You are on page 1of 9

Pullout Strength Models for FRP Anchors

in Uncracked Concrete
Seo Jin Kim1 and Scott T. Smith, M.ASCE2

Abstract: Mechanical anchorage can delay or even prevent premature debonding failure in externally bonded fiber-reinforced polymer
共FRP兲 composite strengthening systems. A promising type of anchor made from FRP, which is known as a FRP spike anchor or FRP
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

anchor among other names, is noncorrosive and can be applied to a wide range of structural elements and externally bonded FRP
strengthening schemes. Experimental investigations have shown FRP anchors to be effective under tension 共pullout兲 and shear loading,
however, few analytical models exist to date. This paper in turn presents analytical models to quantify the pullout strength of FRP anchors.
As existing research on the pullout behavior of metallic anchors is partially relevant to FRP anchors, this paper first presents a review of
current pullout strength models for metallic anchors. These models are then assessed with experimental data of FRP anchors and modified
and recalibrated where appropriate. As a result, simple and rational pullout strength models for FRP anchors are proposed which can also
be used in design. Finally, parametric studies are undertaken and the influence of key variables is identified.
DOI: 10.1061/共ASCE兲CC.1943-5614.0000097
CE Database subject headings: Anchors; Concrete; Fiber reinforced polymers; Strength; Pullout.
Author keywords: Anchors; Concrete; Fiber reinforced polymers; Strength; Pullout.

Introduction mounted rods 共e.g., Gose and Nanni 2000兲, and anchors made
using FRP 共also known as FRP spike anchors but herein referred
Extensive research over the past one and a half decades or so has to as FRP anchors兲 共e.g., Lam and Teng 2001; Orton et al. 2008;
established the effectiveness of strengthening and repairing Kim and Smith 2009兲. FRP anchors are noncorrosive and can be
共herein referred to as strengthening兲 existing reinforced concrete applied to a wide variety of structural shapes such as beams,
共RC兲 structural members with externally bonded fiber-reinforced slabs, and walls.
polymer 共FRP兲 composites 共Hollaway and Teng 2008兲. A com- Fig. 1 demonstrates possible uses of FRP anchors and their
monly reported failure mode of the FRP strengthening is prema- resulting loadings. In this case, FRP anchors are installed to se-
ture separation 共i.e., debonding兲 of the FRP from the concrete cure the free ends of FRP U jackets when used as shear strength-
substrate which originates at an intermediate crack in the member ening for RC flexural members. Depending on the orientation of
the FRP anchor, it can be subjected to predominantly pullout
and propagates along the FRP-to-concrete interface in the con-
forces 共anchor Type A in Fig. 1兲 or shear forces 共anchor Type B in
crete 关i.e., intermediate crack induced debonding, which is also
Fig. 1 which will include a pullout component兲. FRP anchors are
referred to as IC debonding 共Teng et al. 2003; Smith and Gravina
formed from bundles of glass or carbon fibers or rolled fiber
2007兲兴. IC debonding can occur in a relatively sudden manner and
sheets which are then inserted into epoxy filled holes in the con-
can occur well below the ultimate tensile strain capacity of the
crete substrate 共anchor dowel component in Fig. 1兲. The free end
FRP. Such debonding must therefore be controlled and the
of the fiber is splayed 共anchor fan component in Fig. 1兲, then
strengthened member made to fail in the more conventional epoxied onto the surface of the FRP strengthening plate or sheet
modes of concrete crushing or FRP rupture. 共herein FRP plate兲 in order to disperse stresses around the FRP
In order to prevent or delay debonding failures, several differ- anchor and immediate FRP plate vicinity.
ent types of anchorage systems have been investigated to date. To date, FRP anchors have been applied to various structural
They include, but are not limited to, embedded metallic 共e.g., shapes and materials such as RC beams 共e.g., Oh and Sim 2004;
Sharif et al. 1994兲 and FRP threaded rods 共e.g., Galati et al. Eshwar et al. 2005; Orton et al. 2008兲, RC slabs 共e.g., Lam and
2007兲, U jackets 共e.g., Smith and Teng 2003兲, near surface Teng 2001; Seliem et al. 2008兲, and masonry walls 共e.g., Tan and
Patoary 2004; Binici et al. 2007兲. Research on the strength and
1
Senior Research Assistant, Dept. of Civil Engineering, Univ. of Hong behavior of FRP anchors in isolation has however been much
Kong, Pokfulam Rd., China; formerly, Ph.D. Candidate, School of Civil more limited and such research has been predominantly experi-
and Environmental Engineering, Univ. of Technology Sydney, Broadway, mental. Studies to date have reported the pullout strength and
Sydney, New South Wales, Australia. E-mail: skim@hku.hk behavior 共Özdemir 2005; Kim and Smith 2009; Ozbakkaloglu
2
Assistant Professor, Dept. of Civil Engineering, Univ. of Hong Kong, and Saatcioglu 2009兲 as well as the shear strength and behavior
Pokfulam Rd., China 共corresponding author兲. E-mail: stsmith@hku.hk
共e.g., Aiello et al. 2004; Smith and Kim 2010兲 of FRP anchors.
Note. This manuscript was submitted on January 23, 2009; approved
on November 22, 2009; published online on February 3, 2010. Discussion FRP anchors subjected to pullout forces 共Type A anchors in
period open until January 1, 2011; separate discussions must be submitted Fig. 1兲 will form the focus of this paper and from herein will only
for individual papers. This paper is part of the Journal of Composites for be discussed.
Construction, Vol. 14, No. 4, August 1, 2010. ©ASCE, ISSN 1090-0268/ In this paper, a review of the behavior of adhesive metallic and
2010/4-406–414/$25.00. FRP anchors is first presented followed by a review of existing

406 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010

J. Compos. Constr. 2010.14:406-414.


hef

hef

hef

hef

hef
'A' 'B'
'A' 'B'
(a) Concrete cone (b) Adhesive-to- (c) Anchor-to- (d) Mixed interface (e) Anchor failure
failure concrete adhesive interface failure
interface failure failure
Section Elevation (Combined failure) (Combined failure) (Combined failure)

(a) FRP anchor types and applications Fig. 2. Typical adhesive anchor failure modes
Nu

