ARTICLE IN PRESS

International Journal of Adhesion & Adhesives 28 (2008) 101–108

www.elsevier.com/locate/ijadhadh

Analysis of steel/concrete interfacial shear stress

by means of pull out test
L. Bouazaoui, A. Li
Groupe Mécanique, Matériaux et Structures, UFR Sciences, Université de Reims Champagne Ardenne, Moulin de la Housse,
BP 1039, 51687 Reims Cedex, France
Accepted 21 February 2007
Available online 22 July 2007

Abstract

The excellent mechanical performance of epoxy resins has encouraged their wide-spread application in civil engineering. This paper
deals with the prediction of the interfacial shear strength between the steel bar surface and concrete surface of steel rods bonded into
concrete. The pull out test is used in order to determine the ultimate force and the shear stress of steel–concrete specimens. The
relationship between the ultimate force, the diameter and the length embedded into the concrete are investigated by experimental test.
After discussing some theoretical models, a theoretical model is also proposed in this work in order to estimate the shear distribution and
the critical shear stress at the instance that the first cracks appear and of the ultimate failure of steel/concrete structure. A comparison
between the test and the theoretical results is provided.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Concrete; Steel; Stress analysis; Pull out

1. Introduction The single-fibre pull out test is widely used in composite

In reinforced concrete structures, steel/concrete inter-
facial strength plays an important role. The separation of
steel/concrete is one of the principal causes for deteriora-
tion or damage of the civil infrastructure. It is well
recognized that first cracks appear in a reinforced concrete
structure, which then propagate along the steel bar’s
concrete interface under loading. Also, it is well known
that the steel/concrete structure should possess good
interfacial shear strength. However, the shear strength
depends upon the mechanical properties of the steel and
concrete and the surface properties of steel bars and
concrete. To improve the steel/concrete interfacial shear
strength, an adhesive joint is added between the steel bar
surface and the concrete surface. The experimental
program involved using pull out tests for purposes of
investigating the interfacial adhesion quality and interfacial
properties between steel bar and concrete, and the shear
stress along the bonded steel rod/concrete interface.
Corresponding author. Tel.: +33 326 91 3044; fax: +33 326 91 8711.
E-mail address: alex.li@univ-reims.fr (A. Li).
material structures [1–8]. ‘Fibres’ have included metal
rods, carbon filaments, glass fibre and polymer fibres
such as Kevlar and ultra-high density polyethylene. In
this case, the ‘matrix’ materials are concrete and epoxy
resin.
One of the earliest descriptions of the single-fibre pull
out test, which detailed the initial debonding, crack
propagation, completion and fibre pull out, was given by
Kelly in 1970 [9]. Other literature shows that various
analytical and numerical models have been developed for
the initiation of crack propagation between fibre and resin,
subsequent propagation of this crack, evaluation of the
fibre/matrix interfacial shear strength and the load transfer
between a single fibre and an infinite matrix [2–5].
These models may be distinguished between two different
kinds of analytical approaches: (a) the perfect interface
model, which is mainly used for analysis of stress transfer
problems in resin matrix composites, and (b) the cohesive
interface model, which is frequently applied to analyse
cement based matrix composites [10]. The principal
difference between these two approaches is that in the
first case a ‘perfect interface’ between fibre/matrix is

0143-7496/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijadhadh.2007.02.006
 ARTICLE IN PRESS
102 L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108

assumed. There is no slip between fibre and matrix, and

displacement and tractions are continuous at the inter-
face. As research progressed, the interpretation of data
evolved with the development of models through the
shear lag hypothesis as applied by Greszczuk [11] and
interfacial fracture energy as applied by Outwater and
Murphy [12].
Indeed, the fibre pull out problem has been investigated
extensively for the purpose of investigating the interfacial
adhesion quality and interfacial properties between fibres
and matrix and the elastic stress transfer in the fibre pull
out problem [6–8]. However, these studies have not been
applied to bonded steel rod–concrete problem.
The present paper tries to analyse mechanical behaviour
of steel concrete composite system with an epoxy resin.
The specimen configuration consists of a steel rod
Fig. 1. Uniaxial stress–strain relationship of concrete in compression.
embedded in concrete matrix or a steel rod bonded with
concrete by epoxy resin. This paper also discusses the
theoretical analyses of the normal and shear stress
distribution in steel/concrete interface along the rod axial F
axis. The role of epoxy resin layer in improving shear
2r
strength is discussed. The experimental programme was
carried out in order to determinate the influence of L' Strain gauge
85 mm
some parameters, such as steel rod diameter, embedded
LVDT
or bonded length and bonded surface on the shear 15 mm Steel
strength and stress distribution along the rod’s length
direction of steel–concrete structures by using the pull out Adhesion
test.
L

