ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 105-S05

Experimental Study of Intermediate Crack Debonding

in Fiber-Reinforced Polymer Strengthened Beams
by Owen Rosenboom and Sami H. Rizkalla

Interface crack propagation of fiber-reinforced polymer (FRP)- axial stiffness per unit width; 3) strengthening plate termination
strengthened reinforced or prestressed concrete flexural members in regions near the support; or 4) anchorage details. Prestressed
is often initiated from the toes of the intermediate cracks and concrete beams are prone to IC debonding failures because of
propagates toward the supports. This type of FRP delamination is their high shear span-to-depth ratios. They will also have
commonly termed intermediate crack (IC) debonding and is common
slightly improved bond behavior compared with reinforced
for flexural members with high shear span-to-depth ratios. This
paper describes an experimental program where six 30 ft (9.14 m) concrete members due to the existing compressive strain in
long prestressed concrete bridge girders were tested monotonically the beam soffit. This proves to be beneficial because the IC
to failure to evaluate the bond characteristics of carbon FRP debonding propagation is typically due to failure in the
(CFRP) strengthening systems. Four of the beams failed due to IC concrete surface material. Conversely, beams with short
debonding, one failed due to FRP rupture, and the unstrengthened shear span-to-depth ratios or strengthening plates with high
control girder failed due to concrete crushing. The results of this stiffness or thickness may experience debonding propagating
study were combined with the results from other tests in the literature from the plate end (PE debonding), which has been examined in
to create a useful experimental database. The database was used many analytical studies.4-7 Debonding mitigation techniques
to assess the analytical models currently available in the national have been explored by several researchers8-10 and have been
code documents. The analysis indicated that the current models do
codified in China.11
not correlate well with the experimental database and the need for
a new analytical model is highlighted. This paper briefly discusses This study has three main objectives: 1) to create a
a proposed model for IC debonding that predicts the measured description of an experimental program designed to assess
values tested by other researchers well. the bond behavior of CFRP strengthening systems for
prestressed concrete; 2) assemble an experimental data-
Keywords: bond; cracking; debonding; fiber-reinforced concrete; prestressed. base of IC debonding failures from sources in the literature;
and 3) use the measured values and the database to assess the
INTRODUCTION accuracy of the current models of the national code documents
Strengthening of existing structures using lightweight in predicting IC debonding failure. The accuracy of a
composite materials is becoming widespread due to their proposed model by the authors is also presented.
ease of installation and competitive pricing compared with
traditional methods. Reinforced or prestressed concrete RESEARCH SIGNIFICANCE
beams strengthened with fiber-reinforced polymer (FRP) This study provides unique data on the bond behavior
materials often fail in flexure due to concrete crushing or using large-scale 30 ft (9.14 m) long prestressed concrete
FRP rupture.1 This type of failure can be well predicted bridge girders strengthened with FRP, which complements
using cracked section analysis of the strengthened earlier studies using small-scale beams. The experimental results
section using the specified FRP material properties from the presented in this paper, along with the assembled database, are
manufacturer.2 A detailed examination of the flexural useful for the development of rational models and design
behavior of prestressed concrete beams strengthened with procedures that can be used by ACI Committee 44012 to provide
FRP materials can be found elsewhere.2 The behavior under up-to-date guidelines for the repair and strengthening of
fatigue loading of FRP strengthened beams is often controlled by concrete structures using FRP.
the stress range induced in the internal steel reinforcement,
which should be kept within prescribed limits.3
The presence of an FRP strengthening material bonded to
the tension face of a reinforced concrete beam will restrict
but not prevent the opening of intermediate flexural or shear
cracks due to applied loading. Displacements at the toe of the
flexural cracks create stress concentrations at the interface of
the plate and the beam, leading to the development of localized
interface cracking. At higher load levels, the interface cracks
propagate between the flexural cracks and move toward the Fig. 1—Intermediate crack debonding process.
end supports for a simply supported beam as shown in
Fig. 1. This type of FRP delamination is commonly termed
intermediate crack (IC) debonding whether the propagation ACI Structural Journal, V. 105, No. 1, January-February 2008.
MS No. S-2006-358.R2 received February 20, 2007, and reviewed under Institute
is from intermediate flexural cracks or intermediate shear publication policies. Copyright © 2008, American Concrete Institute. All rights reserved,
cracks. It is a failure mode common in structures with: 1) high including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the November-
shear span-to-depth ratios; 2) FRP strengthening with low December 2008 ACI Structural Journal if the discussion is received by July 1, 2008.