关Fig. 2共d兲兴. It is however usually difficult to clearly distinguish


Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Anchor Anchor between the various failure surfaces. In the case of adhesive an-
fan fan chors, the thickness of adhesive layer does not appear to have a
significant influence on the pullout strength of the anchor 共Cook
Anchor Nu
Anchor et al. 1998兲. The remaining failure mode is anchor rupture which
dowel 'A' can occur in anchors with large embedment depths and sufficient
dowel 'B'
bond strength 关Fig. 2共e兲兴. In the case of concrete cone or bond
(b) FRP Anchor types and dowel stress distribution failure, the anchor will generally not have yielded and the anchor
can be treated as essentially elastic. In the case of anchor tensile
Fig. 1. FRP anchors in U-jacket shear strengthening failure which takes place outside of the anchor hole, the tensile
strength rather than the yield strength of the anchor is considered
as the ultimate strength as metal anchors generally do not exhibit
analytical models for both types of anchors. Such a review ex- a well defined yield point 关ACI 349-01; American Concrete In-
plains the fundamental concepts of anchor pullout behavior and stitute 共ACI兲 2001兴.
modeling and sets the scene for the new analytical models pre- In the case of adhesive metal anchors, load is transferred from
sented in this paper for describing the pullout behavior and the anchor, through the adhesive, then into the concrete along the
strength of FRP anchors. The new analytical model is calibrated entire bond surface area 共Cook et al. 1998兲. The bond strength of
with a database of FRP anchor tests under tensile 共pullout兲 load adhesive anchors is derived mostly from chemical adhesion and
and then a design version is proposed. Parametric studies are once the adhesive bond is broken, force transfer is provided by
finally conducted in which the importance of key variables is friction. Such force transfer is strongly affected by transverse
made evident. pressure, concrete shrinkage, and roughness of the anchor surface.
In addition, interfacial wear and tear along the sliding plane can
reduce the confining pressure 共FIB 2000兲.
Review: Pullout Behavior of Metallic
and FRP Anchors FRP Anchors
Metallic or nonmetallic anchors are distinguished by their load- FRP anchors are typically drilled-in 共postinstalled兲 anchors with a
transfer mechanism 共i.e., mechanical interlock, friction or chemi- chemical bond load-transfer mechanism. Considerably more re-
cal bond, or various combinations兲 as well as the method of search has been undertaken on the pullout strength and behavior
installation 共i.e., cast in place, drilled-in anchors, or pneumatically of cast-in place FRP reinforcing bars than adhesive 共postinstalled兲
installed兲 共Eligehausen et al. 2006a兲. FRP reinforcing bars and FRP anchors. Research on cast-in place
FRP reinforcing bars has strived to develop bond-slip and bond
strength models which can be used for the design of internal
Metallic Anchors
reinforcement for concrete flexural members 共e.g., Okelo and
Metallic anchors can be cast-in place 共e.g., headed anchors兲 or Yuan 2005兲. Such tests are considered outside the scope of the
postinstalled 共e.g., adhesive anchors and expansion anchors兲 and a paper and therefore bear no relevance due to surface deformation
comprehensive review of the behavior and strength of such an- on the bars as well as the lack of an adhesive layer causing fun-
chors is given in Eligehausen et al. 共2006a兲. Typical failure modes damental differences in behavior with adhesive FRP anchors.
of these metallic anchors are shown in Fig. 2 共Cook et al. 1998兲. When considering adhesive postinstalled FRP reinforcing bar an-
As shown in Fig. 2共a兲, a concrete cone shaped failure surface chors, the bond failure modes shown schematically in Figs.
commonly occurs at shallow embedment depths 共hef = 3d ⬃ 5d, 2共b–d兲 have been observed as well as the anchor rupture failure
where d is the anchor diameter兲 with the slope of the failure mode shown in Fig. 2共e兲. The main variables of concern of such
face with respect to the surface of the concrete being approxi- testing have been 共1兲 surface features 共and texture兲 of the FRP
mately 35° 共Cook 1999; Eligehausen et al. 2006a兲. For longer bar; 共2兲 concrete compressive strength; 共3兲 concrete cover thick-
embedment depths, a concrete cone failure surface usually ap- ness; and 共4兲 embedment depth 共e.g., Ahmed et al. 2008兲. Such
pears adjacent to the face of the concrete with the remaining experimental results will however not be used in this paper due to
portion of the anchor failing at the bonded interface 关Figs. fundamental differences in fabrication quality of FRP reinforcing
2共b–d兲兴. Bond failure is defined by the interfacial failure surface bars to FRP anchors. In addition, it appears that no analytical
position; namely 共1兲 adhesive-to-concrete interface 关Fig. 2共b兲兴; models to date have been proposed for the design of postinstalled
共2兲 anchor-to-adhesive interface 关Fig. 2共c兲兴; and 共3兲 mixed in- adhesive FRP reinforcing bar anchors.
terface 共adhesive-to-concrete and anchor-to-adhesive interface兲 Research on the pullout strength and behavior of FRP anchors

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010 / 407

J. Compos. Constr. 2010.14:406-414.


has been primarily limited to the studies of Özdemir 共2005兲, Kim Nc = khnef 冑 f ⬘c 共1兲
and Smith 共2009兲, and Ozbakkaloglu and Saatcioglu 共2009兲. The
main parameters varied in these studies were 共1兲 anchor diameter; where Nc = concrete cone pullout resistance; k = empirical multi-
共2兲 anchor embedment depth; 共3兲 anchor fiber content; and 共4兲 plier which largely depends on the anchor type 共see Table 1兲;
concrete compressive strength. All three studies reported experi- f ⬘c = concrete cylinder compressive strength; and the effective em-
mental results, while the first and third studies reported analytical bedment depth hef generally refers to the length of the anchor of
models which were largely based on traditional pullout strength constant diameter 共i.e., does not include the headed portion or
models of adhesive metallic anchors. Pullout tests on FRP an- tapered end of a headed anchor兲. The power n is given in Table 1
chors have revealed similar failure modes to the modes presented as either 1.5 or 2. The concrete cone model adopts n = 2 which is
in Fig. 2 for adhesive metallic anchors. Özdemir 共2005兲 reported derived directly from the area of the circular projected failure
anchors failing by rupture of the FRP, concrete cone failure, as surface 共i.e., the radius of the failure surface is hef 兲. The CCD
well as combined concrete cone and bond failure. The first series models adopt n = 1.5 which is derived from h2ef times h−0.5
ef , where
of tests conducted by Özdemir 共2005兲 mostly failed by FRP rup- h2ef is the area of the square projected failure surface and h−0.5ef
ture, while the second series test, with improved anchor construc- is a size effect factor 共Eligehausen et al. 1992; Cook 1999兲. In
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