Concrete
2. Materials
X
Three materials, steel bar, concrete and epoxy resin, are
used in this work. The steel bar used was a type S275. 2R
Three diameters 12, 16 and 20 mm of steel bar were selected
with five different embedded lengths (100, 150, 200, 250
and 300 mm). The yield strength and Young’s modulus of
the steel bar obtained by test was fy ¼ 340 MPa and F
Es ¼ 198,000 MPa, respectively. The initial state of the
Fig. 2. Geometry of pull out model.
surface was smooth. The surface of the steel bar was then
treated by mechanical sandblasting.
The concrete specimen was a cylinder, which had a by the following formula:
diameter of 160 mm and a length of 320 mm. The concrete
had an average compressive strength at 28 days of 40 MPa. 2f 0c
0 ¼ , (2)
The measured tensile strength was 3.3 MPa and the elastic Ec
modulus Ec was 37,800 MPa. The steel bar was embedded
in the centre of the concrete cylinder. The relationship with Ec being the elastic modulus of concrete.
between strain and stress is presented in Fig. 1. In the When the concrete reaches its ultimate compressive
elastic region, the relationship for concrete can be strength, the strain e0 depends only on the compressive
described by the following formula [13]: strength f0 c and the strain value is equal to 0.003. The
" tensile stress of concrete ft is calculated as ft ¼
 2 #
0 2c c 0.6(1+0.1fc).
fc ¼ fc  , (1) The selected adhesive is an epoxy resin with two
0 0
components. The measured elastic tensile modulus of
where fc is the compressive stress, f0 c is the average epoxy adhesive is Ea ¼ 3600 MPa and its tensile strength
compressive strength of concrete at 28 days. ec is the is sa ¼ 22 MPa.
concrete strain and e0 is the strain corresponding to The geometry of the pull out test is shown in Fig. 2. The
the ultimate compressive strength which is determined diameter and the height of concrete were 160 and 320 mm,
 ARTICLE IN PRESS
L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108 103