ACI Structural Journal/January-February 2008 41

 bridges in Eastern North Carolina. The 30 ft (9.14 m) long
ACI member Owen Rosenboom is a Post-Doctoral Fellow at Hong Kong Polytechnic
University, Hong Kong, China. He received his PhD from North Carolina State girders were prestressed with ten 250 ksi (1724 MPa) 7/16 in.
University, Raleigh, NC, in 2006. He is a member of ACI Committee 440, Fiber-Reinforced (11.1 mm) diameter stress-relieved seven-wire prestressing
Polymer Reinforcement, and Joint ACI-ASCE Committee 423, Prestressed Concrete. strands. Five strands were in each web, the top three of which
Sami H. Rizkalla, FACI, is Distinguished Professor of Civil and Construction were harped with a hold-down point located at the midspan.
Engineering in the Department of Civil, Construction and Environmental Engineering, Details of the girder reinforcing are shown in Fig. 2. Two
North Carolina State University. He is also the Director of the Constructed Facilities types of CFRP material was used in this research: precured
Laboratory and NSF I/UCRC in Repair of Structures and Bridges at North Carolina
State University. He is a member of ACI Committees 118, Use of Computers; 440, strips and wet lay-up sheets. Girders EB1S, EB1SB, and
Fiber Reinforced Polymer Reinforcement; E803, Faculty Network Coordinating EB1SB2 were all strengthened with one 2 in. (51 mm)
Committee; and Joint ACI-ASCE Committees 423, Prestressed Concrete, and 550,
Precast Concrete Structures.
precured strip per web. Girder EB8SB was strengthened
with two 2 in. (51 mm) sheets per web and Girder EB9SB
was strengthened with four 2 in. (51 mm) sheets per web. All
EXPERIMENTAL PROGRAM of the strengthened girders had debonding mitigation,
A total of six girders were tested as part of an experimental consisting of 10 transverse CFRP wet lay-up U-wraps (five
study. Three girders were strengthened with precured carbon on each web) on one side only to promote debonding on the
FRP (CFRP) strips (EB1S, EB1SB, and EB1SB2) and two instrumented side of the girder, except Girder EB1S, which
girders were strengthened with CFRP wet lay-up sheets had the same debonding mitigation placed throughout the
(EB8SB and EB9SB). An unstrengthened control girder (CS) length of the girder. Details of the CFRP strengthening are
was also tested to examine the effectiveness of the various shown in Fig. 2 and summarized in Table 1. Four of the
strengthening schemes used in this investigation. Table 1 beams were purposefully designed without U-wrap
provides a summary of all six tested girders. Detailed anchorage on one side to study the IC debonding failure
information on the test results can be found elsewhere.2 mechanism and not to achieve specific increases in load-
carrying capacity. All materials were tested to determine
Test girders their stress-strain behavior in accordance with the appropriate
The precast prestressed C-channel type bridge girders ASTM standard and summarized results are shown in Table 1.
tested in this study were obtained from two decommissioned
Test setup and instrumentation
Table 1—Summary of material properties All girders were tested using a 490 kN MTS hydraulic
and experimental results actuator mounted to a steel frame placed at the midspan of
the girder. The load was applied using a steel plate
Speciman
designation CS EB1S EB1SB EB1SB2 EB8SB EB9SB approximately 10 x 20 in. (250 x 510 mm [the same area as
Concrete strength,
the design loading area specified by AASHTO]).13 To
10,720 8890 5410 5180 7930 6720
psi (MPa)* (73.9) (61.3) (37.3) (35.7) (54.7) (46.3)
Modulus of elasticity 23.2 23.6 23.6 10.8 10.8
—
of FRP, Msi (GPa)* (160) (163) (163) (74.4) (74.4)
Ultimate strength of 400 383 383 110 110
—
FRP, ksi (MPa)* (2758) (2640) (2640) (758) (758)
Rupture strain of FRP,
— 17,300 16,300 16,300 11,100 11,100
µm/m*
Area of FRP, 0.20 0.20 0.20 0.34 0.68
—
in.2 (mm2) (130) (130) (130) (218) (437)
Yield strength of 207 207 212 212 210 210
prestressing, ksi (MPa)* (1430) (1430) (1464) (1464) (1451) (1451)
Ultimate strength of 230 230 263 263 263 263
prestressing, ksi (MPa)* (1590) (1590) (1812) (1812) (1812) (1812)
Cracking load, 12.6 12.6 13.0 12.9 12.5 13.0
kips (kN) (56) (56) (57.8) (57.6) (55.6) (57.8)
Prestress losses, % 15.3 15.9 15.4 16.6 19.3 17.0
Effective prestress, 16.0 16.2 16.0 15.9 15.2 15.7
kips (kN) (71.2) (72.1) (71.2) (70.6) (67.8) (69.8)
Ultimate load, 33.2 39.6 36.4 36.2 34.5 40.1
kips (kN) (147.7) (176.1) (162) (161) (153.5) (178)
Percent increase in
— 19.3 9.2 8.6 3.6 20.2
capacity
Ultimate concrete
compressive strain, 3000 N/A 1186 1301 1262 1163
µm/m
Ultimate CFRP
— 12,200 9960 10,670 9490 8680
tensile strain, µm/m
Exp/material test
— 0.71 0.61 0.65 0.86 0.79
tensile strain in CFRP

Failure mode† CC IC IC IC Rupture IC then

rupture
*
From material testing.
†CC = concrete crushing; rupture = rupture of CFRP near midspan; and IC = intermediate Fig. 2—Girder reinforcing and CFRP strengthening
crack debonding. configurations (Note: 25.4 mm = 1 in.).

42 ACI Structural Journal/January-February 2008

simulate field conditions, the girder was supported at both Girder EB8SB failed due to rupture of the FRP at a load of
ends on neoprene pad, which in turn rested on a steel plate. 34.5 kips (153.4 kN), slightly lower than Girders EB1SB and
The width of the neoprene pad was 8.5 in. (216 mm), which EB1SB2, which were strengthened with the same axial stiffness
provided a clear span of 28.6 ft (8712 mm) for each tested of FRP. The rupture occurred in the longitudinal CFRP at the
girder. The girders were tested first up to the cracking load, location of the first U-wrap. Just before failure, cracking was
unloaded, and reloaded to failure. This allowed observation heard at the FRP-concrete interface near the midspan
of the crack reopening load, which aided in the computation representing localized debonding around the flexural cracks.
of the effective prestress force.1 The displacement behavior Inspection after the test revealed that the interface failure
of the girders during testing was monitored using two string plane before rupture was at the FRP-concrete interface in the
potentiometers placed at the midspan and two linear potentiom- concrete surface layer. This failure plane developed between
eters to measure vertical displacement over the supports. The the concrete teeth formed between the flexural cracks at
compressive strain in the concrete was measured using a midspan. The maximum recorded strain in the FRP before
combination of PI gauges (a strain gauge mounted to a spring rupture was 9490 μm/m similar to Girders EB1SB and
plate) and two electric resistance strain gauges located at EB1SB2, but much closer to the rupture strain measured
midspan. PI gauges were placed at various locations at the from material testing.
level of the lowest prestressing strand to measure the crack Girder EB9SB failed due to IC debonding followed by
width and to determine the strain profile along the depth of the FRP rupture at a load level of 40.1 kips (178.3 kN). Inspection
girder. The tensile strain profile in the CFRP reinforcement was after the test revealed that the failure plane was in the FRP-
measured using numerous electric resistance strain gauges concrete interface in the concrete surface layer throughout
placed at various locations along the length of the girder.