tion, reported all three failure modes. Ozbakkaloglu and general, CCD models correlate well for a wide range of anchor
Saatcioglu 共2009兲 reported FRP anchor rupture failure, concrete embedment depths, while the concrete cone models can under-
cone failure, combined concrete cone and bond failure, and con- estimate the pullout strength of anchors with small embedment
crete splitting failure. Kim and Smith 共2009兲 in turn observed the depths 共Fuchs et al. 1995兲. Initially the concrete cone model was
same failure modes as reported by Özdemir 共2005兲 and recommended 共e.g., ACI 349-85: ACI 1985兲 however in recent
Ozbakkaloglu and Saatcioglu 共2009兲. FRP anchors therefore pos- years the CCD model is more widely referred to 关e.g., ACI
sess many similar characteristics to metallic anchors although 349-01 共ACI 2001兲; ACI 318-05 共ACI 2005兲兴.
some fundamental differences 关e.g., FRP material properties 共FIB
2000兲兴 exist as well. In particular, the bond strength between the
anchor and surrounding concrete face may be reduced by a Pois- Bond Models
son’s ratio effect in which a larger anchor diameter can lead to Bonded interface resistance models assume that the failure sur-
greater lateral contraction under tensile load and hence a reduced face is adjacent to the surface of the embedded portion of the
frictional and mechanical interlocking strength between the an- anchor at the anchor-to-concrete interface. The most important
chor and concrete hole wall 共Ozbakkaloglu and Saatcioglu 2009兲. parameters are the bond stress of the anchor-to-concrete interface
共which is usually assumed to be uniform兲 and the failure surface
area. Table 1共b兲 provides a summary of the commonly referred to
Review: Analytical Pullout Models for Metallic bond resistance models for adhesive metallic anchors. Such an-
and FRP Anchors chors can consist of threaded rods and reinforcing bars; however,
to date no models appear to have been proposed for FRP anchors.
Pullout strength models, predominantly developed for metallic Both anchor diameter as well as anchor hole diameter can be used
anchors, can be classified as 共1兲 concrete cone models; 共2兲 bond for determining the bond resistance, however, the use of anchor
models; 共3兲 combined concrete cone and bond models; and 共4兲 diameter in design is more favorable as its dimensions are easier
anchor rupture models. to control 共Cook et al. 1998兲. A uniform bond stress is typically
assumed where the stress is evenly distributed along the embed-
ded length of the anchor as shown in Fig. 1共b兲. Shear lag may
Concrete Cone Models
cause the longitudinal stress to vary throughout the thickness of
When anchors pull out of the concrete, the failure surface can the anchor; however, such a phenomenon is not considered in the
form in the shape of a concrete cone or similar shape with the two analytical models developed to date and will not be considered in
main categories of models being 共1兲 concrete cone and 共2兲 con- this paper.
crete capacity design 共CCD兲 models. Concrete cone models 共e.g.,
Cook 1993; CEB 1994兲 assume a projected circular area on the
Combined Concrete Cone and Bond Models
concrete surface based on a failure cone with a side inclination of
45° to the concrete surface. CCD models 共e.g., CEB 1994; Fuchs Such models assume a cone failure surface to form near the face
et al. 1995; ACI 2001, 2005; Eligehausen et al. 2006b兲 assume a of the concrete block with which the anchor is cast into with
projected square area on the concrete surface with side dimen- failure in the remaining portion occurring at the anchor-to-
sions of three times the anchor embedment depth 共i.e., 1.5 times concrete interface. The pullout resistance is therefore the addition
the embedment depth is the distance from the anchor to each side of the concrete cone strength contribution and the bonded inter-
of the square failure area兲 and a failure surface within the con- face strength contribution. Table 1共c兲 presents three models in
crete of approximately 35° to the concrete surface. which the first is for a bonded metallic anchor 共Cook 1993兲 and
Table 1共a兲 presents a summary of several commonly referred the remaining two for FRP anchors 共Özdemir 2005; Ozbakkaloglu
to concrete cone and CCD models for metallic anchors 共to date, and Saatcioglu 2009兲. In these three models an accurate estima-
no models appear to have been proposed for FRP anchors兲. The tion of the concrete cone depth is crucial; however, Cook 共1993兲
concrete cone failure model, developed in the middle 1970s, was reported the concrete cone resistance component to provide a
largely based on experimental results of cast-in-place 共headed兲 minimal contribution to the pullout resistance in particular for
anchors 关e.g., ACI 349-85 as cited in CEB 共1994兲兴 while the CCD anchors with deep embedment depths. Of particular interest to
models developed in the early 1990s were based on theoretical note in the model of Cook 共1993兲 in Table 1共c兲 is the assumption
and experimental models on headed anchors in the 1980s by of a nonlinearly distributed shear stress for the bond model com-
Eligehausen and Sawade 共1989兲 and Eligehausen et al. 共1992兲. All ponent: such a distribution is also referred to as an elastic bond
models adopt the general form given in Eq. 共1兲 stress model 共Cook 1993; Cook et al. 1993兲. The elastic bond

408 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010

J. Compos. Constr. 2010.14:406-414.


Table 1. Summary of Existing Analytical Models
共a兲 Concrete cone models
Projected failure area/
Reference Ncc failure angle k Anchor type

0.33冑 f c⬘Ac, Ac = ␲hef 共hef + dh兲


a
ACI 349-85 Circle/45° — Cast in situ anchor 共headed兲
0.33冑 f c⬘Ac, Ac = ␲h2ef Circle/45° — Postinstalled anchor 共expansion兲
Eligehausen et al. 共1987兲; 冑
ef f c⬘
kh1.5
c
⬇35° 17 共13b兲 Cast in situ anchor 共headed兲
Eligehausen and Fuchs 共1988兲a
Eligehausen et al. 共1987兲a 冑
ef f c⬘
kh1.5
c
⬇35° 15 Postinstalled anchor 共expansion兲
0.92h2ef 冑 f c⬘
a
Eligehausen et al. 共1984兲 Circle/45° — Postinstalled anchor
共adhesive anchor兲
Eligehausen et al. 共2006b兲 kh1.5冑
ef f c⬘
Square/⬇35° 14.7 Postinstalled anchor
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

