Flexural Strengthening of

Prestressed Bridge Slabs

with FRP Systems
Fiber reinforced polymer (FRP) materials offer great potential for cost-
effective retrofitting of concrete structures. In response to the growing
need for strengthening and rehabilitation of concrete structures and
bridges, an experimental program was conducted to investigate the
feasibility of using different strengthening techniques as well as
different types of FRP for strengthening prestressed concrete
members. Half-scale models of a prestressed concrete bridge were
constructed and tested to failure. The test specimens consisted of one
simple span and two overhanging cantilevers. Each specimen was
Tarek Hassan
Ph.D. Candidate tested three times using a different load location in each case. Five
Department of Civil and different strengthening techniques were investigated including near
Geological Engineering
University of Manitoba
surface mounted Leadline bars, C-Bars, CFRP strips and externally
Winnipeg, Manitoba, Canada bonded CFRP sheets and strips. Ultimate capacity, failure mechanism
and cost analysis of various strengthening techniques for concrete
bridges are presented. The applicability of a nonlinear finite element
analysis of post-tensioned bridge slabs strengthened with near surface
mounted FRP reinforcement is enumerated.

n an aggressive environment, con- In the United States, nearly one-third

I crete may be vulnerable to chemical

attacks such as carbonation and
chloride contamination which breaks
of the nation's 581,000 bridges were
found to be structurally deficient or
functionally obsolete.! A large number
down the alkaline barrier in the ce- of these deficient bridges are rein-
ment matrix. Consequently, the steel forced or prestressed concrete struc-
Sami Rizkalla, Ph.D., P.E. reinforcement in concrete structures tures, and are in urgent need of repair
Distinguished Professor becomes susceptible to rusting and and strengthening. In the United King-
Department of Civil Engineering corrosion. Such a phenomenon leads dom, over 10,000 concrete bridges
North Carolina State University
Raleigh, North Carolina
to delamination of the concrete at the need structural attention. In Europe, it
reinforcement level, cracking and is estimated that the repair of struc-
spalling of the concrete under more tures due to corrosion of reinforcing
severe conditions. bars in reinforced concrete structures

76 PCI JOURNAL
 west abutment east abutment
west piers eastpiers
bearing C.L bearing

Precast concrete
bearing piles
'Qt. . . . . . . . . . . . . . . ...,. .F1IO.....r-.-......r-....,............CW....
Cross section

1 ft = 0.3048 m

Fig. 1. Schematic of Bridge No. 444 in Winnipeg, Manitoba, Canada.

costs over $600 million annually. 2 where the maximum shear is located. injecting them with a high strength
A possible solution to combat rein- Bridge replacement costs were esti- epoxy resin adhesive. The midspan
forcement corrosion for new construc- mated to exceed $1.6 million (USD). section was then strengthened using
ion is the use of non-corrosive materi- Therefore, it was decided to consider FRP reinforcement and tested.
tis to replace conventional steel bars. strengthening the bridge using FRP.
lIigh tensile strength, lightweight and Site inspections indicated extensive RESEARCH SIGNIFICANCE
:orrosion resistance characteristics transverse and longitudinal cracking
nake FRP (fiber reinforced polymer) of the top surface of the bridge deck. This paper demonstrates the cost ef-
deal for such applications. FRP also The underside of the deck was gener- fectiveness of five different strength-
JfOvides a cost effective and practical ally in good condition. Based on site ening techniques using FRP for flexu-
echnique for the repair and strength- inspections, the bridge does not need ral strengthening of post-tensioned
ming of structures and bridges using to be demolished and replaced. It can concrete bridge slabs. The signifi-
~xtemally bonded sheets or prefabri- be feasibly and economically rehabili- cance of this project is that the flexu-
:ated laminates. FRP tendons can also tated. ral strengthening is located at the
)e used to strengthen old prestressed Due to the lack of information on zones of maximum shear stresses typi-
the use of near surface mounted FRP cally occurring at the maximum nega-
:oncrete girders as well as to prevent
reinforcement for flexural strengthen- tive moment section of cantilevers and
:orrosion in tendons from occurring in
ing in regions of combined bending continuous beams. The research pro-
alty regions. 3.4
and high shear stresses, the reported vides experimental evidence and de-
The needs to study the most appro-
experimental program in this paper tailed performance of various FRP
Ifiate strengthening technique for pre:
has been undertaken. Three half-scale strengthening techniques.
tressed concrete bridges are initiated
models of the bridge under considera- The paper also provides a cost anal-
Iy the necessity to upgrade a 30-year-
tion were cast and post-tensioned. The ysis for each technique to help engi-
lId concrete bridge in Winnipeg,
specimens were tested in simple span neers to select the most appropriate
tlanitoba, Canada. A bridge rating
with a dOJ.lble cantilever configuration. system for flexural repair and
nalysis conducted using the current
Each specimen was tested three times strengthening of concrete structures
~ASHTO (American Association of
using loads applied at different loca- and bridges. The development of cost-
tate Highway and Transportation Of-
tions in each test. effective and durable restorative sys-
icials) Specifications indicated that
The first and second tests were per- tems will extend the service life of de-
Ie flexural strength of the bridge deck
formed on the two cantilevers where teriorated civil infrastructures and
; not sufficient to withstand modern
the load was applied at the edge of ensure safety of the public.
ruck loads. 3 To accommodate the
~ASHTO HSS30 truck design load, each cantilever. The third test was
Ie analysis indicated a need to in- conducted using a load applied at
midspan. Prior to the midspan test, the
TEST SPECIMENS
rease the flexural strength by approx-
nately 10 percent at the negative mo- cracks resulting from testing the two The test specimens simulate the
lent zone over the pier columns cantilevers were sealed completely by post-tensioned solid slab over the in-

77
anuary-February 2002
Fig. 2. Reinforcement details of test specimens.