Test results
Before strengthening, the prestressed concrete girders
used in this research were similar. Girders EB1S, EB1SB,
EB1SB2, and EB8SB were strengthened with approximately
the same axial stiffness Ef Af of FRP material and, therefore,
performed much the same before and after cracking and after
yielding of the prestressing strands. Because only Girder EB1S
was strengthened with debonding mitigation throughout the
entire length, this indicates that the presence of the U-wraps
did not influence flexural cracking. Girder EB9SB was
strengthened with approximately double the CFRP axial
stiffness of the other girders; therefore, the stiffness was greater
both before and after yielding of the prestressing strands.
Spacing of the cracks around midspan was approximately
8 in. (203 mm) for the control girder (CS) and approximately
6 in. (152 mm) or smaller for the strengthened girders, which
also had numerous other branching cracks around the toes of
the flexural cracks.
Girder EB1S failed due to IC debonding at a load of
39.6 kips (176.1 kN), a 19.3% increase in ultimate load
compared with the control girder (CS). The IC debonding
occurred rapidly and developed despite the presence of
U-wraps throughout the length of the girder. The failure
interface was in the concrete interface at midspan, but was
difficult to discern between the U-wraps in other locations.
The maximum measured strain of the CFRP material at
midspan was 12,200 μm/m, which is 71% of the rupture
strain as determined from material tests. Girders EB1SB and
EB1SB2 failed due to IC debonding at measured strain
values of 9960 and 10,670 μm/m, lower than the value for
Girder EB1S. This indicates that the presence of the U-wraps
throughout the length of the girder delayed the IC debonding
failure. For both girders, the debonding initiated near the
midspan on the side without U-wraps and propagated rapidly
towards the supports. Initially, the failure occurred at the
FRP-concrete interface in the concrete surface layer and then
shifted to the CFRP-adhesive interface outside of the heavily
cracked region. Figure 3 shows the rapid debonding propagation
observed during the test of Girder EB1SB. The measured
ultimate loads were nearly the same for both girders and
represented approximately an 8% decrease from Girder
EB1S with U-wraps along the entire span. Fig. 3—IC debonding failure of Girder EB1SB.

ACI Structural Journal/January-February 2008 43

the debonded length, except for near midspan where the Discussion of test results
formation of small concrete teeth between the branching This section discusses the structural behavior of the six
cracks led to damage at the CFRP-adhesive interface before girders tested as part of the experimental study. A summary
failure. The maximum measured strain at midspan was of the results is provided in Table 1.
8680 μm/m, lower than the other strengthened girders. Girder EB1S had debonding mitigation placed along the
entire length of the girder, which reduced the propensity for
debonding failure and increased the ultimate load and
displacement at failure due to IC debonding. For the other
strengthened girders, the ultimate displacements were lower
than the control girder due to premature debonding failures,
which was designed intentionally to study the failure
mechanism. The applied load-versus-midspan displacement
is shown in Fig. 4 for the all the strengthened girders, along
with the unstrengthened girder. The girder with debonding
mitigation (EB1S) showed considerably more ductility than
the other strengthened girders. The girder with the largest
axial stiffness of FRP (EB9SB) achieved the highest
measured load value, which is slightly higher than Girder EB1S
and 20.2% higher than the unstrengthened girder.
From the continuous monitoring of strain gauges attached
to the CFRP material, the strain profile along the length of
the girder was established for various load levels. The strain
profile for Girder EB1SB is shown in Fig. 5, where the dark
Fig. 4—Applied load versus midspan displacement. lines represent the load of flexural cracking and debonding
failure. The figure indicates that the tensile strain in the
CFRP material at service load levels, which is approximately
the cracking load, are considerably below the strains
measured at debonding failure. The debonding mitigation
provided on one side of Girder EB1SB (the right side of the
figure) indicates that the transverse U-wraps did not affect
the tensile strain profile of the longitudinal CFRP; therefore,
in subsequent tests, the strain gauges were grouped on the
side without U-wraps. Girders EB1SB2 and EB8SB had
similar axial stiffnesses of longitudinal CFRP resulting in
similar strain distributions. The strain profile of Girder
EB9SB, which had twice the axial stiffness of the other
girders tested, is shown in Fig. 6. Comparing the strain
profiles of Girders EB1SB and EB9SB suggests that
externally bonded wet lay-up type systems can have a
slightly more uneven (or jagged) distribution of strain due to
the conformation of the strengthening material to the soffit of
Fig. 5—Tensile strain in CFRP versus length along girder the beam or a result of the increased amount of instrumentation.
for Girder EB1SB. The interface shear stress [τ(x)] along the length of the
girder corresponding to the measured tensile strain in the
CFRP can be evaluated using the following equation

d
τ ( x ) = ------ [ K p ( x )ε p ( x ) ] (1)
dx

where d/dx[εp(x)] is the change in FRP strain along the

length of the beam x and Kp is the axial stiffness of the CFRP
material per unit width. Based on the slope of the measured
tensile strain and the axial stiffness of the material, the interface
shear stress along the length of the beam can be determined.
The interface shear stress versus length for Girders EB1SB
and EB9SB is shown in Fig. 7 and 8, respectively. The more
uniform distribution of measured strain for the externally
bonded FRP precured strips provided positive interface
shear stress distribution over half the girder with a maximum
value of 145 psi (1.0 MPa) as shown in Fig. 7. The interface
shear stress distribution for Girder EB9SB has a much more
Fig. 6—Tensile strain in CFRP versus length along girder variable distribution with a maximum positive value of
for Girder EB9SB. approximately 340 psi (2.34 MPa) and a negative value of