共expansion and undercut兲


16.8 Cast-in anchor 共headed兲
ACI 349-01 共ACI 2001兲 and kh1.5
ef
冑 f c⬘ Square/⬇35° 7b Postinstalled anchor
ACI 318共M兲-05 共ACI 2005兲 共expansion and undercut兲
10b Cast-in anchor 共headed兲
共b兲 Bond models
Reference Ncb ␶ or ␶⬘ 共MPa兲 Comments
Cook 共1993兲 ␶␲d0hef 6.3–14.6 Tested values
CEB 共1994兲 ␶␲d0hef 8 Not specified
Cook et al. 共1998兲 ␶⬘␲dhef ⌿c ␶⬘f =20 MPa 共1 − kCoV兲 For design
c
Cook and Kunz 共2001兲 ␶␲dhef 15.4a Tested value
12b Lower bound 共assumed for design兲
共c兲 Combined concrete cone and bond models
Reference Anchor depth Ncc component Ncb component
Cook et al. 共1993兲 hef ⬍ hc 0.92h2ef 冑 f c⬘ —

hc ⬍ hef ⬍ 40冑d0 + hc 0.92hc2冑 f c⬘ 冋 40冑d0 − 共hef − hc兲


40冑d0
册 ␲␶d0共hef − hc兲

hef ⬎ 40冑d0 + hc ␶max␲d0 冋


␭⬘
冑d0
tanh
␭⬘共hef − hc兲
冑d 0 册 where hc =
␶␲d0
1.84冑 f c⬘
Özdemir 共2005兲 hef ⬍ 50 mm 0.33冑 f c⬘hef 共d0 + hef 兲␲ —
hef ⬎ 50 mm 0.33冑 f c⬘50共d0 + 50兲␲ ␲␶d0共hef − 50兲
Ozbakkaloglu hef ⬎ hc f ct.exphc共d0 + hc兲␲ ␲␶d0共hef − hc兲
and Saatcioglu 共2009兲
Note: f c⬘ = concrete cylinder compressive strength; f ct.exp = tested concrete splitting tensile strength; dh = head diameter of headed anchor; d0 = anchor hole
diameter; d = diameter of embedded anchor; hef =effective embedment depth; hc = concrete cone failure depth; ␶ = uniform bond stress; ␶⬘ = design version
of ␶; ␶max = maximum value of elastic bond stress; ␭⬘ = elastic constant; Ac = projected area of a single anchor; k = calibration factor; and ⌿c
= modification factor for concrete strength.
a
Original source in German but material extracted from CEB 共1994兲 and Eligehausen et al. 共2006a兲.
b
Design factor 共5% fractile兲.
c
Not specifically stated.

stress distribution was assumed to be described by a hyperbolic 共i.e., combined model兲; and 共3兲 bond model. Comparison of these
tangent function. The maximum bond stress of the elastic model three models by Eligehausen et al. 共2006a兲, with limited test data,
has been shown to give a similar prediction to a uniform bond revealed the first model to greatly underestimate and overestimate
stress model 共Cook 1993兲. The uniform bond stress model as well the pullout load 共depending on the embedment depth兲, the second
as elastic model showed overall good correlation with test results model to be predominantly conservative, and the third model to
with the elastic model correlating well at low stress 共load兲 levels
generally provide the best fit. Hence, the bond model appears to
and deep embedment depths 共Cook 1993; Cook et al. 1998兲.
be the preferred choice.
Based on the results of numerical analyses 共McVay et al. 1996兲,
the uniform bond stress model is more appropriate for strength The limited number of FRP anchor analytical models has been
design and in particular adhesive anchors which may fail in con- developed by Özdemir 共2005兲 and Ozbakkaloglu and Saatcioglu
crete cone or combined failure modes 共Cook et al. 1998兲. 共2009兲. These modes, as presented in Table 1, are based on the
Eligehausen et al. 共2006a兲 provided a summary of the devel- model of Cook et al. 共1993兲 and it is obvious that an accurate
opment of three analytical models for adhesive metallic anchors, prediction on the concrete cone depth 共hcone兲 is crucial. Özdemir
namely, 共1兲 concrete cone model; 共2兲 concrete cone+ bond model 共2005兲 assumed hcone to be a constant 50 mm, as observed from