respectively. The steel rod diameter was varied from 12 to
20 mm. The adhesive thickness is 1 mm.
If it can be assured that the shear stress in the adhesive t
does not vary according to the immersed length of steel bar
in the concrete, then the shear stress can be determined by
t ¼ F/(pdL), with F being the load applied, d the diameter
of the steel bar and L the length of steel immersed in the
concrete.
It is also admitted that Hooke’s law is valid for steel
behaviour. The relationship between the applied force F
and the strain es can be described by F ¼ EsesAs, with Es the
elastic modulus and As the area of steel bar.
To study this relationship between the shear stress, the
steel diameter and the embedded steel length in the
concrete, three steel diameters (12, 16 and 20 mm) and
three (140, 210 and 280 mm) or five (100, 150, 200, 250 and
300 mm) lengths of embedded steel in the concrete were
used.
3. Theoretical model
Many theoretical models regarding the pull out of fibres
from a brittle matrix are in the literature. These models are
based on either interfacial fracture energy or shear stress
analysis and have proven insightful. However, the wide use
and application of these models to experimental data are at
best limited. This is partially due to difficulties in
developing micro-scale samples with good experimental
reproducibility and the appropriate recording of the failure
events, especially for small and sometimes for large-scale
samples.
3.1. Models in literature
In 1969, Outwater and Murphy proposed a model based
on the fracture energy of unidirectional laminates [12]. The
relationship between the debonding of the fibre, the tension
in the fibre and the fracture energy of the laminate is shown
as the following:
 2  
p dC
G¼ , (3)
2pd da
where G is the opening mode fracture energy, p is the
tensile load on the fibre, d is the fibre diameter and dC/da is
the compliance of structure as a function of crack length.
In 1982, Laws presented another model based on the
shear lag theory [14]. According to this theory, a pull out
curve can be determined by the bond strength ts and the
frictional slip resistance tf. The load–displacement
curve during the elastic stage before debonding can be
expressed as
bE f Af sinhðbLf Þ
Pf ¼ d, (4)
coshðbLf Þ  1
where Lf is the embedded length; Ef and Af are the Young’s
modulus and the cross-section area of the fibre, respec-
tively; and d is the displacement. However, it is difficulty to
use this formula, because the constant b normally cannot
be determined directly.
Li has simplified this model by applying some
conditions [15]. The bond strength is ignored; instead the
average bond, tav, is assured constant along the debonded
part of the fibre and the stress distribution along the
debonded part of the fibre is assumed linear. So, the
following expression to determine the pull out curve can be
obtained:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
E f d 3f tav d
Pf ¼ p ; d  d0 , (5)
2
2L2f tav
d0 ¼ , (6)
Ef df
where d0 corresponds to the displacement when maximum
load occur and the debonding is completed along the
embedded length of the fibre. df is the fibre diameter.
The average bond strength, tav, is calculated as
p
tav ¼ max . (7)
pd f Lf
In 1985, Piggott [16] proposed a debonding model which
allows estimating the axial strain ef in the embedded region
of the fibre at a distance x from the resin surface and given
by the following:
sinh½nðLe  xÞ=r
f ¼ app , (8)
sinhðnsÞ
where
Em 1 1
n2 ¼ , (9)
E f lnðR=rÞ ð1 þ nm Þ
Em is the matrix tensile modulus, Ef is the fibre tensile
modulus, R is the effective radius of the interface, r is the
fibre radius, nm is the matrix Poisson’s ratio, Le is fibre
embedded length and s is the fibre aspect ratio Le/r.
The interfacial shear stress can be estimated by
n cosh½nðLe  xÞ=r
t ¼ E f app . (10)
2 sinhðnsÞ
3.2. Proposed model
In this work, a cylindrical steel rod of diameter 2r is
bonded and embedded in a matrix of cylindrical concrete
(Fig. 2). The steel rod embedded length is L. r and R
represent the radius of the steel rod and concrete,
respectively. L0 is the free length of the steel rod.
It is assumed that the steel rod, adhesive and the concrete
are elastic and isotropic. The longitudinal direction is
parallel to the steel rod axis.
The applied external stress sF is assumed to be parallel to
the steel rod axis. Under the action of this external stress,
the steel rod and the adhesive deform elastically. For an
element dx of the test specimen (Fig. 3), the difference of
displacements du between the steel rod us and the adhesive
 ARTICLE IN PRESS
104 L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108

Y The mathematical solution of Eq. (17) can be as follows:

a2
ss ðxÞ ¼ C 1 coshða1 xÞ þ C 2 sinhða1 xÞ þ sf , (19)
(1+εa)dx a2
where C1 and C2 are determined by using the boundary
dx
concrete conditions ss(0) ¼ 0 for x ¼ 0 and ss(L) ¼ sf for x ¼ L:
 
adhesive a2 sf sf a2
C1 ¼ ; C2 ¼ 1  2 ð1  coshða1 LÞÞ .
γ γ+dγ
a21 sinhða1 LÞ a1
steel X (20)
(1+εs)dx From Eqs. (20) and (21), the distribution of normal stress
ss(x) can be estimated by
 
Fig. 3. Displacement of steel rod and adhesive joint for an element dx. ss ðxÞ a2 sinhða1 xÞ a2
¼ 2 ð1  coshða1 xÞÞ þ 1  2 ð1  coshða1 LÞÞ .
sf a1 sinhða1 LÞ a1
(21)
joint ua is given by
By using Eqs. (15) and (21), the distribution of interfacial
du ¼ us  ua (11) shear stress ti(x) is to be obtained by
and  
2ti ðxÞ a2 sinhða1 xÞ a1 coshða1 xÞ a2
¼ þ 1  2 ð1  coshða1 LÞÞ .
du rsf a1 sinhða1 LÞ a1
¼ s  a . (12)
dx (22)
So, shear strain is obtained by The applied external stress sf is assumed to be parallel to
dg ðs  a Þ the steel rod axis. Under the action of the external stress sf,
¼ . (13) the specimen deforms elastically.
dx r
tc is a critical shear stress at the fibre emergent end
Applied the Hooke’s law s ¼ Ee and the relationship (at x ¼ 0) for this loading configuration. Therefore, from
between the shear stress and shear modulus t ¼ gG, the Eq. (22), the pull out shear stress tc required to initiate
shear stress can be expressed as interfacial debonding is given by
   
