strain gauges (or any two points along The average slip is then calculated as the incremental sum of
the laminate) can be written as a function of the difference of mea- the CFRP extension. This is expressed in
sured strains, ei, ei+1. This equates to dividing the force difference
ef ;iþ1  ef ;i
between two points by the total area and is represented in Si ¼ DL þ si1 ð5Þ
2
Ef ðef ;iþ1  ef ;i Þt f
s¼ ð4Þ In general, the bond–slip curves have a non-linear ascending and
DL
descending trends. It was found that these trends can be approxi-
where Ef and tf are CFRP elastic modulus and thickness; ef,i+1 and ef,i mately described using Popovics’ equation given by:
are CFRP strains; and DL is the distance between strain gauges.
In order to define the slip distribution along the FRP plate, the s S a
¼ ð6Þ
following assumptions are made: smax S1 ða þ 1Þ þ ðS=S1 Þa
(1) Perfect bonding (no slip) between plate and concrete at last where: smax and s1 are the maximum bond stress and correspond-
strain gauge position. ing slip. The value (a) controls the slope of the ascending and
(2) Deformation of concrete specimen far from external cover is descending branches of the bond slip curve. A value of a = 3 was
negligible with respect to FRP counterpart. established by Nakaba et al. [11].
2744 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746

Fig. 6. Strain vs. distance along laminate; (a) control specimen (WG9); (b) type 1 – anchorage specimen (WG1).

Fig. 7. Load vs. strain distribution (gauges G3 and G4); (a) control specimen (WG9); (b) type 1 – anchorage specimen (WG1).

Fig. 8. Load vs. strain distribution (gauges G5 and G6); (a) control specimen (WG9); (b) type 1 – anchorage specimen (WG1).

Table 4 demonstrates the effect of the 40  40 mm concrete of 118% in bond stress is seen as a result of introducing the chase
chase on the maximum bond strength achieved. A strength gain anchor.
 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746 2745

Fig. 9. Load vs. strain distribution (gauges G7); (a) control specimen (WG9); (b) type 1 – anchorage specimen (WG1).

Table 3 using the two methods show an acceptable level of correlation.

Load/elongation results summary (WG1, WG2 and WG9). Photogrammetry measurements estimate +10% higher bond stres-
Reference Failure load (kN) Max elongation (le) ses and +17.2% higher slips recorded. The softening branches of the
GA/AR GA AR
two bond–slip curves follow comparable descending gradients,
with photogrammetry estimating a lower degree of softening
Control specimen WG9 99.6 2535 2706
resulting in a higher fracture energy and slip.
Stage 1 – anchorage WG1 194.4 4640 4434
WG2 198.5 4881 4733 Although this difference in post peak slip is not shown in
Fig. 10b, the strain behaviour after the onset of de-bonding can
be difficult to measure using photogrammetry. The photogrammet-
ric measurements generally require filtering to smooth out irregu-
Table 4 larity and noise in the raw data. This irregularity can increase after
Max bond stress and corresponding slip results summary (WG1, WG2 and WG9). the onset of de-bonding occurs. It is therefore recommended that
Reference Max bond stress Corresponding slip where a photogrammetry data acquisition system is used, it always
(MPa) (so) be verified with strain gauge data for experiments of this nature.
GA AR GA AR Bond slip correlations for anchorage type 1 specimens demon-
Control specimen WG9 5.0 5.5 0.058 0.10
strate a similar trend of bond–slip curves for both photogrammetry
Stage 1 – anchorage WG1 11.3 11.0 0.20 0.20 and strain gauge data. It can be verified that curves are very similar
WG2 17.5 10.9 0.20 0.23 in terms of peak shear stress and corresponding slip.

4. Effective strain in CFRP laminates used in design

Fig. 10a presents the bond–slip curves obtained from both
strain gauge and photogrammetry measurement techniques. A The effective strain of a CFRP laminate is a governing factor in
Popovics trend line has been fitted to each curve using the mea- the design of FRP systems. It is the maximum strain that can be
sured peak bond stress and corresponding slip obtained from each achieved in the FRP system prior to failure considering all possible
data acquisition technique. The peak bond stresses determined failure modes. ACI 440.2R-02 [1] states that for bonded U-wraps or

Fig. 10. Bond–slip curves (a) control specimen (WG9) with fitted curve following Popovics equation; (b) type 1 – anchorage specimen (WG1).
2746 R. Kalfat, R. Al-Mahaidi / Composite Structures 92 (2010) 2738–2746

Table 5
Maximum FRP elongations and corresponding effective FRP strains and utilisation percentiles.