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010 / 409

J. Compos. Constr. 2010.14:406-414.


his tests, with Ozbakkaloglu and Saatcioglu 共2009兲 measuring Proposed Pullout Strength Models for FRP Anchors
hcone from their tests but not providing a general expression.
Analytical models for predicting the pullout failure mode and
strength of FRP anchors are presented in this section. The general
Anchor Rupture Models form of the model is first given followed by a detailed description
Anchor rupture strength is based on the tensile capacity of the of its various components.
anchor. Kim and Smith 共2009兲 showed hand-made FRP anchors
to possess a tensile strength as low as 65% of the strength of flat General Form of the Strength Models
FRP tensile coupons.
The following analytical models are based on the assumptions of
共1兲 single anchor behavior 共no anchor groupings and multiple
anchors do no interfere with each other兲; 共2兲 monotonically ap-
Experimental Data plied tension load; 共3兲 idealized condition for epoxy cure 共i.e., dry
and clean holes, moderate temperature, and proper curing of
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Cosenza et al. 共1997兲 speculated that a rational approach to de- epoxy兲; and 共4兲 uncracked concrete 共failure surface is in no way
signing internal FRP reinforcement for concrete structures must affected by cracks in the immediate vicinity of the anchor or the
be based on theory largely dependent on the physical behavior of edges of the concrete pullout test blocks兲. Experiments have re-
FRP. Analytical models must be calibrated with embedded FRP vealed similarities between metallic and FRP anchors so analyti-
test results and not by merely altering the material properties of cal models for the former will be used for the later. The analytical
analytical models developed for embedded steel bars. Variability model describing the pullout resistance 共Nu兲 of a single FRP an-
of the anchor geometry, Poisson’s ratio effect, as well as surface chor is therefore given as
texture of FRP bars is likely to explain a lot of the differences
between the pullout behaviors of metallic anchors to FRP rein- Nu = min共Ncc,Ncb,Nar兲 共2a兲
forcing bars. These comments can be extended to FRP anchors,
meaning the pullout resistance is dependent on the properties of
the concrete, FRP anchor, as well as the bond between the con- Ncc = ␣cch1.5冑 c 共concrete cone failure兲
ef f ⬘ 共2b兲
crete and anchor. The components of particular importance are 共1兲
FRP anchor 共anchor or hole diameter, effective embedded depth兲;
共2兲 concrete 共compressive strength兲; and 共3兲 adhesive material Ncb = ␶u␲d0hef 共combined cone-bond failure兲 共2c兲
共strength of the bonded interface兲. Poission’s ratio effect of the
anchor is ignored for simplicity. Nar = ␣arwFRPtFRP f FRP 共anchor rupture failure兲 共2d兲
An experimental database of 101 data points of FRP anchors
under pullout conditions was assembled by Kim and Smith 共2009兲 where hef = effective embedment depth of the anchor 共mm兲; f ⬘c
based on the data obtained from tests by Kim and Smith 共2009兲, = concrete cylinder compressive strength 共MPa兲; d0 = diameter of
Özdemir 共2005兲, and Ozbakkaloglu and Saatcioglu 共2009兲. the anchor hole 共mm兲; wFRP and tFRP = width 共mm兲 and thickness
Eighty-four of these data points are considered herein with the 共mm兲, respectively, of the fiber sheet used in construction of the
remaining 17 data points deemed unsuitable for development of FRP anchor; and f FRP = flat coupon tensile rupture FRP strength
analytical models. The selected data are classified in terms of 共1兲 共MPa兲.
concrete cone failure 共30 data points兲; 共2兲 combined failure 共46 The key elements for the cone, combined, and anchor rupture
data points兲; and 共3兲 anchor rupture 共eight data points兲. The rel- models are the calibration factors denoted by ␣cc, ␶u, and ␣ar,
evant 25 tests from the study of Kim and Smith 共2009兲 reported respectively. In essence, these factors are determined from a sta-
varying anchor hole diameter 共nominal diameter= 12, 14, and 16 tistical analysis of the different test data sets of the 84 data points
mm兲 and embedment depth 共nominal depth= 20, 40, and 60 mm兲. described in the previous section for both best fit 共50% ex-
Sixteen test results were selected from the study of Özdemir ceedence of test results兲 and design 共5% exceedence or 95 per-
共2005兲 in which the anchors failed in a concrete cone or combined centile lower bound兲. The data set is assumed to be normally
manner with the variables being embedment depth 共nominal distributed and all statistical results are presented in Table 2, with
depth= 50, 70, and 100 mm兲 and concrete cylinder compressive the average expressed as a ratio of test to prediction. The standard
strength 共approximately 10 and 16 MPa兲. The majority of the tests deviation 共SD兲, coefficient of variation 共CoV兲, and percentage of
of Özdemir 共2005兲 failed by rupture of the FRP anchors; however, exceedence 共PE兲 are also all given. Each factor in Table 2 is
as tested mechanical properties of the FRP were not reported, calibrated independently 共i.e., no coupling of Ncc, Ncb, and Nar兲.
such ruptured anchor results are not considered herein. The analytical model proposed in Eq. 共2a兲 is largely based on
Ozbakkaloglu and Saatcioglu 共2009兲 reported 73 tests however existing analytical models developed for metallic anchors. The
only 43 results are used herein because all the anchors failing by reason for adopting this strategy and not regressing the relevant
rupture could not be included due to the tested mechanical prop- FRP anchor test data directly is due to several reasons, namely,
erties of the FRP not being reported. Also, test specimens of 共1兲 analytical models derived from an abundance of metallic an-
Ozbakkaloglu and Saatcioglu 共2009兲 which failed in a concrete chor tests over the last couple of decades have established clear
splitting mode as well premature anchor rupture failure due to rational trends in test data thus producing rational models; 共2兲
poor anchor installation were also not included herein as well. limited tests on FRP anchors have revealed many similarities in
The selected 43 tests failed in a concrete cone or combined man- behavior 共i.e., failure兲 to metallic anchors; and 共3兲 the limited
ner in which the variables were anchor hole diameter 共nominal FRP anchor tests data contained are not expected to yield statis-
diameter= 12.7, 15.9, and 19.1 mm兲, embedment depth 共nominal tically meaningful regression results for key parameters. Models
depth= 25, 50, 75, and 100 mm兲, and concrete strength 共nominal regressed directly from FRP anchor test data may however be
strength= 30 and 50 MPa兲. attempted in the future when more test data is available.

410 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010

J. Compos. Constr. 2010.14:406-414.


Table 2. Calibration Factors and Statistical Performance of Analytical Models for FRP Anchors
共a兲 Concrete cone failure
Calibration factors Statistics
Number
Model type ␣cc ␶u1 ␶u2 ␣ar Av SD CoV PE of tests
Raw 14.70 — — — 0.82 0.10 12 97 30
Best-fit 12.04 — — — 1.00 0.12 12 50
Design 9.68 — — — 1.24 0.15 12 5
共b兲 Combined failure
Low compressive strength concrete 共f c⬘ ⬍ 20 MPa兲
Raw — 15.40 — — 0.37 0.04 11 100 11
Best-fit — 5.65 — — 1.00 0.11 11 50
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Design — 4.62 — — 1.22 0.13 11 5


Normal compressive strength concrete 共f c⬘ 艌 20 MPa兲
Raw — — 15.40 — 0.71 0.07 10 100 35
Best-fit — — 10.86 — 1.00 0.10 10 50
Design — — 9.07 — 1.20 0.12 10 5
共c兲 Anchor failure
Raw — — — 0.65 1.11 0.12 11 19 8
Best-fit — — — 0.72 1.00 0.11 11 50
Design — — — 0.59 1.22 0.13 11 5
共d兲 Overall performance
Best fit 12.04 5.65 10.86 0.72 1.01 0.10 10 47 84
Design 9.68 4.62 9.07 0.59 1.23 0.13 11 4
Note: ␣cc, ␶u1, ␶u2, and ␣ar = calibration factors for cone, combined, and rupture models; Av= average of test to prediction; SD= standard deviation;
CoV= coefficient variation 共%兲; and PE= percentage of exceedence 共%兲.