dti G a du Ga ss sa 2tc a1 a2
¼ ¼  , (14) ¼ 1  2 ð1  coshða1 LÞÞ . (23)
dx r dx r Es Ea rsf sinhða1 LÞ a1
where Ga is the shear modulus of adhesive, r is the adhesive The theoretical maximum shear stress tmax recorded is
thickness, ss and sa are the steel rod stress and the adhesive obtained when x reaches the embedded length L such that
stress, respectively. Es and Ea are the steel Young’s  
modulus and the elastic modulus of adhesive, respectively. 2tmax a2 sinhða1 LÞ a1 coshða1 LÞ a2
¼ þ 1  2 ð1  coshða1 LÞÞ .
For the element dx, equilibrium of the interfacial shear rsf a1 sinhða1 LÞ a1
stress and the steel rod axial stress can be written as (24)
follows:
dss ðxÞ 2ti ðxÞ 4. Experimental results
¼ . (15)
dx r
Combining Eqs. (13), (14) and (15), the following equation In order to determine the distribution of the slip between
is obtained: the steel rod and the concrete, a linear variable differential
   transformer (LVDT) was used (Fig. 2). With an aim of
d2 ss ðxÞ 2G a 1 r2 r2 analysing the evolution of the strains and stresses along the
¼ 2 þ ss ðxÞ  sf .
dx2 r E s E a ðR2  r2 Þ E a ðR2  r2 Þ steel rod length, two strain gauges were also installed on
(16) the surface of the steel rod, 15 and 85 mm from the
concrete surface (Fig. 2). During the test, initially, a small
This differential equation can be expressed as load was applied to the specimen to enable checks on the
s00s ðxÞ  a21 ss ðxÞ þ a2 sf ¼ 0, (17) instrumentation. The load was then released and the
proper test started. The specimen was subject continuously
where increasing load until failure of the specimen was observed.
 
2Ga 1 r2 2G a
a21 ¼ 2 þ and a2 ¼ 4.1. Influence of embedded length
r E s E a ðR2  r2 Þ E a ðR2  r2 Þ
(18) Fig. 4 presents the evolution of the ultimate load
and sf is tensile stress of steel bar. according to the embedded length of steel rod in concrete
 ARTICLE IN PRESS
L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108 105

for three steel diameters. These curves show that the
ultimate load Fmax increases according to the embedded
length of steel in concrete. On the other hand, this increase
is very linear. From the test results, the curves give the
following relationship:
F max ¼ aL þ 21,
where Fmax is in kN, L is in mm and a is a constant.
For the steel rod diameter d ¼ 12 mm, a ¼ 0.04; for
d ¼ 16 mm, a ¼ 0.066; for d ¼ 20 mm, a ¼ 0.104. These
results show that in relation to the length, L, the slope of
curve increases linearly according to the steel rod diameter.
4.2. Influence of steel rod diameter
Fig. 5 shows the influence of steel rod diameter on the
ultimate load of the test specimen. It can be observed that
for a constant length L embedded in concrete, the increase
in force is also linear according to the steel rod diameter.
For three lengths of the steel rod embedded in concrete, the
results obtained by tests give a relationship between the
ultimate load and the diameter as follows:
F max ¼ k1 d þ k2 ,
where d is the diameter of steel rod. k1 and k2 depend on the
embedded length L: k1 ¼ 0.075L and k2 ¼ 0.05L+22.
This relationship is similar compared to the relationship
between the ultimate load of the test-specimen and the steel
rod length embedded in concrete.
60
d = 20 mm
50
F(kN) = 0.104 L+21
d = 16 mm
40 F(kN) = 0.066L+21
F (kN)
4.3. Influence of bonded surface
With an aim of analysing the relationship between
the loading and the bond area of the embedded surface
of the steel rod in to the concrete, the curve of ultimate
loading vs. embedded surface bond area was plotted
(25)
(Fig. 6). Based upon the experimental results, the curve
of load vs. embedded surface bond area shows that the
increase in resistance according to embedded surface is not
linear.
4.4. Comparison between the steel/concrete specimens with
adhesive joint and without adhesive joint
In order to study the effectiveness of adhesive joint in
improving the interfacial shear stress between the steel
surface and the concrete surface, the experimental test was
carried out on the specimen that had no adhesive in the
joint. In this case, the steel rod was embedded directly into
the concrete. The diameters of steel rod and the embedded
length in concrete were 12, 100, 200 and 300 mm,
respectively (Fig. 7). It is noted that the ultimate load
(26) increased significantly for the structure having an adhesive
in the joint. The role of adhesive joint to increase the
adhesive strength between the steel surface and the
concrete surface is considerable.
4.5. Force–strain curves
The curves in Fig. 8 present the evaluation of strain in
steel according to the loading. Fig. 8 shows that the steel
has not yielded prior to the failure of structure. For a
length embedded in the concrete of 300 mm, the beginning
of a yield region was detected. It is obvious that the strain
increases linearly in steel according to the loading. On the