termediate pier columns of a bridge The number and layout of the ten- (254 mm) on center in the simply sup-
constructed in the early 1970s in Win- dons were selected to have the same ported span. Reinforcement details of
nipeg, Manitoba, Canada. The bridge stress level of the bridge under service the specimens are shown in Fig. 2.
consists of four spans, as shown in load conditions. The critical bending Bursting reinforcement was pro-
Fig. 1, and was designed for moments for the bridge were evalu- vided using six pairs of No.3 looped
AASHTO HSS20 truck design load- ated based on a linear elastic finite ele- bars spaced at 3 in. (75 mm) on center.
ing. 5 The thickness of the bridge slab. ment analysis using a commercial Twelve 0.6 in. (15 mm) seven-wire
along the spans is 32 in. (810 mm), computer program, SAP2000. 6 The strands were used for post-tensioning
cast in place, partially voided and loss of the prestressing force was cal- the specimens. The compressive
post-tensioned. The solid slab over the culated according to the current strength of concrete after 28 days
intermediate pier columns is post-ten- AASHTO Specifications. 5 ranged from 6500 to 7200 psi (45 to
sioned transversely to resist the nega- The specimens were reinforced with 50 MPa) for the three slabs.
tive moments at the columns and the four No.5 mild steel bars on the top
positive moment at midspan. surface and five No.5 mild steel bars
STRENGTHENING
To simulate the combined effect of on the bottom surface. The bars have a
high flexural and shear stresses at the yield strength of 60 ksi (400 MPa) and TECHNIQUES
intermediate supports of the bridge, an elastic modulus of 29,000 ksi (200
three half-scale models of the solid GPa). The number of bars in the top Slab Sl
slab designated as Sl, S2, and S3 were surface was selected to represent the One cantilever of Slab S 1 was
constructed. The support configuration same reinforcement ratio in the can- strengthened using near surface
of the specimens was designed to ex- tilever portion of the bridge. mounted Leadline CFRP bars while the
amine the FRP repair system at the Shear reinforcement consisted of other cantilever remained unstrength-
zones of maximum negative moment No.3 U-shaped stirrups. The stirrups ened. The Leadline bars were produced
at the support, which coincides with were spaced at 5 in. (127 mm) on cen- by Mitsubishi Chemicals Corporation,
the maximum shear. ter in the cantilever span and lOin. Japan. The bars have a modulus of

78 PCI JOURNAL
 elasticity of 21,000 ksi (147 GPa) and
an ultimate tensile strength of 285 ksi
(1970 MPa) as reported by the manu-
facturer. Based on equilibrium and "

compatibility conditions, six 3/8 in. (10

mm) diameter Leadline CFRP bars
were used to achieve a 30 percent in-
crease in the ultimate load carrying ca-
pacity of the slab.
To strengthen the cantilever slab
: using near surface mounted bars,
grooves had to be cut on the top sur-
face of the concrete. The location of Fig. 3. Cutting
the grooves was first marked using a grooves for near
chalk line. The grooves were 8 in. surface mounted
(200 mm) apart. A concrete saw was CFRP bars.
used to cut six grooves approximately
0.6 in. wide and 1.2 in. deep (16 x 30
mrn) at the tension surface of the can- grooves and gently pressed to displace Slab S2
tilever as shown in Fig. 3. the bonding agent as shown in Fig. 4. The second slab, S2, was used to in-
The groove ends were widened to The grooves were then filled com- vestigate the performance of both near
provide wedge action and to prevent pletely with the epoxy. Quality control surface mounted and externally
possible slippage of the bars. Tapering was achieved through continuous in- bonded CFRP strips in strengthening
the ends of the grooves was intended spections and measurements during of concrete bridges. Recently, test re-
to induce inclined frictional forces at the installation procedures. sults of near surface mounted CFRP
the concrete-epoxy interface. These in- After completion of the first two strips reported good and uniform bond
clined forces provided radial confining tests on both cantilevers, the resulting distribution of the strips to the con-
forces on the bars and, consequently, cracks were injected with epoxy. crete and full utilization of the
increased the pullout resistance. Based on an equilibrium and compati- strength of the strip up to rupture.?
Kemko 040 epoxy adhesive was bility approach, ten 3/8 in. (10 mm) di- The analysis indicated a need of six
used for bonding the CFRP bars to the ameter Lead1ine CFRP bars were used CFRP strips, 2 in. wide and 0.055 in.
surrounding concrete. The epoxy was to achieve a 30 percent increase in the thick (50 x 1.4 mm), to achieve a 30
produced by ChemCo Systems, Inc., ultimate capacity of the midspan sec- percent increase in the ultimate capac-
United States. The adhesive is com- tion of the simply supported slab. The ity of the cantilever slab.
monly used for grouting bolts, dowels, grooves on the bottom surface were The first cantilever was strength-
and steel bars in concrete. The epoxy 4.7 in. (120 mm) apart. The same pro- ened using six externally bonded
was pressure injected into the grooves cedures as described before for cut- CFRP strips. The concrete substrate
to cover two-thirds of the groove ting the grooves and placing the bars was prepared by grinding the surface
height. The bars were placed in the were applied. at the locations of the strips. The

10 "
I l----J 0.6 tf
~II~:~-.:......Ii

Detail of groove widening

1 in. =25.4 nun

Fig. 4. CFRP bars inserted in epoxy.

January-February 2002 79
Fig. 5. Bonding CFRP strips to surface.

epoxy was then placed over the strips The CFRP strips were produced by Slab S3
and on the concrete surface. Finally, S&P Clever Reinforcement Company,
A widespread method for the reha-
the strips were placed on the concrete Switzerland. The strips had a modulus
surface and gently pressed into the of elasticity of 21,800 ksi (150 GPa) bilitation of concrete structures is the
epoxy using a ribbed roller as shown and an ultimate tensile strength of 290 externally bonded CFRP sheets. This
in Fig. 5. ksi (2000 MPa) as reported by the method can be seen as a state-of-the-
The second cantilever was strength- manufacturer. art technique despite some detailing
ened also using six CFRP strips in- After testing both cantilevers, the problems and design aspects that
serted into grooves cut at the top sur- areas above the supports were sub- could influence the failure modes.
face of the concrete. In order to insert stantially cracked. To enable further To investigate the effectiveness of
the strips within the concrete cover testing of the midspan, the cracks this strengthening method in compari-
layer, the strips were cut into two were injected similar to Slab S 1. son to the three previously prescribed
halves, each 1 in. (25 mm) wide. Eighteen near surface mounted CFRP techniques, the simply supported span
Using a concrete saw, grooves approx- strips, 1 in. wide and 0.055 in. thick and one cantilever of Slab S3 were
imately 0.2 in. wide and 1 in. deep (5 (25 x 1.4 mm), spaced by 2.6 in. (66 both strengthened using externally
x 25 m11.l) were cut at the tension sur- mm) on center were used to achieve a bonded CFRP sheets. The sheets were
face of the second cantilever. 30 percent increase in the ultimate ca- manufactured by Master Builders
The grooves were injected with the pacity of the simply supported slab. Technologies, Ltd., Ohio. The re-
epoxy adhesive to provide the neces- En-Force CFL was used in bonding quired area of CFRP sheets was calcu-
sary bond with the surrounding con- both near surface mounted and exter- lated to achieve a 30 percent increase
crete as shown in Fig. 6a. The strips nally bonded CFRP strips to the con- in the flexural capacity of the can-
were then placed in the grooves and crete. The epoxy was produced by tilever slab.
completely covered with the epoxy as Structural Composites, Inc., United For the first cantilever, the sheets
shown in Fig. 6b. States. were applied in two plies. The first ply