Calibration of the Concrete Cone Model 20 MPa 共i.e., data of Özdemir 2005兲, ␶u ranged from 4.6 to 6.6
MPa and for compressive strength greater than 20 MPa 共i.e., data
Eligehausen’s research group proposed the concrete cone model
of Ozbakkaloglu and Saatcioglu 2009; Kim and Smith 2009兲, ␶u
of Eq. 共2b兲 with varying values of ␣cc proposed over several years
ranged from 8.7 to 14.2 MPa. The corresponding ␶u values for the
of research and publications. Most recently, Eligehausen et al.
concrete cone failing specimens presented in the test database
共2006b兲 proposed ␣cc = 14.7 for postinstalled anchors which in-
range from 6.5 and 10 MPa 共concrete strength⬎ 20 MPa兲, while
cludes expansion, undercut, adhesive, and grout anchors. Such
those for the anchor rupture failure cases presented in the test
anchors bear close resemblance to FRP anchors and are therefore
database range from 7.5 to 13.1 共concrete strength⬎ 20 MPa兲. It
of most relevance here. Table 2 reveals that when ␣cc equals 14.7
is therefore logical that the largest uniform bond stresses are
the pullout strength of FRP anchors is overestimated 共i.e., PE
achieved in the specimens failing in a predominantly bonded in-
= 97%兲 when assessed with the relevant test data. The correspond-
terface manner. In order to assess the suitability and accuracy of a
ing best-fit and design values for ␣cc are 12.04 and 9.68 共Table 2兲
more recently published analytical model for metallic anchors
with the CoV being only 12%. The statistical performance of the
with FRP anchor data, the model of Cook and Kunz 共2001兲 is
concrete cone model with size effect omitted 共i.e., Nc = ␣ch2ef 冑 f ⬘c analyzed under the “raw” model type for the data sets containing
form of equation兲 produces a best fit and design value of ␣cc of low concrete strength 共Table 2兲 and normal concrete strength
2.31 and 1.48, respectively, however with a much higher CoV of 共Table 2兲. The model is way too unconservative with percentages
22% which in turn justifies using Eq. 共2b兲. of exceedence equal to 100%. The model is therefore recalibrated
to produce the best-fit and design values of ␶u1 共low strength
Calibration of the Combined „Bond… Model concrete; Table 2兲 of 5.65 and 4.62 and ␶u2 共normal strength con-
crete; Table 2兲 of 10.86 and 9.07, respectively. The CoVs are an
Tests showed that anchors of sufficient length should fail in a
acceptable 11 and 10%, respectively. It should be noted that Cook
combined manner. The analytical approach adopted in this paper
and Kunz 共2001兲 reported the effect of concrete strength to have
is to ignore the resistance provided by the concrete cone and
minimal effect on the bond stress of adhesive metallic anchors
instead assume all resistance to be provided at the anchor-to-
while the type of coarse aggregate highly influenced the bond
concrete interface. Resistance provided by the cone has been
stress. On account of very limited information existing on the
shown for metallic anchors to contribute little to the pullout
type of aggregate used in all tests reported in the database of Kim
strength of the anchor 共Cook 1993兲. In addition, the depth of the
and Smith 共2009兲, this issue will not be considered and can be left
cone for longer anchors is most difficult to estimate due to the
for future research.
large scatter in test results.
The uniform bond stresses 共i.e., ultimate bond strength兲 at
pullout failure, ␶u, of all 46 combined failure test results are Calibration of the Anchor Rupture Model
calculated and presented by Kim and Smith 共2009兲 and a notice- The calibration factor, ␣ar, accounts for the differences in tensile
able trend can be observed. For compressive strengths less than strength between flat FRP coupon tensile test coupons to FRP

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010 / 411

J. Compos. Constr. 2010.14:406-414.


70 70
60 60
50 50

Nu,pre (kN)
Nu,pre (kN)
40 40
30 CC 30 CC
20 CB 20 CB
10 AR 10 AR
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Nu,test (kN) Nu,test (kN)

Fig. 3. Predicted versus test results 共best-fit model兲 Fig. 4. Predicted versus test results 共design model兲
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

anchors failing by tensile rupture. Kim and Smith 共2009兲 estab- cent to the concrete surface below the FRP plate. It does not
lished the tensile strength of FRP anchors in isolation to range consider the connection between the FRP anchor and the FRP
from 65 to 94% of the tensile strength of flat FRP coupon test plate which is typically via the anchor fan. In the case where the
coupons. As the FRP anchors were handmade, the fibers tended to force in the anchor is developed by transfer of stresses to the
be bendt and twisted and the reduction in strength was attributed anchor via the anchor fan, the bond strength of the anchor fan to
to the poor fiber alignment. It was hypothesized that the anchor FRP plate is needed.
quality should be improved though if mass produced with im- Fig. 5 provides a three-dimensional representation of Eq. 共3a兲.
proved manufacturing quality. According to Table 2, ␣ar = 0.65 The concrete cylinder compressive strength is assumed to be 30
gives an average of 1.11 while the values for best fit and design MPa, the anchor rupture strength Nar = 37.8 kN 关calculated design
are 0.72 and 0.59, respectively. All CoVs are an acceptable 11%. strength of 200-mm-wide anchors of Kim and Smith 共2009兲 with
0.117-mm sheet thickness兴, ␣cc = 9.68 MPa 共design兲, and ␶u
Overall Performance of the Calibrated Model = 9.07 MPa 共design兲. Figs. 6共a and b兲 show the relationships be-
tween failure load, failure mode, anchor hole diameter, hef / d0
Table 2 reveals the overall best-fit average of Nu to be 1.01 ratio, and concrete strength. Several observations can be made
共CoV= 10% and PE= 47%兲 while the corresponding design from Figs. 5 and 6, namely, 共1兲 the anchor pullout strength is
average= 1.23 共CoV= 13% and PE= 4%兲 when used to assess all ultimately limited by the anchor rupture strength; 共2兲 anchors with
eight-four data points. The predicted load 共vertical axis兲 versus large anchor hole diameters are susceptible to concrete cone fail-
test results 共horizontal axis兲 for the best-fit and design cases are ure 共e.g., anchors with larger than 18-mm hole diameter兲; 共3兲
subsequently shown in Figs. 3 and 4, respectively. In the deter- combined failure dominates anchors of smaller diameters; and 共4兲
mination of Nu, 87% of the failure modes were correctly predicted concrete strength has a minor effect on the failure mode but can
for the best-fit case 共i.e., 74 out of 84 tests correctly predicted兲 affect the failure load.
and 84% of cases for the design case predictions 共71 out of 84
tests correctly predicted兲. Failure modes were incorrectly pre-
dicted when the predicted failure loads for different failure modes Conclusions
were similar. This shortcoming is to be expected due to the inher-
ent scatter in the test data which is reflected in the calibration This paper has investigated analytically the pullout resistance of
factors. More test data will surely increase the reliability of failure FRP anchors in uncracked concrete. A review of the behavior of
mode prediction. The failure loads can however be predicted with metallic anchors was initially made and similarities 共and differ-
confidence and reliability. ences兲 with FRP anchors were identified. Existing analytical mod-
els, originally developed for metallic anchors, were critically