44 ACI Structural Journal/January-February 2008

100 psi (–0.69 MPa). The large variability is mainly due to large drop-off in strain that occurs just away from the CFRP
complex cracking behavior of reinforced/prestressed concrete directly below the loading location.
and the resulting stress concentrations occurring around the The variation of the reported debonding strain value could
toes of the flexural cracks. The distribution of the interface also be due to the time at which the strain was measured. In
shear stress, in particular Fig. 7, indicates that the some cases, researchers report the strain at the initiation of
maximum interface shear stress at debonding failure is likely debonding and other just before IC debonding failure, a
to occur at the location along the girder corresponding to first difference that could be as much as 230%.14 Due to the wide
yielding of the prestressing strands, in this case, approximately variation in reported debonding strain, some have tried back-
10.5 ft (2790 mm) from the support. calculation of the debonding strains using strain compatibility
The average measured debonding strain for Girders EB1SB, and equilibrium and the measured values for applied load14;
EB1SB2, and EB8SB that were strengthened with the same however the results of this exercise were inconclusive.
axial stiffness of FRP was approximately 10,000 μm/m as Out of a total of 38 experimental studies, only 17 had one
shown in Table 1. Using simple cracked section analysis or more specimens satisfying the previous requirements,
(described in detail elsewhere1) that satisfies equilibrium and resulting in a total of 47 beams or slabs included in the data-
compatibility of the section for the strengthened prestressed base. Including the four prestressed concrete C-channels
concrete girders and debonding strain of 10,000 μm/m, tested as part of this research that satisfy the requirements,
the load versus deflection behavior was predicted as shown there are a total of 51 beams or slabs reported in the data-
in Fig. 9. The figure indicates that if the debonding strain is base. The selected members represent a wide cross section of
known, the behavior of the girders can be well predicted. shapes and sizes, with depths varying from the lower limit of
5.9 in. (150 mm) up to 32.5 in. (825 mm). There are five T-beam
INTERMEDIATE CRACK DEBONDING DATABASE sections and spans ranging from 59 to 360 in. (1500 to
There have been numerous experimental studies examining 9140 mm). The shear span-to-depth ratio of the tested beams
the intermediate crack debonding phenomenon. In construction varies from 2.65 (near the lower prescribed limit) to 10.04. The
of the IC debonding database presented herein, reported test reinforcement ratio varies from 0.34 to 2.7%. Some of the
results were not included if they had any of the following beams were heavily reinforced for shear, and others had no
characteristics:
1. Beams less than 5.9 in. (150 mm) in depth;
2. Beams with a shear span-to-depth ratio less than 2.5;
3. Plate-end debonding;
4. Beams with anchorage details;
5. Studies where no debonding strains were provided; and
6. Cantilever beams or slabs, beam-end, or other specimens.
It should be noted that most of the reported strains were
measured from electrical resistance strain gauges placed on
the outside surface of the CFRP after strengthening normally
within the constant moment region or high moment regions.
Due to induced stress concentrations on the externally
bonded plates due to flexural cracking, these reported strains
could vary significantly, even if they are placed very close
together. For in the test configuration of four-point bending,
the most reliable reported debonding strains are those that
are averaged over the constant moment region. For the test
configuration using three-point bending, the most reliable
value is the highest recorded value at midspan due to the Fig. 8—Interface shear stress versus length along girder for
Girder EB9SB.

Fig. 7—Interface shear stress versus length along girder for Fig. 9—Analysis versus experimental for Girders EB1SB,
Girder EB1SB. EB1SB2, and EB8SB.

ACI Structural Journal/January-February 2008 45

shear reinforcement. The database includes three different reinforced or prestressed concrete members, although this
FRP materials: CFRP, aramid FRP (AFRP), and glass FRP value was of a small magnitude.
(GFRP), which were installed using both the wet lay-up ACI Committee 44012 proposes the use of a bond reduction
method and externally bonding a precured laminate using factor κm multiplier to determine the debonding strain εdb in
structural adhesive. The properties of the FRP vary signifi- terms of the ultimate rupture strain εfu of the material to give
cantly and represent both the laminate and fiber properties for the design debonding strain of the material. The value of κm
depends on the axial stiffness nEftf of the FRP material per
wet lay-up systems. Several assumptions were made in the
unit width and the ultimate rupture strain of the material.
construction of the database: 1) if the depth to the tensile
steel d was not specified, then it was assumed to be 0.9h
nE f t f ⎞
κ m = ------------ ⎛ 1 – --------------------
1
(where h is the height of section); and 2) if the shear modulus of - ≤ 0.90 for nE f t f ≤ 180, 000 (2)
the adhesive was not provided it was assumed to be 145 ksi 60ε fu ⎝ 360, 000⎠
(1000 MPa). The IC debonding database is given in Tables 2
and 3. A full review of all 38 studies can be found elsewhere,1 otherwise
but the studies included in the database are referenced
herein.1,14-30 1 90, 000
κ m = ------------ ⎛ ------------------⎞ ≤ 0.90 (3)
60ε fu ⎝ nE f t f ⎠
COMPARISON OF EXISTING MODELS
This section evaluates the existing national and international The predicted debonding strain based on the aforemen-
code equations currently available to predict the tensile tioned equations correlates poorly to the measured
strain at IC debonding. The strain in the beam soffit before debonding strain from the database as shown in Fig. 10. The
strengthening was included in the calculations for the prediction with the reduction factor of ψ = 0.85 is also