Fig. 6a. Filling grooves with epoxy. Fig. 6b. Inserting CFRP strips inside grooves.

80 PCI JOURNAL
 covered the entire width of the slab
while the second ply covered 19 in.
(480 mm) and was centered along the
width of the slab. Installation proce-
dures are illustrated in Fig. 7. The
sheets had a modulus of elasticity of
J3,000 ksi (228 GPa) and an ultimate
tensile strength of 620 ksi (4275 MPa)
as reported by the manufacturer.
The second cantilever was strength-
ened using eight near surface mounted
C·BAR CFRP bars. The bars were
manufactured by Marshall Industries
Composites Inc., United States. Based
(a) Grinding of the surface
on testing, the bars had a modulus of
elasticity of 16,100 ksi (111 GPa) and
an ultimate tensile strength of 280 ksi
(1918 MPa). The bars were sand-
blasted to enhance their bond to the
epoxy adhesive. The bars were then
inserted inside grooves cut at the top
surface of the concrete. The grooves
were 6 in. (150 mm) apart. The groove
dimensions were 0.6 in. wide and 1.2
in. deep (16 x 30 mm).
The simply supported span was
strengthened with externally bonded
CFRP sheets after injecting the cracks
resulting from cantilever tests. Three
plies of CFRP sheets were used to
achieve a 30 percent increase in flexu-
ral capacity. The first two plies cov-
ered the entire width of the slab, while
the third ply covered 16 in. (400 mm)
and was centered along the width of
the slab. Detailed information about
the tested specimens is provided in
Table 1. The designation of the tested
specimens are C or SS for cantilever
or simply supported specimens, re-
spectively.

TESTING SCHEME
The slabs were tested under static
loading conditions using a uniform
,.. , ,.'
•
(c) Bonding the CFRP sheets

line-load acting on a width equivalent Fig. 7. Installation procedures for externally bonded CFRP sheets.
to the width of a tire contact patch ac-
cording to the AASHTO HSS30 de-
sign vehicle. A closed-loop MTS 1200 Table 1. Details of test specimens.
kip (5000 kN) testing machine was Slab No. Specimen Strengthening technique
used to apply the load using stroke CI Control specimen
I C2 6 No.3 near surface mounted Leadline bars
control mode with a rate of 0.02 10 No.3 near surface mounted Leadline bars
SSI
in.lmin (0.5 mmlrnin) up to failure. C3 6 Externally bonded CFRP strips (2 x 0.055 in.)
Neoprene pads were placed between 2 C4 12 near surface mounted CFRP strips (l x 0.055 in.)
the steel beam and the slab to simulate SS2 18 near surface mounted CFRP strips (l x 0.055 in.)
the contact surface of a truck tire and C5 2 plies of externally bonded CFRP sheets
3 C6 8 No.3 near surface mounted C-BAR CFRP bars
to avoid local crushing of the concrete. 3 plies of externally bonded CFRP sheets
SS3
For the cantilever tests, the load was

January-February 2002 81
 (1) E/astomeric bearings (Neoprene pads) 16" xl6" xO.16"
(2) Steel plate 8" xl" (Length=4')
(J) H8S JItx2"xO.251t (Length=10") MTS
(4) HSS 4"x4ItxO.2jlt (Length=50")
(5) Prestressed DYWIDAG bar (Diameter 0.6", Prestressing/orce=40 kips)
:t=

(6) H8S 4"x4"xO.25" (Lengrh=4')

27'-10·
1 ft =0.3048 m
1 in. =25.4 mm
Fig. 8a. Schematic of test setup for cantilever specimens.

with near surface mounted Leadline

bars (C2), CFRP strips (C4), and C-
Bars (C6) is compared to the un-
strengthened Specimen (Cl) as shown
in Fig. 10. Test results indicate identi-
cal behavior for all the specimens until
cracking occurred at a load level of
40.5 kips (180 kN) for the unstrength-
ened cantilever and 42.7 kips (190 kN)
for the strengthened cantilevers.
After cracking, a nonlinear behavior
was observed up to failure. The mea-
sured stiffnesses for the strengthened
specimens were higher due to the ad-
dition of the CFRP reinforcement.
The presence of CFRP reinforce-
ment precluded the flattening of the
Fig. 8b. Testing of cantilever specimens in laboratory. load-deflection curve, which was clear
in the control specimen at the load
range of 99 to 105 kips (440 and 466
applied at a distance of 13 in. (330 TEST RESULTS AND kN). Prior to yielding of the steel rein-
mm) from the cantilever edge. DISCUSION forcement at a load level of 99 kips
To prevent possible damage to the (440 kN), the stiffnesses of all
other cantilever during the fIrst test, an This section presents the test results strengthened cantilevers were almost
intermediate support was provided as of both cantilever and simply sup- the same and were 1.5 times higher
shown in Fig. 8a. The test setup for ported tests. The general behavior of than the stiffness of the un strength-
the cantilever tests is shown in Fig. each specimen is summarized in the ened cantilever.
8b. For the midspan tests, the slab was following subsections. The presence of the CFRP rein-
simply supported with a span of 16 ft forcement provided constraint to
(4.90 m), and the load was applied at Cantilever Tests opening of the cracks. Therefore, the
the center of the slab as shown in Figs. The load-deflection behavior of deflections were reduced and conse-
9aand 9b. cantilever specimens strengthened quently appeared to increase the stiff-

82 PCI JOURNAL
 (1) ElastQmeric bearings 1fi"x16"xO.1fi"

1 :ft "" 0.3048 m

1 in. "" 25.4 mm

4. •
"

