You are on page 1of 13

Engineering Fracture Mechanics 74 (2007) 109–121

www.elsevier.com/locate/engfracmech

Digital image correlation analysis of interfacial


debonding properties and fracture behavior in concrete
a,*
David Corr , Matteo Accardi b, Lori Graham-Brady c, Surendra Shah a

a
Northwestern University, McCormick School of Engineering and Applied Science, Department of Civil and Environmental
Engineering, 2145 Sheridan Road, Evanston, IL 60208-3109, United States
b
University of Palermo, Sicily, Italy
c
The Johns Hopkins University, United States

Available online 3 April 2006

Abstract

The use of digital image correlation (DIC) as a fracture mechanics tool is described, for two projects currently under-
way. The goal of the first project is to examine the bond between carbon fiber reinforced polymers (CFRP) and concrete
substrates. The second project involves the interfacial transition zone (ITZ) of plain concrete, and the softening and frac-
ture behavior of this region. For both projects, DIC allows for precise measurement of the surface displacements of the
deforming materials. The resulting strain data are higher in resolution than is possible with other experimental techniques.
For both projects, the DIC results are being used to determine bond constitutive laws, which will be the basis for fracture
mechanics studies of the materials.
 2006 Published by Elsevier Ltd.

Keywords: Image correlation; Interface; Fracture; Debonding

1. Introduction

Digital image correlation (DIC) is a computer vision technique that is used to track the surface displace-
ments of deforming materials. DIC is a fully non-destructive and non-contact measurement tool; all that is
required to perform the technique are digital images of the specimen surface before deformation (reference
image) and after deformation (deformed image) [1]. From these two images, DIC computes the displacement
of a regular grid of points on the specimen surface. DIC has been widely used to examine the deformations of
engineering materials, including concrete and related cement-based materials [2,3].
Two ongoing research projects, focusing on interface properties and utilizing DIC, will be described in this
paper. The goal of the first project is to examine the interfacial fracture behavior between carbon fiber rein-
forced polymer (CFRP) sheets and concrete substrates. The use of CFRP for external strengthening of con-
crete and masonry structures has become widespread in recent years, due to the relative ease and low cost of

*
Corresponding author. Tel./fax: +1 847 467 4105.
E-mail address: d-corr2@northwestern.edu (D. Corr).

0013-7944/$ - see front matter  2006 Published by Elsevier Ltd.


doi:10.1016/j.engfracmech.2006.01.035
110 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

the reinforcement technique [4]. The effectiveness of the technique depends on the mechanical properties of the
bond between the two materials. In addition, in order to analyze and model the behavior of CFRP strength-
ened structures, an analytical model of the bond deformation and fracture properties is essential. In the cur-
rent work, DIC will be used to examine the surface displacements of CFRP sheets as they are loaded in a
single shear pulling test [5].
The second project discussed in this paper centers on the deformation and fracture properties of the inter-
facial transition zone (ITZ) in concrete. The ITZ is a region of cement paste, surrounding the large aggregate
particles, that has a different microstructure and composition than the bulk cement paste [6]. The ITZ forms
due to the packing density and directional formation of hydration products at the interface between aggre-
gates and hydrated cement paste. The resulting microstructure is higher in calcium hydroxide, a crystalline
phase with relatively weak bond properties, and higher in porosity. The result is that the ITZ is a commonly
known as the ‘‘weak link’’ in the composite description of concrete, and is often referred to as a third phase in
the material [7]. The goal of this project is a direct examination of the ITZ through evaluation of the mechan-
ical performance of the bond between aggregate and cement paste. DIC is employed as the central tool for
evaluation of the properties of the bond.