Specimen Max elongation (le) Effective FRP strain, ei0 ACI440.2R-08 (le) FRP rupture strain (le) Fibre utilisation
GA AR
Control specimen WG9 2535 2706 1890 14,000 18–19%
Stage 1 – anchorage WG1 4640 4434 NA 14,000 31–33%
WG2 4881 4733 NA 14.000 33–35%

bonded face plies; the maximum strain in the FRP which may be Acknowledgements
used in design can be expressed in the following equation:
efe ¼ kv efu  0:004 ð7Þ The authors acknowledge the financial support provided by the
Westgate Bridge Strengthening Alliance and the services provided
where kv is a bond reduction coefficient applicable to shear and efu
by the Department of Civil Engineering at Monash University.
is the ultimate strain of the CFRP plate at rupture.
When this is computed using current specimen material prop-
erties (control specimen) the effective CFRP strain efe is calculated References
as: 0.00189 which is 75% of the observed maximum elongation re-
corded. This value is considered acceptable for use in design with a [1] ACI 440.2R-02. Guide for the design and construction of externally bonded FRP
factor of safety of 0.75. Table 5 compares the effective CFRP design systems for strengthening concrete structures. American Concrete Institute,
Farmington Hills, Michigan; 2002.
strain with the maximum elongations recorded in experimental [2] BBR ISO 9001. A guide to the design and application of BBR FRP strengthening
testing. No prediction model currently exists in literature for the systems. BBR Systems Ltd., Bahnstrasse 23, Poostfach, CH-8603
anchorage solution which was tested. Schwerzenbach, Switzerland, No. 1. p. 5.1.
[3] Bizindavyi L, Neale KW. Transfer lengths and bond strengths for composites
bonded to concrete. J Compos Constr 1999;3(4):153–60.
5. Conclusions [4] Camata G, Spacone E, Al-Mahaidi R, Saouma V. Analysis of test specimens for
cohesive near-bond failure of fiber-reinforced polymer-plated concrete. J
This paper presented the results of anchorage bond strength of Compos Constr 2004;8(6):528–38.
[5] Chajes MJ, Finch Jr WW, Januszka TF. Bond and force transfer of
CFRP laminates bonded to concrete with modified substrate condi- composites materials plates bonded to concrete. ACI Struct J
tions. The development of an effective anchorage solution for the 1996;93(2):209–17.
improvement of overall substrate and bond strength properties [6] Chen JF, Teng JG. Anchorage strength models for FRP and steel plates bonded to
concrete. J Struct Eng 2001;127(7):784–91.
has been presented using an experimental study. [7] Ferracuti B, Savoia M, Mazzotti C. Interface law for FRP–concrete delamination.
The results and discussions presented allow the following con- Compos Struct 2007;80(4):523–31.
clusions to be made: [8] GOM mbH. ARAMIS v5.4.1. Gom optical measuring techniques; 2005.
[9] Hii AKY, Al-Mahaidi R. Experimental investigation on torsional behavior of
solid and box-section RC beams strengthened with CFRP using
 Design guidelines can limit the CFRP material strains to levels as photogrammetry. J Compos Constr 2006;10(4):321–9.
low as 10–25% of the ultimate material strain at rupture. [10] Maeda T, Asano Y, Sato Y, Ueda T, Kakuta Y. A study on bond mechanism
of carbon fiber sheet. Non-metallic (FRP) reinforcement for concrete
 By anchoring the ends of CFRP laminates or sheets it is possible struct. In: Proc 3rd int symp, Japan Concrete Institute, Sapporo 1. p. 279–
to achieve higher bond strength capacities and degrees of util- 85.
isation of the CFRP material resulting in increased economy. [11] Nakaba K, Kanakubo T, Furuta T, Yoshizawa H. Bond behaviour between fibre-
reinforced polymer laminates and concrete. ACI Struct J 2001:1–9.
 The strength of the concrete substrate is a key factor affecting
[12] Pham HB, Al-Mahaidi R. Prediction models for debonding failure loads of
the de-lamination mode and overall bond strength. carbon fiber reinforced polymer retrofitted reinforced concrete beams. J
 The introduction of a mechanical chase cut into the concrete over Compos Constr 2006;10(1):48–59.
the anchorage length is an effective way to improve the strength [13] Täljsten B. Plate bonding: strengthening of existing concrete structures with
epoxy bonded plates of steel or fibre reinforced plastics. Doctoral thesis, Lulea
of the concrete substrate, resulting in higher CFRP maximum University of Technology, Sweden; 1994.
elongations, bond stress, slip and load carrying capacities [14] Van Gemert D. Force transfer in epoxy-bonded steel–concrete joints. Int J
 The effect of the chase is a 95–100% increase in ultimate capac- Adhes Adhes 1980;1:67–72.
[15] Yao J, Teng JG, Chen JF. Experimental study on FRP-to-concrete bonded joints.
ity, 118% increase in bond stress and 83–93% increase in the Article; 2005.
maximum strain level reached prior to failure. [16] Yuan H, Wu Z. Interfacial fracture theory in structures strengthened with
 3D non-contact measuring technique based on image correla- composite of continuous fiber. In: Proc symp of China and Japan: sci and
technol of 21st century, Tokyo 1; 1999. p. 142–55.
tion photogrammetry is a viable measurement technology [17] Yuan H, Wu ZS, Yoshizawa H. Theoretical solutions on interfacial stress
which results in good correlation with measurements obtained transfer of externally bonded steel/composite laminates. JSCE J Struct Mech
using discrete strain gauges. Earthq Eng 2001;675(I-55):27–39.