I"
21'-10"

Fig. 9a. Schematic of test setup for simply supported specimens.

ness. After yielding of the tension

steel reinforcement, the stiffness of the
cantilever specimen strengthened with
Leadline bars, Specimen C2, was three
times higher than that of the un-
strengthened one.
Using C-BAR CFRP bars instead of
Leadline bars increased the stiffness
by an extra 20 percent. However,
using near surface mounted CFRP
strips yielded a stiffness increase of 35
percent compared to the Leadline bars.
For the control specimen, the increase
in the applied load was negligible after
yielding of the steel reinforcement.
For strengthened cantilevers, the load
resistance and deflection increased
until the concrete crushed in the com- Fig. 9b. Testing of simply supported specimens in laboratory.
pression zone. This is due to addi-
tional strength and stiffness provided
by the CFRP reinforcement. Specimens C3 and C5 up to a load tion curves shown in Figs. 10 and 11.
Fig. 11 shows the load-deflection level of 112 kips (500 kN). Initial cracking in the concrete sub-
behavior of cantilever specimens, C3 After yielding of the steel reinforce- strate at the anchorage zone was ob-
and C5, strengthened with externally ment, the stiffness of Specimen C5 served at a load level of 90 kips (400
bonded CFRP strips and sheets, re- was about 3.3 times higher than that kN) for Specimen C3, as shown in
spectively. The behavior of the control of the un strengthened cantilever. Fig. 12. Upon additional loading, the
specimen, Cl, is also shown for com- Loading of the cantilever specimens cracks continued to widen up to a load
parison. The figure clearly indicates was paused every 11 kips (50 kN) to level of 119 kips (530 kN), where un-
that the strength, stiffness, and ductil- manually record the strain in demec stable delamination occurred resulting
ity were significantly improved with points attached to the concrete sur- in peeling of the strips. The load
the addition of CFRP reinforcement. face. This was reflected by the contin- dropped to a level corresponding to
Identical behavior was observed for uous rise and fall in the load-deflec- the yield strength of the cross section

January-February 2002 83
Fig. 12. Initial cracking in concrete substrate of Specimen C3 at 90 kips (400 kN).

shown in Fig. 15. To investigate dif- Table 2. Experimental results of cantilever specimens.
ferent strengthening techniques, it was Percent
decided not to test a control specimen Per Ller p. LI. increase in
for the midspan and to rely on the re- Specimen Strengthening technique (kips) (in.) (kips) (in.) capacity
sults obtained from nonlinear finite el- Cl N.A. (Control) 40.5 0.36 107 3.64 -
C2 Near surface mounted Leadline bars 42.5 0.33 145.5 4.02 36
ement analysisY
C3 Externally bonded CFRP strips 43.2 0.36 119* 1.54 11
The finite element model was care- C4 Near surface mounted CFRP strips 42.0 0.33 153 3.66 43
fully calibrated to the behavior of the C5 Externally bonded CFRP sheets 43.6 0.36 154 4.41 44
un strengthened cantilever specimen. C6 Near surface mounted C-BAR 44.3 0.33 149 3.94 39
Strengthening of the specimens Note: 1 kip = 4.448 kN; 1 m. = 25.4 mm.
slightly increased the cracking load. * Specimen C3 failed due to delamination of CFRP strips, followed by crushing of concrete.
P" = cracking load
However, this behavior is also influ- P u = ultimate load
enced by the value of the tensile LI" = deflection at cracking
Llu = deflection at failure
strength used in the analysis in com-
parison to the actual values.
Results indicated a considerable in-
crease in stiffnesses and ultimate loads unstrengthened specimen. Identical be- ing of the steel. Prior to yielding of the
with the addition of the CFRP rein- havior was observed for all specimens bottom tension steel reinforcement,
forcement. Fig. 15 shows that the until cracking occurred at a load level the stiffnesses of all strengthened
CFRP reinforcement did not contribute of 78 kips (348 kN) for the unstrength- slabs were almost the same and were
very much to an increase in stiffness in ened slab and 81 kips (360 kN) for 1.5 times higher than the stiffness of
the elastic range of the slabs. However, strengthened slabs. the un strengthened slab. Specimens
the stiffnesses of the strengthened slabs The midspan deflection curves SS 1 and SS2, strengthened with near
were significantly enhanced it} the showed traditional nonlinearities due surface mounted Leadline bars and
post-cracking region compared to the to cracking of the concrete and yield- near surface mounted CFRP strips, re-

Fig. 13. Overview and closeup of typical failure due to crushing of concrete for cantilever specimens.

January-February 2002 85
 COST ANALYSIS
One of the prime objectives of
investigation was to provide a
fective analysis for each strfmgthelning
technique considered in this study ..
should be mentioned that all ~",,,hni,~,,,••
were designed to increase the
by 30 percent using the ch'lfaIGtelnsti.~
of each FRP material. An aPr)rmcim:ate
cost for each strengthening ''''-'l1111'1lW)
used for the cantilever specimens
given in Table 4. The total """~t1·n,,jCinn.
cost accounts for the cost of materilals,
equipment, and labor.
The percentage increase in the flexu·
ral capacity and the construction
for each of the strengthening tech·
niques considered in this study for
Fig. 14. Load versus crack width for cantilever specimens. cantilever specimens are shown in Fig.
17a. The figure indicates that using
near surface mounted CFRP strips and
spectively, showed comparable stiff- Fig. 16. The unstrengthened slab ex- externally bonded CFRP sheets pro·
nesses up to failure. hibited classical failure due to crush- vided the maximum increase in
After yielding of the tension steel ing of the concrete at a load level of strength. The construction cost of ex-
reinforcement at a load level of 148 167 kips (741 kN). Experimental re- ternally bonded CFRP sheets was only
kips (660 kN), the stiffnesses of Spec- sults of simply supported specimens 25 percent in comparison to near sur-
imens SS1 and SS2 were three times are given in Table 3. face mounted strips. Using either near
higher than that of the unstrengthened Strengthening the slab using near surface mounted Leadline bars or C-
slab. Using externally bonded CFRP surface mounted Leadline bars in- BAR CFRP bars provided approxi-
sheets in Specimen SS3 increased the creased the strength by 34 percent in mately the same increase in ultimate
stiffness by an extra 25 percent. comparison to the design value of 30 load carrying capacity.
Traditional flexural failure due to percent. Using near surface mounted With respect to cost, strengthening
crushing of the concrete at the CFRP strips instead of Leadline bars using C-BAR CFRP bars was 50 per-
midspan section was observed for all increased the strength by 38 percent. cent less. Using an efficiency scale (E) .
three specimens. Typical crack pattern Using externally bonded 'CFRP sheets defined by Eq. (1), the efficiency of
development for the strengthened sim- provided the highest increase III each technique was evaluated as shown
ply supported specimens is shown in strength by 50 percent. in Fig. 17b:

E = Percent increase in strength

300,-------------------------------------------~ Construction cost in USD
1 kip = 4.448 leN
1 in. = 25.4 mm (1)
/
250
The results show that strengthening
using externally bonded CFRP sheets
200
a is the most efficient technique in terms
of strength improvement and construc-
;g 150 tion cost. The estimated cost of the re-
~
oS habilitative work for the bridge under
100 consideration was approximately $1
million (US D), which was 60 percent
50 of the cost of demolition and replace-
_ _________ ~~.Di~~"_!9~_~1!:£!i.P~_(t~~_!® ____________ ment of the eiisting structure.
O+---~~r_~--_+--~--_r--~--;_--~--+_--~~

o 0.5 1.5 2 2.5 3 ANALYTICAL MODEll NG

Mid-spanddlection (in) This section summarizes the nonlin-
ear finite element analysis, conducted
Fig. 15. Load-deflection behavior of simply supported specimens. to simulate the behavior of the post-

86 PCI JOURNAL
 tensioned bridge slabs strengthened
with FRP. The analytical prediction is
compared to the experimental results.
Up-to-date but very limited informa-
tion has been reported on the use of
nonlinear finite element techniques to
simulate the overall behavior of pre-
stressed concrete members strength-
ened with FRP. Some analyses are re-
ported on the normal and shear stress
distributions at the end zones of the
FRP strip/sheet.
The finite element modeling de-
scribed in this paper was conducted
using the program ANACAP (Version
2.1). ANACAP is known for its ad-
vanced nonlinear capabilities of the
concrete material model. I2 The
ANACAP software employs the clas-
sical incremental theory of plasticity Fig. 16. Typical crack pattern development for strengthe,ned simply supported
that relates the increment of plastic specimens.
strain to the state of stresses and stress
increment.
Formulation of the yield surfaces, Interfacial failures are not consid- the symmetry of the cantilever slab,
loading, and failure surfaces take into ered in ANACAP. Consequently, only one-half of the slab in the longi-
account the effect of confinement on peeling failures cannot be predicted tudinal direction was modeled. The
the concrete behavior. The concrete and the analysis will be limited to fail- concrete was modeled using 20-node
material is modeled by the smeared ures due to crushing of the concrete or isoparametric brick elements with a
cracking methodology in which pro- rupture of the FRP reinforcement. The 2 x 2 x 2 reduced Gauss integration
gressive cracking is assumed to be dis- analysis is conducted using an incre- scheme. Each node has three transla-
tributed over an entire element.13 Over mental-iterative solution procedure, in tional degrees of freedom. The slab
the past 35 years, the distributed which the load is incrementally in- was supported on elastic springs hav-
smeared crack model has been used creased. Within each increment, equi- ing the same stiffness of the neoprene
for plane stress, plane strain and three- librium is iteratively achieved. pads used in the test program.
dimensional solid systems. 14• 15 Iteration is repeated until internal The load was applied as a uniform
Cracks are assumed to form perpen- equilibrium conditions are sufficiently pressure acting on an area of 9 x 24 in.
dicular to the principal tensile strain fulfilled and convergence is obtained. (228 x 610 mm). The load was applied
direction in which the cracking crite- At the end of each step, the ANACAP gradually using a step-by-step analy-
rion is exceeded. When cracking oc- program adjusts the stiffness matrix to sis. The number of load steps and step
curs, the stress normal to the crack di- . reflect the nonlinear changes in the size were chosen based on the experi-
rection is reduced to zero, which stiffness. Verification of the ence gained through different analyti-
/
results in redistribution of stresses ANACAP program using independent cal simulations conducted on bridge
around the crack. The ability of experimental results can be found deck slabs. The influence of the step
cracked concrete to share the tensile elsewhere. 16. 17 size at failure is performed and re-
forces with the steellFRP reinforce- ported elsewhere. IS
ment between cracks is modeled in Three analytical simulations were
Modeling of the Cantilever Slab carried out by varying the size of the
ANA CAP by means of the tension
softening model. The descending The cantilever slab and adjacent elements at the anticipated failure
branch of the tensile stress-strain panel were modeled to account for the zone and the number of layers within
curve is assumed to follow an expo- continuity effect. Taking advantage of the slab thickness. The number of ele-
nential function. Cracks are allowed to
form in the three principal directions. Table 3. Experimental results of simply supported specim~ns.
The program also accounts for the Percent
reduction in shear stiffness due to P cr J.,. Pu ,1,. increase in
cracking and further decay as the crack Specimen Strengthening technique (kips) (in.) (kips) (in.) capacity'"
SSI Near surface mounted Leadline bars 79 0.26 223 2.5 34
opens. 12 The reinforcement is assumed
SS2 Near surface mounted CFRP strips 83 0.26 230 2.4 38
to be distributed throughout the con- SS3 Externally bonded CFRP sheets 81 0.27 251 2.6 50
crete element. Full bond is assumed be- Note. 1 kip 4.448 leN, 1 m. 25.4 mm.
tween the reinforcement and concrete. * Ultimate load of unstrengthened specimen was based on nonlinear finite element analysis.