2. Experimental methods and materials

2.1. CFRP bond project

The experimental setup for the CFRP project involves a concrete block as substrate for a carbon fiber sheet
that is bonded to the concrete surface with epoxy. The concrete substrate is a type I normal strength concrete
of size 75 mm by 150 mm by 200 mm, with mix design summarized in Table 1, and compressive and tensile
strength summarized in Table 2. The carbon fiber sheets are supplied by Sika Corporation and have a modulus
of elasticity of 230 GPa. The carbon fiber sheet is attached to the concrete substrate with epoxy using the wet
lay-up technique and no surface pre-treatment, thus forming a carbon fiber reinforced polymer (CFRP). The
epoxy is Sikadur 300, a pore-penetrating epoxy. The CFRP composite has an elastic modulus of 146.5 GPa
and an ultimate strength of 2200 MPa (ASTM D3039 / D3039M-00). The epoxy is cured for at least 72 h
before testing the specimen.
The test setup used for the CFRP project is a single shear pulling test (SPT), with schematic shown in
Fig. 1a, and picture in Fig. 1b. The CFRP sheet is gripped by the upper crosshead of a 20 KN MTS frame,
and the block is restrained by the steel bracing as shown in Fig. 1a. The test is instrumented by measuring
relative global slip between the CFRP sheet and concrete substrate with linearly varying displacement trans-
formers (LVDT). The average of the two LVDTs (on either side of the CFRP sheet) displacements are used as
feedback for the closed-loop system. The experiment is therefore ‘‘slip controlled’’ at a rate of 0.00065 mm/s.

Table 1
Concrete specimen (75 · 150 · 200 mm) and six cylinders
Constituents. Concrete mixture from ACI 211.1 Amount
Portland cement 2200 gr (ratio 1)
Water 1100 gr (ratio 0.5)
Natural sand, 4.75 mm. Nominal maximum size 4400 gr (ratio 2)
Coarse aggregate, 15.875 mm. Nominal maximum size 4400 gr (ratio 2)

Table 2
Mechanical properties of concrete
Average strength Values (MPa)
Compressive strength ðfc0 Þ, from C39/C39M-99 5.4
Splitting tensile strength (T), from (C496-96) 40.4
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 111

Fig. 1. CFRP project: (a) specimen schematic and (b) picture of specimen setup.

2.2. Concrete interfacial fracture project

The specimens used for the interfacial fracture project are two-dimensional ‘‘model concrete’’ specimens. A
schematic of these specimens is shown in Fig. 2a. Granite aggregates are placed in a mold and surrounded by a
mortar with a mixture design of 0.45:1:2.3 (w:c:s). The mortar and granite have moduli of elasticity of 25 GPa
and 60 GPa, respectively. The visible cross-section of the specimen, seen schematically in Fig. 2a in a single-
aggregate specimen, is constant through the 25 mm thickness of the specimen. Due to this constant cross-sec-
tion, the surface deformations of this model concrete specimen will approximate the internal deformations of
the material. This aggregate arrangement was chosen so that there would be a relatively constant level of stress
through the thickness of the sample, across the interface. Fig. 2b shows the imaged surface of a two-aggregate
specimen. Note that the surface of the specimen has been spray-painted to provide a unique pixel pattern to
improve the performance of DIC. The visible surface of the specimen is 75 mm by 75 mm, and the total
specimen dimensions are 200 mm by 75 mm by 25 mm.
The specimens are tested in direct tension [8] by affixing steel plates to the specimen for mounting in a direct
tension grip. The experiment is closed-loop displacement controlled, with the average of two LVDTs mounted
on either side of the specimen as the feedback signal. The displacement rate is 0.0025 mm/min, and the gauge
length the LVDTs measure is 75 mm.

2.3. Digital image correlation

In both experimental testing programs, digital images are acquired continuously as the specimens are
loaded. The digital camera has a resolution of 1024 · 1280 pixels. For the CFRP project, the camera is
mounted approximately 1 m from the specimen, in order to image the entire 75 mm width of the specimen.
At this location, the camera images 100 mm of the CFRP length, sufficient to capture the fracture process zone
and minimum transfer length of the CFRP-concrete bond. For this camera alignment, the pixels in the image
represent an approximately 70 lm square on the specimen surface. For the interfacial fracture project, spec-
imens are imaged with the camera in two locations. First, the same camera location as the CFRP project is
112 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

Fig. 2. Interface fracture project: (a) single-aggregate specimen schematic and (b) double-aggregate specimen image.

used to examine the full-field of specimen displacement. The second camera location is closer to allow high-
resolution images to be acquired at the aggregate–mortar interface; here pixels represent approximately 5.2 lm
squares. For both projects, images are acquired once per second until the specimen fails, or until 1000 speci-
mens are stored.
Digital image correlation (DIC) compares the deformed images acquired during loading to reference
images acquired before loading begins. DIC proceeds as follows:

1. Apply a regular grid of nodes (e.g. 10 pixels apart) to the reference image.
2. For each node, identify a ‘‘subimage’’ of fixed pixel size (e.g. 40 pixels).
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 113

3. In the deformed image, compare the subimage to windows of the same size. This is done by calculating the
pixel intensity correlation coefficient:
PP
i j F ði; jÞT ði  m; j  nÞ
rðm; nÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
PP ffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
PP ð1Þ
2 2
i j ½F ði; jÞ i j ½T ði  m; j  nÞ

where F and T are the pixel greyscale values in the reference and deformed images, respectively, and r(m, n)
is the correlation parameter at location (m, n).
4. Move the window through a defined area of the deformed image, and calculate the correlation coefficient
for each location.
5. The location with the highest correlation is the new location of the node. Subtract the new location from the
old to determine the displacement of the node.
6. Repeat steps 2–5 for all nodes in the reference image.

This procedure can be repeated for each image acquired during loading; the reference (unloaded) image is the
same for each analysis. DIC is capable of determining displacements with an accuracy of 1/20 pixel [3]. This
corresponds to ±3.5 lm for the low-resolution images, and ±0.25 lm for the high-resolution images. Because
the gauge length is related to the size of the pixel in the original image, the accuracy of the strain measurements
across the entire image is approximately 50 microstrain. There is a low level of noise in the displacement cal-
culations; this noise is smoothed using bi-cubic spline smoothing in MATLAB. More details on the develop-
ment of DIC and its use in analysis of displacement and strain can be found in the literature [9–11].
Using the smoothed displacements, strain can be calculated by defining a square element consisting of 4
nodes at the corners, and using a finite element approximation:
e ¼ Bu ð2Þ
where e is the 2D plane strain vector, B contains linear interpolation shape functions and u are the 8 nodal
orthogonal nodal displacements. It should be noted that there are other possibilities for strain interpolation,
including higher-order elements consisting of 9 or more nodes. However, since the displacement measures are
smoothed using bi-cubic spline interpolation before the strain calculation, unlikely that these higher-order
strain interpolations will yield further improvement in strain accuracy.

3. Results and discussion

3.1. CFRP project

The single shear pulling tests show two different mechanism of failure: concrete failure and concrete-to-
adhesive interfacial failure. Fig. 3 shows a sample load vs. time relationship for a CFRP specimen typical
of the first failure mode. The specimen shows a linear-elastic portion of the response, followed by a non-linear
curve, and then a gradual hardening as fracture propagates a few millimeters beneath concrete/adhesive inter-
face. The peak load of the specimen is of interest, particularly as it pertains to the width of the CFRP sheet.
Fig. 4 shows a plot of peak load vs. CFRP width for a number of specimens in this research program. Also
shown in this figure are a number of theoretical models that have been proposed to examine the size effect of
the maximum load. Note that there is a wide range of predicted maximum loads for these models; the experi-
mental results tend to agree with the lower predicted maximum loads.
For a fracture mechanics analysis of the bond between CFRP and concrete, a bond constitutive law, relat-
ing local slip to bond stress at the interface, is essential. The local slip and bond stress can be calculated as
follows:
Z
si ðxÞ ¼ eðxÞdx ð3Þ
i

ECFRP ACFRP ðeiþ1  ei Þ


sðxÞ ¼ ð4Þ
bFRP Dx
114 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

310
7.0
230
6.0

5.0
160

Load (kN)
4.0 460

3.0

2.0
184
1.0

0.0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300
Time (sec.)

Fig. 3. CFRP load vs. global slip (elapsed time in seconds marked).

Fig. 4. Maximum load vs. CFRP width (see also [4,16–18]).