Table 2—IC debonding database—Part 1

Longi-
Test Shear Concrete Bar Tension Yield tudinal Ply No. Rupture FRP
Refer- Height, Width, span, span, strength, size, steel strength, FRP Modulus, thickness, of Width, strain Failure strain,
No. ence no. Specimen mm mm mm mm MPa mm number MPa type MPa mm plies mm µm/m type µm/m
1 1 EB1SB 432 127 8928 4337 61.4 Varies Varies 1241 CFRP 164,785 1.194 1 102 17,000 IC 10,600
2 1 EB1SB2 432 127 8928 4337 61.4 Varies Varies 1241 CFRP 164,785 1.194 1 102 17,000 IC 10,000
3 1 EB8SB 432 127 8928 4337 61.4 Varies Varies 1241 CFRP 95,834 2.032 1 102 10,000 Rupture 9500
4 1 EB9SB 432 127 8928 4337 61.4 Varies Varies 1241 CFRP 95,834 2.032 2 102 10,000 IC 8680
5 14 III-1 155 203 2000 1000 22.1 10 2 346 CFRP 257,000 0.165 1 50 17,600 IC 5508
6 14 III-2 156.5 199 2000 1000 20.9 10 2 373 CFRP 257,000 0.165 1 100 17,600 IC 7692
7 14 III-4 153.5 150.1 2000 1000 22.1 10 2 351 CFRP 257,000 0.165 1 50 17,600 IC 11,291
8 15 B 455 205 4880 1982 35 25 2 456 GFRP 37,200 6.00 1 152 10,753 IC 5200
CFRP 152,000
9 16 A3.1 300 140 4800 1800 24.9 16 2 435 1.20 1 80 15,789 IC 7000
strips
BF-04/
10 17 300 150 3000 1500 33 10 2 449 CFRP 165,000 1.20 1 40 17,000 IC 5800
0.5S
11 17 BF-06/S 300 150 3000 1500 32.5 12 2 490 CFRP 165,000 1.20 1 80 17,000 IC 5400
12 17 B-08/S 300 150 3000 800 33.8 12 3 490 CFRP 165,000 1.20 1 80 17,000 IC 5000
13 17 BO-08/S 300 150 3000 800 36.5 12 3 490 CFRP 165,000 1.20 1 80 17,000 IC 5500
CFRP
14 18 BF2 450 200 3800 1250 36.5 16 4 590
strips 159,000
1.20 1 100 18,500 IC 6700

CFRP 159,000
15 18 BF3 450 200 3800 1250 34.9 16 4 590 1.20 1 100 18,500 IC 7200
strips
CFRP 159,000
16 18 BF4 450 200 3800 1250 30.8 16 4 590 strips 1.20 1 100 18,500 IC 6800

CFRP
17 18 BF5 450 200 3800 1250 37.4 16 4 590 strips 159,000 1.20 1 100 18,500 IC 5700

CFRP 159,000
18 18 BF8 450 200 3800 1250 39.4 16 2 590 1.20 1 100 18,500 IC 5800
strips
19 19 A5 150 200 2100 750 54 10 2 575 CFRP 127,000 0.80 1 150 12,100 IC 7200
20 19 B3 150 200 2100 750 54 10 2 575 CFRP 127,000 0.40 1 150 12,100 IC 9700
21 19 B6 150 200 2100 750 54 10 2 575 CFRP 127,000 1.20 1 150 12,100 IC 5500
CFRP
22 20 C-3 254 152 2440 839 55.2 16 2 415 49,250 1.90 1 152 14,000 IC 6700
sheets
CFRP
23 20 C-1 254 152 2440 839 55.2 16 2 415 28,333 0.13 1 152 12,000 Rupture 11,000
sheets
24 21 B1 200 150 1500 550 24.96 8 3 288 CFRP 235,000 0.11 1 150 14,894 IC 7834
25 21 B2 200 150 1500 550 24.96 8 3 288 CFRP 235,000 0.11 1 150 14,894 IC 7100
26 21 B3 200 150 1500 550 24.96 8 3 288 CFRP 235,000 0.11 1 150 14,894 IC 5868
Note: 1 in. = 25.4 mm; 1 ksi = 6.895 MPa.

46 ACI Structural Journal/January-February 2008

unconservative over a wide range of FRP axial stiffnesses as concrete (kc = 0.67 for FRP bonded to concrete with low
shown in Fig. 10. The ACI equation is unconservative for compaction, for example, faces not in contact with formwork
low FRP stiffnesses and becomes slightly more conservative during casting, otherwise kc = 1.0). The value c1 can be
as the FRP stiffness increases. determined experimentally or can be set equal to 0.64. The
The Concrete Society31 and the Fédération Internationale value of kb is a width factor equal to
du Béton (fib)32 have similar analytical models for IC
debonding based on a fracture mechanics approach with
different calibration constants. The debonding strain εdb 2 – -----f
b
according to the fib code is bc
k b = 1.06 ------------------ ≥ 1 (5)
bf
1 + -------- -
ft ′ 400
ε db = αc 1 k c k b ---------
- (4)
nEt f
where bf /bc ≥ 0.33 and ft′ is the mean value of the concrete
where α is a reduction factor accounting for the influence of tensile strength. The prediction according to the fib equation
inclined cracks on the bond strength (α = 1.0 for beams with and the collected database is given in Fig. 11. The comparison
sufficient shear reinforcing and for slabs, otherwise α = 0.9). suggests that the debonding prediction becomes less
The value kc is a factor accounting for the state of compaction of conservative as the axial stiffness of the FRP decreases. It

Table 3—IC debonding database—Part 2

Longi-
Test Shear Concrete Bar Tension Yield tudinal No. Rupture FRP
Refer- Speci- Height, Width, span, span, strength, size, steel strength, FRP Modulus, Ply thick- of Width, strain, Failure strain,
No. ence no. men mm mm mm mm MPa mm number MPa type MPa ness, mm plies mm µm/m type µm/m
27 22 B1 356 203 2690 1065 37.2 16 2 440 CFRP 230,000 0.17 2 75 15,000 IC 7200
28 22 C1 406 203 3000 1220 35.1 16 2 440 CFRP 62,000 1.04 2 50 12,000 IC 7600
29 22 C2 406 203 3000 1220 35.1 16 2 440 CFRP 62,000 1.04 2 50 12,000 IC 7000
30 22 D2 406 203 3000 1220 37.2 16 2 440 CFRP 155,000 1.19 1 50 14,000 IC 4800
31 23 B-083m 300 150 4200 1400 34.4 12 2 436 CFRP 65,400 0.38 3 150 15,000 IC 6810
32 23 B-08S 300 150 4200 1400 32.3 12 2 493 CFRP 172,000 1.20 1 50 17,000 IC 6170
33 23 B-08M 300 150 4200 1400 37.3 12 2 493 CFRP 220,000 1.40 1 120 12,400 IC 5060
CFRP
34 24 A-C1 250 150 2600 1050 31.5 16 2 407 230,000 0.17 1 130 14,800 Rupture 14,800
sheets
AFRP
35 24 A-AT 250 150 2600 1050 31.5 16 2 407 78,500 0.38 1 130 29,900 IC 17,342
sheets
AFRP
36 24 A-AK 250 150 2600 1050 31.5 16 2 407 118,000 0.29 1 130 17,500 IC 13,825
sheets
CFRP
37 24 B-C1 400 150 2600 1050 31.5 16 2 407 230,000 0.17 2 130 14,800 IC 8880
sheets
CFRP
38 24 B-C2 400 150 2600 1050 31.5 16 2 407 sheets 440,000 0.19 1 130 5500 Rupture 5500