January-February 2002 87
Table 4. Cost analysis for cantilever specimens.
Strengthening Material Total Epoxy Labor Equipment Total
Specimen technique cost perft material cost cost hours* cost cost
C2 Near surface mounted 12.6 1058 150 7 67 1394
Leadline bars
C3 Externally bonded 17.0 1224 Included! 5 None 1309
CFRP strips
C4 Near surface mounted 17.0 1224 Includedt 9 67 1444
CFRP strips
C5 Externally bonded 7.0 252 150 4 34 354
CFRP sheets
C6 Near surface mounted 3.5 336 Included! 9 100 739
C-BAR
Note: All costs are in U.S. dollars.
* Labor cost is based on $17 USD per hour ($25 CAD per hour).
t Cost of epoxy is inclnded in the material's cost.

ments varied from 324 elements in the

60% 2000
first mesh to 924 elements in the third
~ .-
• % Increase in capacity mesh for the un strengthened speci-
50010 • Construction costs in U.S. dollars
men. Mesh dimensions for the three
,.-.. 1600 cases are shown in Fig. 18. The varia-
441)-
~
e..;.,
~
40010 391)- tion in element size was employed to
.€ 1200 .g provide a fine mesh around the area of
~ maXImum bendmg and shear stresses.
30% 00
::5 The predicted load-deflection be-
800 .S havior of the three numerical simula-
20010
~ tions is shown in Fig. 19, where the
U results are compared to the measured
10% 400
values for the un strengthened speci-
men, Cl. The response was relatively
0% o brittle when small size elements were
Externally Near surface Near surface Near surface
Externally
bonded CFRP mounted mountedC- mounted CFRP bonded CFRP used. Deformation at failure increased
strips Leadline bars BARCFRP strips sheets with the increase of the element size.
bars As expected, the deformational be-
Strengthening technique havior was quite similar for the vari-
ous simulations. The influence of the
Fig. 17a. Cost analysis of various strengthening techniques. mesh size on the predicted failure
loads was noticeable. The predicted
failure load was 97 kips (430 kN)
when coarse mesh was used. Refining
14 the finite element mesh resulted in a
/
12.3 considerable increase in the predicted
12 failure loads.
Compared to the test results, it
10
~ emerged for the second mesh that the
~ 8 flexural stiffness was predicted with
....5
(,)
6 5.2
sufficient accuracy up to failure. From
the above discussion, it is concluded
~ 4 3.0
that using the second mesh in model-
2.6 ing cantilever specimens revealed suf-
2 ficient accuracy. It should be men-
tioned that the corresponding
0
execution time was 50 percent less
Externally Near surface Near surface Near surface ExtemaIly than that of the third mesh.
bondedCFRP mounted mounted C- mounted CFRP bonded CFRP
strips Leadline bars . BAR CFRP strips sheets
bars Modeling of the Simply
Strengthening technique Supported Slab
Taking advantage of the symmetry
Fig. 17b. Efficiency of various strengthening techniques. of the simply supported specimen,

88 PCI JOURNAL
~?~~ MEIH1
....• .
•

", (324 elements) -

II
. liO ......
-
.,,~. ~~~..L..j-...;.I...~""""--'-"""'+-li""""'''''''''''''''''''''''''''''----''""-.......,..I-..,.~o:----.... ,§leyatjon

~i;~ _~I" "~~,.::~ ~'~r.~~,~,.i~c.~g~~~_~:_ __ .

;~ . , •......' .~ lL2'
Il~ .' .::!. . I-'++'I"+-!-I--+-+~~"""""'l-'.-+'-+-+""+-'-r-l-+-+-+-""""'-+-~-+"-"--+-~i-'----';+-"'-"-I flau.
~';,~>_~:cc _ , . ,f··. I I
I , 1 , I ;
0 .,
.:8.4Jt' 9" ,.
I
, l' 'S" J
I I