30 d = 12 mm
F(kN) = 0.04 L+21
other hand, the slope of the curve varied according to the
20 steel rod length embedded in concrete. In the case of a
10
given load, Fig. 8 shows that the strain in steel length
embedded in concrete of 300 mm is greater. At the moment
0 of failure, for the specimen with 300 mm of length, the
0 50 100 150 200 250 300 350
L (mm)
maximum strain was 2000 mm/m. Then, the maximum

Fig. 4. Ultimate load–embedded steel rod length relationship. 60

50 F = 0.0006S2+0.5S+23
60
L = 280 mm 40
50
F (kN)

40 L = 210 mm 30
F (kN)

30 L = 140 mm 20
20
10
10

0 0
0 5 10 15 20 25 0 40 80 120 160 200
d (mm) S (cm2)

Fig. 5. Ultimate load–steel rod diameter curves. Fig. 6. Load–embedded steel rod surface relationship.
 ARTICLE IN PRESS
106 L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108

40 35
35 30 L = 200 mm
with adhesive
30 25
L = 100 mm
L = 300 mm
20

F (kN)
25
F (kN)

20 without adhesive 15

15 10

10 5

5 0
0 2 4 6 8
0 Slip (mm)
0 50 100 150 200 250 300 350
L (mm) Fig. 9. Load–slip between the steel and concrete surfaces (d ¼ 12 mm).

Fig. 7. Comparison of ultimate load for the structures with and without
adhesive joint between the steel rod and concrete surfaces (d ¼ 12 mm).
40

h = 15 mm
40 30
h = 85 mm

F (kN)
20
30 L = 300 mm
L =1 50 mm
10
F (kN)

L = 100 mm
20
0
0 300 600 900 1200 1500
10
Strain (μm/m)

Fig. 10. Comparison of strain measured by strain gauges installed on the

0
steel surface of a location 15 and 85 mm from the concrete surface for
0 400 800 1200 1600 2000
d ¼ 16 mm.
Strain (μm/m)