Here, si(x) is local slip, s(x) is bond stress, and e(x) is the smoothed axial strain calculated from DIC. ECFRP,
ACFRP and bCFRP are the modulus, cross-sectional area, and width of the CFRP sheet, respectively, and Dx is
the length between two measured points. Use of Eqs. (3) and (4) requires some assumptions: the concrete sub-
strate is rigid in comparison to the CFRP and that the CFRP remains linear elastic. Fig. 5 shows contour
maps of bond stress for the 4 points indicated in Fig. 3. The width bond stress ‘‘ridge’’ in the contour plots
is the anchor length, across which all of the stress is transferred from the CFRP to the substrate. Note the
movement of this ridge as fracture propagates through the bond.
Taking a line along the length of the CFRP sheet and plotting bond stress for a number of load steps gives
an indication of fracture propagation rates. This is plotted in Fig. 6 for a few load steps along the centerline of
the CFRP sheet; similar results have been observed for other lines not in the center. Note that the peak stress
increases at the same location (15 mm) until 182 s, at which time the interfacial crack begins to propagate.
Then, the peak stress begins to move down the length of the CFRP sheet. The anchor length is defined as the
length over which the bond stress rises from a nominal level to the peak and drops back to zero. Note that
the anchor length does not change once the fracture begins to propagate along the length of the CFRP sheet.
The crack tip is probably in the vicinity of the bond stress peak, although its precise location cannot be
determined from this data.
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 115

Fig. 5. Bond stress contour maps.


116 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

9.0

8.0

7.0

6.0

Bond Stress (MPa) 5.0

4.0

3.0

2.0

1.0

0.0
0 10 20 30 40 50 60 70 80 90 100
-1.0
Distance from Loaded CFRP end (mm)
160 seconds 182 seconds 184 seconds 275 seconds 620 seconds

Fig. 6. Bond stress profiles at indicated elapsed times.

7
184

6
Bond Stress (MPa)

4 230

2 265 660
310 460
1
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
-1
Slip (mm)

Fig. 7. Bond constitutive law (elapsed in seconds indicated).

Taking a single point on the CFRP surface and plotting local slip vs. bond stress for a number of load steps
established the bond constitutive law. A sample bond constitutive law is shown in Fig. 7. The curve peaks at
approximately 7 MPa at a slip of 0.025 mm, and then drops to a constant level of approximately 1 MPa. This
bond constitutive law is in agreement with other experimental laws available in the literature [12–14]. One area
of disagreement is the 1 MPa constant stress once the crack front has passed. This measurement is repeatable
throughout our studies but does not specifically indicate a bond property. The source of this residual stress has
not been established, although it could be caused by a ‘‘complementary friction’’ effect on the fractured surface
which originates from variability in the alignment of CFRP sheet with respect to the loading direction [15].

3.2. Interface fracture

Two aggregate arrangements are examined in the interfacial fracture project: a single aggregate, 1.500 wide
by 100 deep (visible), and a two-aggregate specimen, with side-by-side 0.7500 wide by 100 deep aggregates,
separated by 0.500 . The aggregate volume fraction for these two specimens is the same. Fig. 8 shows a sample
bulk stress vs. bulk strain (as calculated from the load cell and LVDT data) for a single-aggregate specimen.
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 117

2.0

1.5
Stress (MPa)

1.0

0.5

0.0
0 50 100 150 200 250 300 350 400 450 500
Strain (µstrain)

Fig. 8. Sample stress–strain curve for interfacial fracture study.

Table 3
Summary of direct tension specimens
Specimen Aggregate type E (MPa) Peak r (MPa) e at Peak (lm/mm)
dh1 Single 28692 2.42 106.34
dh2 Single 29469 1.73 48.82
dh3 Single 20447 1.41 61.75
dh4 Single 23277 1.44 65.09
dh5 Single 27447 1.48 56.85
dh7 Single 31318 2.24 66.21
dh8 Single 29680 1.28 21.40
dh9 Single 28759 1.54 54.84
dh10 Single 26506 1.66 63.09
dh12 Single 28341 1.97 90.06
dh13 Single 24303 1.75 73.57
d6 Single 17172 1.21 66.65
d7 Single 22394 1.28 35.00
d8 Single 22450 1.28 26.75
d9 Single 26093 1.89 63.31
d10 Single 28557 2.43 88.95
d11 Single 29327 2.74 82.71
d12 Single 27674 1.88 68.66
d13 Single 25126 1.32 50.83
d14 Single 26573 1.87 72.90
d15 Single 22990 1.58 94.07
d16 Single 27080 2.00 68.66
d17 Single 25552 2.11 80.70
e1 Double 27592 2.09 79.36
e2 Double 27069 1.75 65.76
e3 Double 26975 1.54 52.83
e4 Double 27090 2.05 79.36
e5 Double 24179 2.26 103.21
e6 Double 24742 1.40 45.92
e7 Double 26738 2.49 86.72
Mean 26120 1.80 67.35
Standard deviation 3017 0.42 20.42
Coefficient of variation 0.12 0.23 0.30
118 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