t f, 1 3·_~Jt
kt r 1l S ..7.. ·· J
'1< •.
':"t
+
I J I" .1 1

!i~
~~~• (4J~!~)!e" "H-+++1~I--+-+-+-t'-'. +-'
~m-nt._rx~7rzJJr -t'+ '+-'H'-+-I'-t'"+'M···,...,.···-.-I----+---.----+-----+--+--:---:-I

Sf_1"

"';' ":;, ,-

, ',- -', , "", : -!C' - : 0 '-' _, ~ , ',,/', -

~;~~.c~.i.!}'_~'4<.-·• . M·"",.>..,
,.ro.'.,.,'<!O:-··
••....., ~·~
•....• .. ....."..>..."jNJ
......!'"""'r_ .... e+.·~T,".t·.·r ·~I":~'·., · ~iT·X1':T"'n.X~. ;. ~..~--,.~"""!-
... ... ·. .;.· ·,.... - ... E!e..'tIation
...........,....... . •.•.
,.(•., ..•.• . ..•.•. " ;I:
,.·i;~~i.;'· ..••• 1
---I
J

Fig. 18. Investigation of influence of mesh size.

January-February 2002 89
 cantilever and simply supported
160 specimens. A comparison between
Specimen CI: Unstrengthened cantilever analytical and experimental results is
140 presented.
120
Cantilever Specimens
'<ii' 100 Experimental To validate the finite element model,
~ 80 Mesh 2 the unstrengthened cantilever was
110 modeled first. Predicted load-deflec-
....:l tion curves were compared to the ex-
60
.3"%4.72"<.,-,3" perimental results as shown in Fig. 20.
40 The predicted load-deflection behavior
using the finite element analysis com-
20 4"x5.9 "x5.33" 1 kip = 4.448 kN pared very well with the measured val-
1 in. = 25.4 rom ues. The predicted failure load using
0 the finite element analysis underesti-
0 0.5 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 mated the measured value by 7 per-
Cantilever deflection (in) cent. Failure was due to crushing of the
concrete at the face of the support.
Fig. 19. Influence of mesh size on load-deflection behavior. The second stage of the model de-
velopment involved the addition of
CFRP reinforcement. Six 3/g in. (10
only one-quarter of the slab was mod- slab was supported on spring elements mm) diameter Leadline bars were
eled using 20-node brick elements. To in the vertical direction to simulate the added to model the behavior of Speci-
focus on the slab behavior and to re- neoprene pads. The load was applied men C2, which was strengthened with
main within a realistic computer exe- as a uniform pressure acting on an near surface mounted Leadline bars.
cution time, the cantilevers were not area of 4.5 x 24 in. (114 x 609 mm). The epoxy was not modeled in the
included in the model since they are The load was applied gradually using analysis as no slip between the epoxy
not loaded. The specimen was dis- a step-by-step analysis. and the bars was observed during the
cretized into 255 elements. test. The CFRP bars were considered
The element size at the anticipated to be bonded to the concrete.
failure zone was set to 3 x 4.72 x 5.33
RESULTS OF THE
The predicted load-deflection be-
in. (76 x 120 x 135 mm), the same di- NUMERICAL ANALYSIS
havior is shown in Fig. 21. Compared
mensions that were previously de- This section discusses the results to the test results, the flexural stiffness
scribed in the cantilever model. The of the finite element analysis for both of the cantilever slab was simulated
with a very high accuracy using the fi-
nite element analysis. The predicted
failure load using the finite element
160
1 kip = 4.448 kN
analysis underestimated the measured
Specimen CI: Unstrengthe1ied cantilever
140 1 in. = 25.4 rom value by 6 percent.
/ Specimens Cl and C2 were ana-
120 lyzed also using a strain compatibility
~_E.......
xperimenta1 approach to predict the flexural behav-
'<ii' 100 ANACAP
ior up to failure. The concrete is as-
~ 80
sumed to be subjected to uniform uni-
axial strains over the entire width of
110
....:l 60
the slab. Strains were assumed to vary
linearly over the depth of the section.
40 The stress-strain relationship of the
concrete was modeled using a
20 parabolic relati9nship in compression.
The internal compression force in the
concrete was evaluated using the
0.5 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 stress-block parameters introduced by
Cantilever deflection (in) Collins and Mitchell in 1991.19
The stress-strain behavior of CFRP
Fig. 20. Comparison of predicted load-deflection behavior with laboratory results reinforcement was assumed to be lin-
(Specimen C1). early elastic up to failure. The deflec-

90 PCIJOURNAL
 was calculated by integrating the
at each load increment. Pre-
rnf'""hllrp.

loads and deflections at crack- 160Tr=~~~~~==~====~~----------------~

ing and at failure are given in Table 5. Specimen C2: Strengthened with
The predicted failure loads using 140 near swface mounted Leadline bars
strain compatibility approach underes-
timated the measured values forSpeci- 120
1 kip = 4.448 kN
mens Cl and C2 by 6 and 10 percent, 1 in. = 25.4 mm
respectively. Predicting the deflection
values using strain compatibility ap-
proach underestimated also the initial
stiffness of the slabs by 13 percent as 60
given in Table 5.
40

Simply Supported Specimens 20

The predicted load-deflection be-
O~~+-~+-~+-~+-~r-L-r-L-r-L-~~~~~~~~
havior of Specimen SSI, strengthened
with near surface mounted Leadline o 0.5 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
bars is shown in Fig. 22, compared to Cantilever deflection (in)
the measured values. Even though the
predicted behavior was very satisfac- Fig. 21. Comparison of predicted load-deflection behavior with laboratory results
tory, the predicted initial stiffness was (Specimen C2).
overestimated by 11 percent. Such a
phenomenon is a direct consequence
of previous bending tests conducted Table 5. Predicted results for Specimens Cl and C2.
on the two cantilevers.
Pc, L1., Pu L1u Pu predicted
In general, the predicted behavior Specimen Method _of analysis (kips) (in.) (kips) (in.) Pu experimental
was in a good agreement in terms of Experimental 40.5 0.36 107 3.64 -
cracking load, ultimate load and flexu- Cl Finite element 45.0 0.37 100 3.3 0.93
ral stiffness after cracking of the con- Strain compatibility 46 0.47 101 2.0 0.94
Experimental 42.5 0.33 145.5 4.02 -
crete. Failure was due to crushing of C2 Finite element 46 0.39 137 4.0 0.94
the concrete>at the location of the ap- Strain compatibility 47 0.42 131 2.7 0.90
plied load. The analysis predicted a Note: 1 kip = 4.448 kN; 1 tn. = 25.4 mm.
failure load of 204 kips (907 kN),
which was 9 percent less than the
measured value. Compared to the ex-
perimental results, the overall behav-
ior was well simulated using the 300~---------------------------------------------'

ANACAP program. 1 kip = 4.448 kN

1 in. = 25.4 mm
250
/ Experimental
CONCLUSIONS 881
Based on the findings of this inves- . - 200
tigation, the following conclusions can CIl
Q,
be drawn: ;g =-----AANN,AlMC;:AP: Unstrengthened slab
150
1. The use of near surface mounted ]
CFRP reinforcement is feasible and H
cost effective for strengthening or re- 100
pairing prestressed concrete girders
and slabs. 50 16'-00" 5'-11" I
2. Both the stiffness and strength of 27'-10"
t ;'

concrete slabs strengthened with O+---~--+---~--;---~---r--~---r--~---r--~---;

CFRP materials were substantially in-
creased. The ultimate load carrying o 0.5 1 1.5 2 2.5 3
capacity of the slabs can be increased Mid-span deflection (in)
by as much as 50 percent.
3. The magnitude of strength in- Fig. 22. Comparison of predicted load-deflection behavior with laboratory results
crease was influenced by the type and (Specimen SS1).

January-February 2002 91
configuration of the CFRP materials. 70 percent compared to the un- behavior. Analytical models des:cnt)Jngi!
In general, near surface mounted CFRP strengthened specimen. the load transfer mechanism of near