Fig. 8. Force–strain curves for the steel rod diameter 12 mm.
stress calculated by using the Hooke’s law is 396 MPa. This
value is more than that of yield strength of steel rod,
340 MPa.
Fig. 9 shows the evolution of the slip between the steel
rod and the concrete according to the applied load. In this
case, the relative slip increases according to the embedded
length of the steel rod. The curves show that the value of
maximum slip is approximately 3 mm for a 100 mm
embedded length and 5 mm for a 300 mm embedded length.
Two strain gauges are installed on the steel surface at the
position of h ¼ 15 and 85 mm from the concrete surface, in
order to measure the strain variation during the loading.
The results are shown in Fig. 10. The experimental values
show that the strain in steel at the position of 85 mm is
more than that at the position of 15 mm from the concrete
surface for a given load. It can therefore be assumed that
the steel near the concrete surface resists the tensile force
and shear force.
4.6. Modes of failure
In this configuration of specimen, the failure occurred in
the three principal regions: in the concrete, at the
steel–concrete interface and failure of the steel rod. It is
known that the steel rod is in tension. The concrete is
acting in tension and shear. The adhesive joint is assumed
to be acting in shear. It is then significant to know if the
stress in the material is in tension or in shear in order to
determine the mode of failure. Assuming that the tensile
stress in steel is higher than the tensile stress and shear
stress in the concrete and in the adhesive joint, the failure
will occur in the steel. In the case where the strength of the
steel is sufficient, failure of the interface is produced by the
ruin of the concrete or the slip of the steel rod. The
appearance of either of these two modes of failure depends
on the relationship between the values of tensile stress in
steel rod and the tensile and shear stress in the concrete.
For the first specimen (rod diameter 12 mm), the mode of
failure of the test specimen is due to the slip between the
adhesive surface and steel surface. For the other specimens,
it is observed that the failure was due to cracks in the
concrete (Fig. 11). Their crack length is between 200 and
230 mm.
The direction of propagation of the first cracks PO and
PQ (Fig. 11) in the concrete was perpendicular compared
to the axis of steel rod. As the loading was increased,
longitudinal cracks PRS appeared at the interface between
the steel surface and concrete surface. Then the cracks
propagated at the same time, longitudinally and transver-
sally until failure of specimen occurred.
 ARTICLE IN PRESS
L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108 107

O Q
P

Fig. 11. Failure mode of steel–concrete specimen.

Table 1
Theoretical and test results

Test results Theoretical results

d (mm) L (mm) F (kN) sf (MPa) tc(c) (MPa) tav (Eq. 7: t (Eq. 10: tc (Eq. 23) tmax (Eq. 24) tmax/sf
Laws–Li) (MPa) Piggott) (MPa) (MPa) (MPa)

12 100 24.8 219 4.0 6.58 9.4 6.00 12.0 0.055

12 150 28 248 1.9 4.95 9.3 4.80 12.7 0.051
12 200 29.9 264 1.1 3.97 9.0 4.46 13.4 0.051
12 250 31.5 279 0.7 3.34 9.4 4.46 14.0 0.050
12 300 34.1 302 0.6 3.02 10.1 4.74 15.1 0.050

16 140 31 154 2.7 4.41 6.5 5.16 7.8 0.051

16 210 35.5 177 1.4 3.36 6.6 4.90 8.3 0.047
16 280 40.3 200 1.0 2.86 7.7 5.25 9.3 0.046
20 140 36.5 116 3.1 4.15 5.6 5.45 5.9 0.051
20 210 42.4 135 1.8 3.21 5.8 5.37 6.0 0.045
20 280 51 162 1.4 2.90 6.6 6.11 7.0 0.043