39 24 B-AT 400 150 2600 1050 31.5 16 2 407 AFRP 78,500 0.38 2 130 29,900 IC 10,465
sheets

40 24 B-AK 400 150 2600 1050 31.5 16 2 407 AFRP 118,000 0.29 2 130 17,500 IC 10,675
sheets
CFRP
41 25 V2 180 120 1800 750 41 8 2 533 240,000 0.111 2 70 15,000 IC 8070
sheets
CFRP
42 25 V4 180 120 1800 750 41 8 2 533 HM 200,000 1.4 1 20 11,000 IC 6870
strips
CFRP
43 26 B3 700 350 6000 2500 37 20 4 557 160,000 1.40 1 240 14,000 IC 9800
strips
CFRP
44 27 B5 200 150 2000 750 42.9 10 2 531 235,000 0.22 1 150 17,872 IC 10,489
sheets
CFRP
45 27 B6 200 150 2000 750 42.9 10 2 531 sheets
235,000 0.22 1 150 17,872 IC 9399

46 28 L1 250 150 4537 2269 23.3 12.7 3 429 CFRP 155,000 1.40 1 25 18,000 IC 5300
47 28 L2 250 150 4537 2269 23.3 12.7 3 429 CFRP 155,000 1.40 1 50 18,000 IC 6688
48 28 L2x1 250 150 4537 2269 23.3 12.7 3 429 CFRP 155,000 1.40 1 50 18,000 IC 7878
49 28 L4 250 150 4537 2269 23.3 12.7 3 429 CFRP 155,000 1.40 1 100 18,000 IC 6595
50 29 CS 825 343 8840 4420 45 Varies Varies 364 CFRP 155,000 1.40 1 200 18,000 IC 9400
51 30 CS 254 152 4600 2300 29.9 12.7 3 446 CFRP 155,000 1.40 1 50 18,000 IC 6400
Maximum 700 350 6000 2500 55.2 25 4 590 CFRP 440,000 6 3 240 29,900 — 17,342
GFRP,
Minimum 150 120 1500 550 20.9 8 2 288
AFRP 28,333.3333
0.11 1 20 5500 — 4800
Note: 1 in. = 25.4 mm; 1 ksi = 6.895 MPa.

ACI Structural Journal/January-February 2008 47

was also observed that as the shear span-to-depth ratio In the current draft of the Australian standard for the retro-
increases, the model becomes more conservative. fitting of reinforced concrete with FRP, an equation is
Prediction of the debonding strain according to the provided to predict the maximum IC debonding strain in a
Chinese code is given by the following equation11 beam with a single crack.33 The equation contains a unique
variable defined as the perimeter length of the idealized fracture
f′ plane Lper , as illustrated in Fig. 13. The guideline admits that the
ε db = λk b ----t ⎛ ---------------
1 - – 0.2-------⎞ (6) IC debonding resistance may be higher in beams with multiple
2 ⎝ nE t L d ⎠ flexural cracks, but a lower bound value is estimated as
f f

where kb is a width factor related to the width of the FRP τ f δ f L per

material bf to the base material bc as follows ε db = --------------------------
- (8)
nE f t f bf

b
2.25 – -----f The depth x (in Fig. 14) is equal to the thickness of the
b concrete paste layer during IC debonding failure, which
kb = ---------------------c- (7)
b separates from the concrete surface and is recommended as
1.25 + -----f
bc 0.0394 in. (1 mm). The value nEf tf bf is the axial stiffness of
the FRP material. The value τf δf is the maximum interface
shear stress multiplied by the maximum interface slip that
The value of Ld in Eq. (6) is the distance from the plate end to
the section where the FRP plate is fully used. This is assumed to
be equal to the shear span minus the distance from the end of the
FRP plate to the support. The value λ is a factor to account for
the effect of FRP U-wrap anchors. When U-wraps are used
throughout the length of the beam, a factor of 1.3 is
recommended. If the designer has sufficient experience, then λ
can be set equal to 1.5. When no U-wraps are used, the factor
should be set equal to 1.0. Comparison of the predicted strain
according to the Chinese code is generally conservative and
correlates well with the database as shown in Fig. 12.

Fig. 12—Experimental versus predicted debonding strains

using Chinese10 model.

Fig. 10—Experimental versus predicted debonding strains

using ACI 44012 model. Fig. 13—Perimeter length of idealized failure plane.33

Fig. 11—Experimental versus predicted debonding strains Fig. 14—Predicted versus experimental debonding strains
using fib32 model. using Australian draft model.33