The scatter in the data is due to the precision of the closed-loop control system and the precision of the LVDT
feedback signals. Because the aggregate volume fractions of the two specimen types are the same, Fig. 8 is a
typical response for either of these types of specimens. This specimen shows a peak stress of approximately
1.54 MPa at a strain of 54.84 microstrain. The peak stress and peak strain for all tested specimens is summa-
rized in Table 3. The average peak stress is 1.80 MPa, and strain at peak of 67.35 microstrain, with coefficients
of variation of 0.23 and 0.30, respectively.
Fig. 9 shows contour plots of axial strain for an entire two-aggregate specimen using the low-resolution
images and DIC. Fig. 9a was acquired near the peak stress, and Fig. 9b was acquired well into the post-peak
of the specimen. Note that small concentrations of strain are visible at the tops and bottoms of the aggregates
in Fig. 9a. The strains are larger at the tops of the aggregates, indicating that the failure macrocracks may have
already formed in this image and begun to dominate the microcracks below the aggregates. In Fig. 9b, the
failure crack is clearly visible above the aggregates. It is important to also note that the failure crack clearly
forms at the aggregate interface first, then propagates out to the sides of the specimen.
Fig. 10 shows the results of high-resolution images of a single-aggregate specimen, with 3D mesh plots of
axial strain for the specimen with stress–strain curve shown in Fig. 8. The camera was positioned to image the

Fig. 9. Low resolution, full-specimen strain maps: (a) pre-peak and (b) post-peak.
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 119

Fig. 10. High-resolution interfacial strain: (a) 40 s, (b) 80 s, (c) 112 s and (d) 113 s.

-214

-215

-216
y-displacement (microns)

-217

-218

-219

-220

-221

-222

-223

-224
0 10 20 30 40 50 60 70 80 90
y position (pixels)

Fig. 11. Displacement profile near peak load.


120 D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121

40

35

30

crack opening (microns) 25

20

15

10

-5
0 100 200 300 400 500 600 700 800
elapsed time (seconds)

Fig. 12. Crack opening vs. load step.

aggregate–mortar interface. Fig. 10a is at approximately 35% of peak stress, Fig. 10b at 65%, Fig. 10c at the
peak stress, and Fig. 10d immediately after the peak. The initiation and growth of the interfacial crack is evi-
dent; even in Fig. 10a, a concentration of strain is apparent at the interface. It is important to note that no
macrocrack is visible in the images associated with Fig. 10a, and the increased strain near the interface is likely
due to distributed microcracking in the interfacial zone.
As a macrocrack forms at the aggregate–mortar interface, the axial strain measurement becomes less
meaningful; the strain estimation becomes dependent on the distance between nodes. Instead, we examine
the width of the crack opening measured directly from the DIC displacement measurements. Fig. 11 shows
a profile of vertical displacement along a line perpendicular to the interface at the peak stress of Fig. 8. The
quick jump in the vertical displacement is due to the crack opening, and subtracting the beginning and end-
ing values of this jump gives a measurement of the crack width. Fig. 12 shows the crack opening width vs.
elapsed time. The crack location is determined by looking at images late in the analysis where the failure
crack is visible; this determines a location, usable in all images for that specific test, which defines the loca-
tion of the crack. The crack opening is very small until the peak stress where the crack width jumps
abruptly and then increases more quickly. In all tests, the crack formed directly and the interface between
the aggregate and mortar phases. Work is underway to obtain an accurate approximation of the stress at
the interface using the known elastic modulus of the mortar matrix and measurements of strain in the mor-
tar immediately in the vicinity of the interface, using either strain gauges or DIC calculations. Once this
stress approximation is obtained, a crack stress vs. crack opening interfacial constitutive law can be
determined.