Recommendations for Design and Parametric Study

Based on the statistical results reported in Table 2, the design


AR
model recommendations are given as follows: 1.0

Nu = min共Ncc,Ncb,Nar兲 共3a兲 0.8


N u / N ar

Ncc = 9.68h1.5冑 c 共cone failure兲


ef f ⬘ 共3b兲 0.6

0.4 CB
Ncb = 4.62␲d0hef 共f ⬘c ⬍ 20 MPa兲 共combined failure兲 共3c兲 0.2
0.0
Ncb = 9.07␲d0hef 共f ⬘c 艌 20 MPa兲 共combined failure兲 共3d兲 22 20
20 15
18
Nar = 0.59wFRPtFRP f FRP 共anchor failure兲 共3e兲 d 16 10
h ef
/d o
0 (m 14
m) 12 5
The ranges of test parameters used to calibrate these models are CC
共1兲 17.5 mm艋 hef 艋 100.0 mm; 共2兲 10.4 MPa艋 f ⬘c 艋 60.0 MPa; 10
0
f'c = 30 MPa

and 共3兲 11.8 mm艋 d0 艋 20.0 mm. In addition, the model is con-
fined to the embedded region of the anchor and the region adja- Fig. 5. Parametric study 共design model兲

412 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010

J. Compos. Constr. 2010.14:406-414.


1.2 Compos. Constr., 12共6兲, 596–607.

Pa

Pa

Pa
0M

0M
Aiello, M. A., De Lorenzis, L., Galati, N., and La Tegola, A. 共2004兲.

M
AR

20
=4

=3

=
f'c = 20, 30, 40 MPa “Bond between FRP laminates and curved concrete substrates with

f'c

f'c

f'c
1.0
anchoring composite spikes.” Proc., Innovative Materials and Tech-
nologies for Construction and Restoration (IMTCR 2004), Lecce,
0.8
CC Italy.
CB American Concrete Institute 共ACI兲. 共1985兲. “Code requirements for
Nu / Nar

f'c = 20, 30, 40 MPa


0.6 nuclear safety related concrete structures.” ACI 349-85, Detroit.
American Concrete Institute 共ACI兲. 共2001兲. “Code requirements for
0.4
nuclear safety related concrete structures and commentary.” ACI 349-
01, Detroit.
American Concrete Institute 共ACI兲. 共2005兲. “Building code requirements
0.2
for structural concrete and commentary.” ACI 318M-05, Detroit.
d0 = 12 mm Binici, B., Ozcebe, G., and Ozcelik, R. 共2007兲. “Analysis and design of
FRP composites for seismic retrofit of infill walls in reinforced con-
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

0.0
0 2 4 6 8 10 crete frames.” Composites, Part B, 38共5–6兲, 575–583.
hef / d0 CEB. 共1994兲. Fastenings to concrete and masonry structures: State-of-
the-art report, Comité Euro-International Du Béton, Lausanne, Swit-
(a) Small anchor hole diameter (d0 = 12 mm) zerland.
Cook, R. A. 共1993兲. “Behavior of chemically bonded anchors.” J. Struct.
Eng., 119共9兲, 2744–2762.
Pa
Pa

Pa

1.2
Cook, R. A. 共1999兲. Strength design of anchorage to concrete, Portland
40M
30M

0M
=2

AR
Cement Association, Ill.
f'c =
f'c =

f'c = 20, 30, 40 MPa


f'c

1.0 Cook, R. A., Doerr, G. T., and Klinger, R. E. 共1993兲. “Bond stress model
for design of adhesive anchors.” ACI Struct. J., 90共5兲, 514–524.
0.8 Cook, R. A., Kunz, J., Fuchs, W., and Konz, R. C. 共1998兲. “Behavior and
design of single adhesive anchors under tensile load in uncracked
CC
Nu / Nar

concrete.” ACI Struct. J., 95共1兲, 9–26.