4.7. Comparison of test and theoretical results
Table 1 shows the test results and the estimated shear
stress by using theoretical models. F is the ultimate force.
tf is normal tensile stress acting on the steel. The average
shear stress tav is estimated by Eq. (7). tc is interfacial shear
stress between the steel rod and the adhesive obtained by
using Eq. (23). The maximum shear stress tmax is calculated
by using Eq. (24). Shear stress t is calculated by using
Eq. (10) for x ¼ 0. tc(c) represents the interfacial shear
stress between the steel surface and the concrete surface
without adhesive joint.
The test results show that the ultimate force and normal
tensile stress increase according to the embedded length for
a given steel diameter. The maximal shear stress obtained
by theoretical calculus also increases according to the
embedded length. However, the values of average shear
stress, tav and tc obtained by the theoretical models,
decrease according to the embedded length.
In comparison with the values of average shear stress tav,
the values obtained by Eqs. (23) and (10) are very different.
The interfacial shear stress values tc estimated by using
Eq. (23) varied between 4.46 and 6.11 MPa. More
importantly shear stress values t estimated by using
Eq. (10), varied between 5.6 and 10.1 MPa. In the case of
diameter 12 mm, the value of t is almost two times more
than that of tc. As increasing the steel rod diameter, the
difference between these two shear stresses decreases. For
the steel diameter 20 mm, the difference between the two
shear stresses is less significant. The maximum shear stress
tmax obtained by using Eq. (24) is varied between 5.9 and
15.1 MPa for all steel rod diameters. As increasing the steel
rod diameter, the value of maximum shear stress decreases.
It is due to the value of tensile stress sf in steel that varies
also with steel diameter. However, the ratio tmax/sf in
Table 1 shows that the value is about constant, 0.05. So,
the maximum shear stress depends also on the tensile stress
of steel rod. It is interesting to indicate that the interfacial
shear stress tc(c), between the steel and the concrete surfaces
without adhesive joint, is very small. Therefore, the effect
of adhesive in improving the shear stress between the steel
and the concrete surfaces is clearly significant.
 ARTICLE IN PRESS
108 L. Bouazaoui, A. Li / International Journal of Adhesion & Adhesives 28 (2008) 101–108
5. Conclusion
The test and theoretical results obtained by this work
give the following conclusions:
 The connections assured by an adhesive joint between
the steel surface and the concrete surface provided
higher joint strengths. The adhesive joint appeared to
significantly improve the shear and tensile stress
distribution along the interface between the steel surface
and the concrete surface.
 The ultimate force depended linearly on the diameter
and the embedded length of the steel rod. On the other
hand, the ultimate force increased parabolically accord-
ing to the bonded surface.
 The first cracks in the concrete appeared at the
embedded end of steel rod and then propagated
transversally and longitudinally.
 The average shear stress estimated did not describe the
distribution of shear stress well. The proposed model in
this work shows that the critical shear stress at the
instance of the first cracks in concrete was about
4.5–6.1 MPa and the ratio between the maximum shear
stress and the applied tensile stress was about 0.05 for
steel rod diameter 12 mm.
References
[1] Chapman RA, Shah SP. Early-age bond of glass fiber rods embedded
in concrete and cement grout. RILEM Mater Struct 1987;26:167–75.
[2] Gao Y-C, Mai Y-W, Cotterell B. Fracture of fiber-reinforced
materials. J Appl Math Phys (ZAMP) 1988;39:550.
[3] Kim JK, Baillie C, Mai Y-W. Interfacial debonding and fiber pull-out
stresses. Part I: critical comparison of existing theories with
experiments. J Mater Sci 1992;27(31):43–54.
[4] Zhou LM, Kim JK, Mai Y-W. Interfacial deboning and fiber pull-out
stresses. Part II: a new model based on the facture mechanics
approach. J Mater Sci 1992;27(31):55–66.
[5] Hsuch CH. Interfacial debonding and fiber pull-out stresses of fiber-
reinforced composites. VII. Improved analysis for bonded interfaces.
Mater Sci Eng 1992;A154:125–32.
[6] Zhandarov SF, Pisanova EV. The local bond strength and its
determination by fragmentation and pull-out tests. Compos Sci
Technol 1997;57:957–64.
[7] Yue CY, Looi HC. Factors which influence the reliability of the
assessment of interfacial bonding in fibrous composites using the pull-
out test. Int J Adhes Adhes 2001;21:309–23.
[8] Wang C. Fracture mechanics of single-fiber pull-out test. J Mater Sci
1997;3:483–90.
[9] Kelly A. Interfacial effects and the work of fracture of a fibrous
composite. Proc R Soc 1970;A319:95.
[10] Banholzer B, Brameshuber W, Jung W. Aanlytical simulation of pull-
out tests—the direct problem. Cem Concr Compos 2005;27-1:93–101.
[11] Greszczuk LB. Theoretical studies of mechanics of the fiber–matrix
interface in composites. In: Interface in composites. Philadelphia, PA:
American Society for Testing and Materials, ASTM STP 452; 1969.
p. 42–58.
[12] Outwater JO, Murphy MC. On the fracture energy of unidirectional
laminates. In: Proceedings of the 24th annual technical conference on
reinforced plastics/composites division. The Society of the Platics
Industry Inc., Section 11-C, 1969. p. 1–8.
[13] ASTM C 24-91a. Standard test method for comparing concretes on
the basis of the bond developed with reinforceing steel. Philadelphia,
PA: Americain Society for Testing and Materials, 1991.
[14] Laws V. Micromechanical aspects of the fibre-cement bond.
Composites 1982;13:145–54.
[15] Li VC. Post-crack scaling relations for fiber reinforced cementitious
composites. J Mater Civil Eng 1990;4(1):41–57.
[16] Piggott MR. Load bearing fibre composites. Oxford, UK: Pergamon
Press; 1980.