4. Conclusions

Digital image correlation (DIC) is an effective tool for the determination of fracture behavior of materials
through the examination of surface displacements. Specifically, projects are currently underway to study the
CFRP-concrete bond properties, and to study the interfacial transition zone at the aggregate–cement paste
bond in concrete. The CFRP project has yielded information on bond stress, local slip, fracture propagation,
and an experimental constitutive law for the CFRP-substrate interface. The interface fracture project has
shown the distributed softening at the interface that occurs ahead of macrocrack formation. Further research
towards a bond constitutive law for the interface region is currently underway.
D. Corr et al. / Engineering Fracture Mechanics 74 (2007) 109–121 121

Acknowledgements

We would like to thank the National Science Foundation (NSF grant CMS-0332356, Program Director P.
Balaguru) for supporting this work. We would also like to recognize Nathan Tregger and Sandy Yoon, stu-
dents at ACBM, for their help in data collection and analysis in the interfacial fracture project.

References

[1] Sutton MA, McNeill SR, Helm JD, Chao YJ. Advances in two-dimensional and three-dimensional computer vision. Top Appl Phys
2000;77:323–72.
[2] Choi S, Shah SP. Measurement of deformation on concrete subject to compression using image correlation. Exp Mech 1997;37(3):
307–13.
[3] Lawler JS, Keane DT, Shah SP. Measuring three-dimensional damage in concrete under compression. ACI Mater J 2001;98(6):
465–75.
[4] Neubauer U, Rostasy FS. Design aspects of concrete structures strengthened with externally bonded CFRP plates. In: Proceedings of
the seventh international conference on structural faults and repairs, Edinburgh, Scotland: ECS Publications; 2, p. 109–18.
[5] Horiguchi T, Saeki N. Effect of test methods and quality of concrete on bond strength of CFRP sheets. In: Proceedings of the third
international symposium, Sapporo: Japan Concrete Institute, 1, p. 475–82.
[6] Mitsui KZ, Li Z, Lange DA, Shah SP. Relationship between microstructure and mechanical properties of the paste–aggregate
interface. ACI Mater J 1994;91(1):30–9.
[7] Nilsen AU, Monteiro PJM. Concrete: a three phase material. Cem Concr Res 1993;23:147–51.
[8] Shah SP, Choi S. Nondestructive techniques for studying fracture processes in concrete. Int J Fract 1999;98:351–9.
[9] Peters WH, Ranson WF. Digital imaging techniques in experimental stress analysis. Opt Engng 1982;21(3):427–31.
[10] Sutton MA, Cheng M, Peters WH, Chao YJ, McNeill SR. Application of an optimized digital image correlation method to planar
deformation analysis. Imaging Vision Comput 1986;4(3):143–51.
[11] Helm JD, McNeill SR, Sutton MA. Improved 3D image correlation for surface displacement measurement. Opt Engng 1996;35(7):
1911–20.
[12] Ali-Ahmad M, Subramaniam K, Ghosn M. Fracture analysis of the debonding between FRP and concrete using digital image
correlation. In: Li, et al. editors, Fracture mechanics of concrete structures, IA-Framcos, 2, 2004. p. 787–93.
[13] Chajes MJ, Finch WW, Januszuka TF, Thomson TA. Bond and force transfer of composite-material plates bonded to concrete. ACI
Struct J 1996;93(2):209–17.
[14] De Lorenzis L, Miller B, Nanni A. Bond of fiber-reinforced polymer laminates to concrete. ACI Mater J 2001;98(3):256–64.
[15] Accardi M. Strengthening of masonry structural elements subjected to out-of-plane loads using CFRP reinforcements. PhD thesis,
University of Palermo, Sicily 2004. p. 150–5.
[16] Brosens K, Van Gemert D. Anchoring stresses between concrete and carbon fibre reinforced laminates. Non-metallic (FRP)
reinforcement for concrete structures. Proceedings of the third international symposium, 1. Sapporo: Japan Concrete Institute; 1997.
p. 271–8.
[17] Chen JF, Tang JG. Anchorage strength models for FRP and steel plates bonded to concrete. J Struct Engng 2001;127(7).
[18] Wu Z, Yuan H, Niu H. Stress transfer and fracture propagation in different kinds of adhesive joints. J Engng Mech 2002;128(5).

You might also like