48 ACI Structural Journal/January-February 2008

 CONCLUSIONS
This research provides unique information related to the
bond behavior of prestressed concrete strengthened with
externally bonded CFRP, especially related to the IC
debonding phenomenon. The collected database and test
results of the experimental program were compared against
the analytical models in national code documents as well as
a proposed model introduced by the authors. The following
conclusions can be reached from this research:
1. For long-span, flexural prestressed concrete members
strengthened with CFRP without transverse U-wraps, the
common failure mode is IC debonding;
2. Precured CFRP systems are more prone to IC
Fig. 15—Predicted versus experimental debonding strains debonding due to their large tensile strain capacity and high
using proposed design model.34 stiffness due to large fiber volume fraction;
3. The IC debonding process will initiate for both precured
represents twice the fracture energy. This value can be and wet lay-up CFRP systems in the concrete layer at the
determined from material testing or may be taken equal to FRP-concrete interface;
the following equation given by the code commentary for
4. During rapid IC crack propagation, the failure plane
beams strengthened with externally bonded sheets or strips
may shift through the adhesive layer and continue along the
plate-adhesive interface outside of the heavily cracked
τ f δ f = 0.73 ⎛ -------------⎞ ( f c′ )
1 0.5 0.67 region for precured CFRP systems;
(9)
⎝ b f + 2⎠ 5. Transverse CFRP U-wraps placed throughout the girder
length can increase the tensile strain in the precured longitudinal
The predicted values for debonding strain based on the CFRP at IC debonding failure by as much as 20%.
Australian approach are similar to those found using the 6. The analytical model from ACI Committee 44012 for IC
approach of fib code32 but are more conservative as shown in debonding is currently unconservative and should be reevaluated;
Fig. 14. 7. The most conservative model currently available is the
model introduced by the Australian Draft Design
PROPOSED DESIGN MODEL Guideline33; and
The following design model, proposed by the authors, 8. The proposed model provides a relatively accurate
characterizes the interface shear stress based on the nature of the prediction of IC debonding and addresses both reinforced
applied loading and the stress concentrations at the toes of the and prestressed concrete structures.
flexural cracks.1 The model requires calculation of the flexural
yielding moment My and the corresponding strain level in the ACKNOWLEDGMENTS
FRP material (εf@y). The moment corresponding to debonding The authors would like to thank the contribution of time and energy from
P. Zia, K. Harries, M. Dawood, A. Miller, and C. Walter. Research funds
failure Mdb can be calculated by assuming a value of debonding received from the North Carolina Department of Transportation are greatly
strain (εdb). The maximum interface shear stress τcmax can appreciated along with the many other contractors, FRP manufacturers, and
then be calculated using the following equation installers who generously donated time and materials. The authors would
also like to thank the dedicated staff of the Constructed Facilities Laboratory
(J. Atkinson, B. Dunleavy, and A. Yonai) whose help at all stages of this
ε db – ε f @y⎞ M
τ cmax = nE f t f ⎛ ------------------------ + 3 ⎛ 1.1 – --------y-⎞ f c′
research has been immeasurable.
(10)
⎝ s – xy ⎠ ⎝ M db⎠
REFERENCES
1. Rosenboom, O. A., “Behavior of FRP Repair/Strengthening Systems
where s is the shear span and xy is the distance from the for Prestressed Concrete,” PhD thesis, North Carolina State University,
support to the location of first yielding of the internal tensile Raleigh, NC, 2006, 400 pp.
steel, which is dependent upon the loading conditions. The 2. Rosenboom, O. A.; Hassan, T. K.; and Rizkalla, S. H., “Flexural
value of debonding strain is iterated until τcmax reaches the Behavior of Aged Prestressed Concrete Girders Strengthened with Various
FRP Systems,” Construction and Building Materials, V. 21, No. 4, 2007,
critical value equal to 1.8 ft′. The total strain in the CFRP pp. 764-776.
(εdb) is the summation of the debonding strain and the strain 3. Rosenboom, O. A., and Rizkalla, S. H., “Behavior of Prestressed Concrete
due to stress concentrations (εsc) that can be estimated as Strengthened with Various CFRP Systems Subjected to Fatigue Loading,” Journal
of Composites for Construction, ASCE, V. 10, No. 6, 2006, pp. 492-502.
4. Oehlers, D. J., and Moran, J. P., “Premature Failure of Externally
M f c′
ε sc = 0.342 ⎛ 1.1 – --------y-⎞ -----------
Plated Reinforced Concrete Beams,” Journal of Structural Engineering,
(11)
⎝ ⎠
M db nE f t f
V. 116, No. 4, 1990, pp. 978-995.
5. El-Mihilmy, M. T., and Tedesco, J. W., “Prediction of Anchorage
Failure for Reinforced Concrete Beams Strengthened with Fiber-Reinforced
Polymer Plates,” ACI Structural Journal, V. 98, No. 3, May-June 2001,
If the total strain (εdb + εsc) exceeds the rupture strain of pp. 301-314.
the material, then rupture will occur before debonding. 6. Teng, J. G., and Yao, J., “Plate End Debonding in FRP-Plated RC
Complete details on the proposed model can be found Beams—II: Strength Model,” Engineering Structures, V. 29, No. 10,
elsewhere.1 The prediction according to the proposed model 2007, pp. 2472-2486.
is shown in Fig. 15. It should be noted that the uniqueness of 7. So, M., and Harmon, T. G., “Cover Delamination of R/C Members
with Surface Mounted FRP Reinforcement,” ACI Structural Journal.
the proposed model is the ability to model the behavior of the (accepted for publication)
prestressed concrete section explicitly through calculation of the 8. Tumialan, G.; Serra, P.; Nanni, A.; and Belarbi, A., “Concrete Cover
parameters My and Mdb. Delamination in Reinforced Concrete Beams Strengthened with Carbon