0.6
Cook, R. A., and Kunz, R. C. 共2001兲. “Factors influencing bond strength
of adhesive anchors.” ACI Struct. J., 98共1兲, 76–86.
0.4 Cosenza, E., Manfredi, G., and Realfonzo, R. 共1997兲. “Behavior and
modeling of bond of FRP rebars to concrete.” J. Compos. Constr.,
0.2
1共2兲, 40–51.
Eligehausen, R., Bouška, P., Červenka, V., and Pukl, R. 共1992兲. “Size
d0 = 20 mm
effect on the concrete cone failure load of anchor bolts.” Proc., 1st Int.
0.0
0 2 4 6 8 10
Conf. on Failure Mechanics of Concrete Structures, FraMCoS1,
Elsevier Applied Science, Colo., 517–525.
hef / d0
Eligehausen, R., Cook, R. A., and Appl, J. 共2006b兲. “Behaviour and
design of adhesive bonded anchors.” ACI Struct. J., 103共6兲, 822–831.
(b) Large anchor hole diameter (d0 = 20 mm) Eligehausen, R., and Fuchs, W. 共1988兲. “Tragverhalten von Dübelbefes-
tigungen bei Querzug-, Schrägzug- und Biegebeanspruchung 共Load-
Fig. 6. Parametric study 共design model兲 bearing behaviour of anchor fastenings under shear, combined tension
and shear or flexural loading兲.” Betonwerk Fertigteil-Tech., 88共2兲,
48–56.
reviewed and a best-fit and design model for use with FRP an- Eligehausen, R., Fuchs, W., and Mayer, B. 共1987兲. “Tragverhalten von
chors, which was calibrated against a test database consisting of Dübelbefestigungen unter Zugbeanspruchung 共Bearing behaviour of
eight-four data points, was then provided. While the proposed anchor fastenings under tension.” Betonwerk Fertigteil-Tech., 87共12兲,
model provides the first accurate representation of the pullout 826–832; 1988, 共1兲, 29–35.
Eligehausen, R., Mallée, R., and Rehm, G. 共1984兲. “Befestigungen mit
resistance of FRP anchors, not all the failure modes were always
Verbundankern 共Fastenings with bonded anchors兲.” Betonwerk
correctly predicted. More tests on parameters both inside and out-
Fertigteil-Tech., 84共10兲, 682–692; 共11兲, 781–785; 共12兲, 825–829.
side of the range of existing test data are required in order to
Eligehausen, R., Mallée, R., and Silva, J. F. 共2006a兲. Anchorage in con-
improve the accuracy of the models and extend their range of
crete construction, Ernst & Young, Germany.
applicability. Eligehausen, R., and Sawade, G. 共1989兲. “A fracture mechanics based
description of the pull-out behaviour of headed studs embedded in
concrete.” Fracture mechanics of concrete structures. From theory to
Acknowledgments applications, Report of RILEM Committee TC90–FMA, Chapman and
Hall, London, 281–299.
Funding provided by the Hong Kong Research Grants Council in Eshwar, N., Ibell, T. J., and Nanni, A. 共2005兲. “Effectiveness of CFRP
the form of General Research Fund HKU 715907E is gratefully strengthening on curved soffit RC beams.” Adv. Struct. Eng., 8共1兲,
acknowledged. 55–68.
FIB. 共2000兲. “Bond of reinforcement in concrete.” Bulletin No. 10, Task
Group on Bond Models, Int. Federation for Structural Concrete 共FIB兲,
Lausanne, Switzerland.
References Fuchs, W., Eligehausen, R., and Breen, J. E. 共1995兲. “Concrete capacity
design 共CCD兲 approach for fastening to concrete.” ACI Struct. J.,
Ahmed, E. A., El-Salakawy, E. F., and Benmokrane, B. 共2008兲. “Tensile 92共1兲, 73–94.
capacity of GFRP postinstalled adhesive anchors in concrete.” J. Galati, D., Rizzo, A., and Micelli, F. 共2007兲. “Comparison of reinforced

JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010 / 413

J. Compos. Constr. 2010.14:406-414.


concrete beams strengthened with FRP pre-cured laminate systems Ozbakkaloglu, T., and Saatcioglu, M. 共2009兲. “Tensile behaviour of FRP
and tested under flexural loading.” Proc., 8th Int. Symp. on Fiber anchors in concrete.” J. Compos. Constr., 13共2兲, 82–92.
Reinforced Polymer Reinforcement for Concrete Structures (FRPRCS- Özdemir, G. 共2005兲. “Mechanical properties of CFRP anchorage.” MS
8), Univ. of Patras, Greece. thesis, Middle East Technical Univ., Turkey.
Gose, S. C., and Nanni, A. 共2000兲. Anchorage system for externally Seliem, H. M., Sumner, E. A., Seracino, R., and Smith, S. T. 共2008兲.
bonded FRP laminates using near surface mounted rods, Center for “Field testing of RC slabs with openings strengthened with CFRP.”
Infrastructure Engineering Studies, Univ. of Missouri–Rolla, Mo. Proc., 4th Int. Conf. on FRP Composites in Civil Engineering (CICE
Hollaway, L. C., and Teng, J. G. 共2008兲. Strengthening and rehabilitation 2008), Empa, Zürich, Switzerland.
of civil infrastructures using fibre-reinforced polymer (FRP) compos- Sharif, A., Al-Sulaimani, G., Basunbul, I. A., Baluch, M. H., and Ghaleb,
ites, Woodhead, Cambridge, U.K. B. N. 共1994兲. “Strengthening of initially loaded reinforced concrete
Kim, S. J., and Smith, S. T. 共2009兲. “Behaviour of handmade FRP an- beams using FRP plates.” ACI Struct. J., 91共2兲, 160–168.
chors under tensile load in uncracked concrete.” Adv. Struct. Eng. Smith, S. T., and Gravina, R. J. 共2007兲. “Prediction of debonding failure
12共6兲, 845–865. in FRP flexurally strengthened RC members using a local deformation
Lam, L., and Teng, J. G. 共2001兲. “Strength of RC cantilever slabs bonded model.” J. Compos. Constr., 11共2兲, 184–191.
with GFRP strips.” J. Compos. Constr., 5共4兲, 221–227. Smith, S. T., and Kim, S. J. 共2010兲. “Strength and behavior of impreg-
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

McVay, M., Cook, R., and Krishnamurthy, K. 共1996兲. “Pullout simulation nated carbon FRP anchors in FRP-to-concrete joint assemblies.” Adv.
of postinstalled chemically bonded anchors.” J. Struct. Eng., 122共9兲, Struct. Eng., accepted.
1016–1024. Smith, S. T., and Teng, J. G. 共2003兲. “Shear-bending interaction in deb-
Oh, H.-S., and Sim, J. 共2004兲. “Interface debonding failure in beams onding failures of FRP-plated RC beams.” Adv. Struct. Eng., 6共3兲,
strengthened with externally bonded GFRP.” Compos. Interfaces, 183–199.
11共1兲, 25–42. Tan, K. H., and Patoary, M. K. H. 共2004兲. “Strengthening of masonry
Okelo, R., and Yuan, R. L. 共2005兲. “Bond strength of fiber reinforced walls against out-of-plane load using fiber-reinforced polymer rein-
polymer rebars in normal strength concrete.” J. Compos. Constr., forcement.” J. Compos. Constr., 8共1兲, 79–87.
9共3兲, 203–213. Teng, J. G., Smith, S. T., Yao, J., and Chen, J. F. 共2003兲. “Intermediate
Orton, S. L., Jirsa, J. O., and Bayrak, O. 共2008兲. “Design considerations crack induced debonding in RC beams and slabs.” Constr. Build.
of carbon fibre anchors.” J. Compos. Constr., 12共6兲, 608–616. Mater., 17共6–7兲, 447–462.

414 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / JULY/AUGUST 2010

J. Compos. Constr. 2010.14:406-414.

You might also like