strips and externally bonded CFRP 9. The predicted results using non- surface mounted FRP reinforcement to
sheets provided superior strength in- linear finite element analysis were in concrete are urgently needed.
crease for both cantilever and simply excellent agreement with the experi- Work is currently under way by the
supported specimens. The overall cost mental results. The error is less than authors to provide complete design
of strengthening using CFRP sheets is 10 percent. guidelines regarding the development
only 25 percent of that using near sur- 10. Test results indicated full com- length needed for the proposed FRP
face mounted CFRP strips. posite action between the near surface strengthening techniques; these will be
4. Strengthening using either near mounted CFRP reinforcement and the reported in a future paper. The authors
surface mounted Leadline bars or C- concrete and no slip occurred through- recommend also that future research
BAR CFRP bars provided approxi- out the tests. be focused on examining the durabil-
mately the same increase in strength. ity aspects of various FRP strengthen-
With respect to construction cost, ing systems under severe environmen-
RECOMMENDATIONS AND
strengthening using C-BAR CFRP tal conditions.
bars is considerably less. FUTURE RESEARCH
5. Strengthening using externally Test results of the experimental
bonded CFRP strips provided the least program and predicted values using ACKNOWLEDGMENTS
increase in strength by 11 percent due numerical modeling provided suffi- The authors wish to acknowledge
to peeling of the strips from the con- cient evidence and confidence of the the support of the Network of Centres
crete surface. Using the same amount proposed strengthening technique of Excellence, ISIS Canada, program
of strips but as near surface mounted using near surface mounting as a new of the Government of Canada and the
reinforcement enhanced the ultimate and promising technology. Delamina- Natural Science and Engineering Re-
load carrying capacity by 43 percent. tion-type failures, occasionally ob- search Council. The writers gratefully
Groove dimensions of 0.2 in. wide by served by using externally bonded re- acknowledge the support provided by
1 in. deep (5 x 25 mm) were adequate inforcement, can be precluded using Mitsubishi Chemical Corporation,
to prevent splitting of the epoxy cover. this technique. Japan, Marshall Industries Composites
6. Strengthening using externally Epoxy adhesives commonly used for Ltd., United States, S&P Clever Rein-
bonded CFRP sheets is the most effi- bonding steel bars have proved to be forcement Company, Switzerland, and
cient technique in terms of strength effective in bonding CFRP bars. How- Master Builders Technologies Ltd.,
improvement and construction cost. ever, characterizing the bond perfor- United States, for providing the mate-
7. Delamination of the CFRP strips mance of the adhesives is compulsory rials used in the test program.
occurred at a strain of 0.54 percent, prior to any application. It is recom- The authors would like also to ac-
which is equivalent to 41 percent of mended to use the groove dimensions knowledge the support provided by
the rupture strain of the strips as re- outlined in this paper for !lear surface Vector Construction Ltd., Winnipeg,
ported by the manufacturer. The de- mounted bars and strips for strengthen- Canada, for performing all the
lamination strain is 18 percent less ing reinforced and prestressed concrete strengthening work. Special thanks are
than the value recommended by ACI structures. owed to M. McVey for his assistance
Committee 440F. Additional research is needed to du- during the fabrication and testing of
8. Using near surface mounted ~plicate the findings of this research the specimens. Finally, the authors
CFRP strips or externally bonded program and to determine the effect of want to thank the PCI JOURNAL re-
CFRP sheets for flexural strengtlr'en- different parameters, such as fatigue viewers for their thoughtful and con-
ing reduced the crack width by 50 to loading and temperature, on the overall structive comments.

92 PCI JOURNAL
 REFERENCES
1. u.s. Department of Transportation (DOT), Bureau of Trans- 10. Canadian Standards Association, Canadian Highways Bridge
portation Statistics, Transportation Statistics Annual Report, Design Code, Section 16, Fiber Reinforced Structures, Ottawa,
Washington DC, 1997,286 pp. Ontario, 1996,28 pp.
2. Tann, D. B., and Delpark, R., "Experimental Investigation of 11. Hassan, T., Horeczy, G., Svecova, D., Rizkalla, S., Shehata,
Concrete Beams Reinforced with Narrow Carbon Strips," Pro- E., and Stewart, D., "Flexural Strengthening of Post-Tensioned
ceedings of the International Conference on Structural Faults Bridge Slab Using FRP," Proceedings of the International
and Repair, CD-ROM, 1999. Conference on Advanced Composite Materials for Bridges and
3. Corry, R., and Dolan, C. W., "Strengthening and Repair of a Structures (ACMBS-llI), Ottawa, Ontario, 2000, pp. 291-298.
Column Bracket Using Carbon Fibre Reinforced Polymer 12. James, R. G., ANA CAP Concrete Analysis Program Theory
(CFRP) Fabric," PCI JOURNAL, V. 46, No.1, January-Febru- Manual, Version 2.1, Anatech Corp., San Diego, CA, 1997.
ary 2001, pp. 54-63. 13. Rashid, Y. R., "Analysis of Prestressed Concrete Pressure Ves-
4. Burke, C. R., and Dolan, C. W., "Flexural Design of Pre- sels," Nuclear Engineering Design, V. 7, No.4, 1968, pp. 334-
stressed Concrete Beams Using FRP Tendons," PCI JOUR- 344.
NAL;V. 46, No.2, March-April 2001, pp. 76-87. 14. Scordelis, C. A., "Past, Present and Future Development,"
5. AASHTO, Standard Specifications for Highway Bridges, Seminar on Finite Element Analysis of Reinforced Concrete
American Association for State Highway and Transportation Structures, Japan Concrete Institute, V. 1, 1985, pp. 203-212.
Officials, Washington, DC, 1998. 15. Gerstle, K. H., "Material Modeling of Reinforced Concrete,"
6. Integrated Finite Element Analysis and Design of Structures, IABSE Colloquium on Advanced Mechanics of Reinforced
SAP2000 Graphic User Inteiface Manual, Version 7.0, Com- Concrete, Introductory Report, Delft, Netherlands, 1981.
puter and Structures Incorporation, Berkeley, CA, 1998. 16. Hassan, T., Abdelrahman, A., Tadros, G., and Rizkalla, S.,
7. Blaschko, M., and Zilch, K., "Rehabilitation of Concrete "FRP Reinforcing Bars for Bridge Decks," Canadian Journal
Structures with Strips Glued Into Slits," Proceedings of the of Civil Engineering, V. 27, No.5, 2000, pp. 839-849.
12th International Conference on Composite Materials, Paris, 17. Mufti, A., Hassan, T., Memon, A., and Tadros, G., "Analytical
France, CD-ROM, 1999. Study of Punching Shear Strength of Restrained Concrete
8. ACI Committee 440F, "Guide for the Design and Construction Slabs," Proceedings of the Canadian Society of Civil Engi-
of Externally Bonded FRP Systems for Strengthening Con- neering Annual Conference (CSCE), Victoria, British
crete Structures," Draft Report, American Concrete Institute, Columbia, CD-ROM, 2001.
Farmington Hills, MI, October 2001,150 pp. 18. Hassan, T., "Behavior of Concrete Bridge Decks Reinforced
9. Japan Society of Civil Engineers (JSCE), 1997b, "Recommen- With FRP," M.Sc. Thesis, University of Manitoba, Canada,
dations for Design and Construction of Concrete Structures 1999,226 pp.
Using Continuous Fiber Reinforcing Materials," Concrete En- 19. Collins, M., and Mitchell, D., Prestressed Concrete Structures,
gineering Series, No. 23, Tokyo, Japan, 1997,325 pp. Prentice Hall, Englewood Cliffs, NJ, 1991,766 pp.

January-February 2002 93