ACI Structural Journal/January-February 2008 49

Fiber Reinforced Polymer Sheets,” Proceedings of the Fourth International 22. Breña, S. F.; Benouaich, M. A.; Kreger, M. E.; and Wood, S. L.,
Symposium on Fiber Reinforced Polymer Reinforcement for Reinforced “Fatigue Tests of Reinforced Concrete Beams Strengthened Using Carbon
Concrete Structures (FRPRCS4), MD, 1999, pp. 725-735. Fiber Reinforced Polymer Composites,” ACI Structural Journal, V. 102,
9. Reed, C. E., and Peterman, R. J., “Evaluation of Prestressed Concrete No. 2, Mar.-Apr. 2005, pp. 305-313.
Girders Strengthened with Carbon Fiber Reinforced Polymer Sheets,” 23. Kotynia, R., and Kaminska, M. E., “Ductility and Failure Mode of
Journal of Bridge Engineering, ASCE, V. 9, No. 2, 2004, 185-192 pp. RC Beams Strengthened for Flexure with CFRP,” Report No. 13, Department
10. Chahrour, A., and Soudki, K., “Flexural Response of Reinforced of Concrete Structures, Technical University of Lodz, Poland, 2003, 51 pp.
Concrete Beams Strengthened with End-Anchored Partially Bonded Carbon
24. Zhang, G. F.; Kishi, N.; and Mikami, H., “Influence of Material
Fiber Reinforced Polymer Strips,” Journal of Composites in Construction, V. 9,
No. 2, 2005, pp. 170-177. Properties of FRPs on Strength of Flexural Strengthened RC Beams,”
11.Ye, L. P.; Lu, X. Z.; and Chen, J. F., “Design Proposals for the Proceedings of the Sixth International Symposium on FRP Reinforcement
Debonding Strength of FRP Strengthened RC Beams in the Chinese for Concrete Structures (FRPRCS6), Singapore, 2003, pp. 327-336.
Design Code,” Proceedings of the International Symposium on Bond 25. Dias, S.; Juvandes, L.; and Figueiras, J., “Strengthening of Reinforced
Behavior of FRP in Structures (BBFS 2005), Hong Kong, China, Dec. Concrete Structures in Bending with CFRP,” Revista IBRACON de Estructuras
2005, pp. 55-62. (Instituto Brasileiro de Concreto), V. 1, No. 1, 2004, pp. 1-19.
12. ACI Committee 440, “Design and Construction of Externally Bonded 26. Khoman, N.; Foster, S. J.; and Smith, S. T., “Debonding Failure in
FRP Systems for Strengthening Concrete Structures (ACI 440.2R-02),” CFRP Strengthened Concrete Beams,” Proceedings of the Second Interna-
American Concrete Institute, Farmington Hills, MI, 2002, 45 pp. tional Conference on FRP Composites in Civil Engineering (CICE 2004),
13. American Association of State Highway and Transportation Officials, Adelaide, Australia, 2004, pp. 505-512.
“AASHTO LRFD Bridge Design Specifications,” AASHTO, 2004, 27. Pan, J. L., and Leung, C. K. Y, “FRP Debonding under the Presence
1450 pp. of Multiple Cracks Along a Concrete Beam,” Proceedings of the International
14. Yao, J.; Teng, J. G.; and Lam, L., “Experimental Study on Intermediate Symposium on Bond Behavior of FRP in Structures (BBFS 2005), Hong
Crack Debonding in FRP-Strengthened RC Flexural Members,” Advances Kong, China, 2005, pp. 283-290.
in Structural Engineering, V. 8, No. 4, 2004, pp. 365-395.
15. Saadatmanesh, H., and Ehsani, M. R., “RC Beams Strengthened with 28. Reeve, B. Z., “Effect of Adhesive Stiffness and CFRP Geometry on
GFRP Plates, I: Experimental Study,” Journal of Structural Engineering, the Behavior of Externally Bonded CFRP Retrofit Measures Subject to
ASCE, V. 117, No. 11, 1991, pp. 3417-3433. Monotonic Loads,” master’s thesis, University of Pittsburgh, Pittsburgh,
16. Spadea, G.; Bencardino, F.; and Swamy, R. N., “Structural Behavior PA, 2005, 110 pp.
of Composite RC Beams with Externally Bonded CFRP,” Journal of 29. Aidoo, J.; Harries, K. A.; and Petrou, M. F., “Full-Scale Experimental
Composites for Construction, ASCE, V. 2, No. 3, 1998, pp. 132-137. Investigation of Repair of Reinforced Concrete Interstate Bridge Using
17. Kaminska, M. E., and Kotynia, R., “Experimental Research on RC CFRP Materials,” Journal of Bridge Engineering, ASCE, V. 11, No. 3,
beams Strengthened with CFRP Strips,” Report No. 9, Department of Concrete 2006, pp. 350-358.
Structures, Technical University of Lodz, Poland, 2000, 55 pp. 30. Quattlebaum, J.; Harries, K. A.; and Petrou, M. F., “Comparison of
18. Matthys, S., “Structural Behavior and Design of Concrete Members Three CFRP Flexural Retrofit Systems Under Monotonic and Fatigue Loads,”
Strengthened with Externally Bonded FRP Reinforcement,” dissertation, Journal of Bridge Engineering, ASCE, V. 10, No. 6, 2004, pp. 731-740.
University of Ghent, Belgium, 2000.
31. The Concrete Society, “Design Guidance for Strengthening Concrete
19. Rahimi, H., and Hutchinson, A., “Concrete Beams Strengthened with
Structures Using Fibre Composite Materials,” Technical Report 55,
Externally Bonded FRP Plates,” Journal of Composites for Construction,
UK, 2004, 128 pp.
ASCE, V. 5, No. 1, 2001, pp. 44-56.
20. Grace, N. F.; Abdel-Sayed, G.; and Ragheb, W. F., “Strengthening of 32. Fédération Internationale du Béton, “Externally Bonded FRP
Concrete Beams Using Innovative Ductile Fiber-Reinforced Polymer Fabric,” Reinforcement for RC Structures,” fib Bulletin 14, Task Group 9.3, FRP
ACI Structural Journal, V. 99, No. 5, Sept.-Oct. 2002, pp. 692-700. Reinforcement for Concrete Structures, 2001, 138 pp.
21. Fang, T. Q., “Study on U-Shaped Sheet Behavior of Anti-Debonding 33. Oehlers, D. J.; Seracino, R.; and Smith, S. T., “RC Structures Retro-
in the Concrete Beam Reinforced Flexurally with FRP,” master’s thesis, fitted with FRP and Metal Plates: Beams and Slabs,” Standards Australia,
Tsingua University, China, 2002. 108 pp. (in press)

50 ACI Structural Journal/January-February 2008