Composites: Part B 31 (2000) 445±452

www.elsevier.com/locate/compositesb

Behavior of full-scale reinforced concrete beams retro®tted for shear

and ¯exural with FRP laminates
D. Kachlakev a,*, D.D. McCurry b
a
Department of Civil, Construction and Environmental Engineering, Oregon State University, 202 Apperson Hall, Corvallis, OR 97331-2302, USA
b
CH2M HILL, 2300 Northwest Walnut Boulevard, Corvallis, OR 97330-3538, USA
Received 21 February 2000; accepted 12 April 2000

Abstract
Four full-scale reinforced concrete beams were replicated from an existing bridge. The original beams were substantially de®cient in shear
strength, particularly for projected increase of traf®c loads. Of the four replicate beams, one served as a control and the remaining three were
implemented with varying con®gurations of carbon ®ber reinforced polymers (CFRP) and glass FRP (GFRP) composites to simulate the
retro®t of the existing structure. CFRP unidirectional sheets were placed to increase ¯exural capacity and GFRP unidirectional sheets were
utilized to mitigate shear failure. Four-point bending tests were conducted. Load, de¯ection and strain data were collected. Fiber optic gauges
were utilized in high ¯exural and shear regions and conventional resistive gauges were placed in eighteen locations to provide behavioral
understanding of the composite material strengthening. Fiber optic readings were compared to conventional gauges.
Results from this study show that the use of ®ber reinforced polymers (FRP) composites for structural strengthening provides signi®cant
static capacity increases approximately 150% when compared to unstrengthened sections. Load at ®rst crack and post cracking stiffness of all
beams was increased primarily due to ¯exural CFRP. Test results suggest that beams retro®t with both the designed GFRP and CFRP should
well exceed the static demand of 658 kN m sustaining up to 868 kN m applied moment. The addition of GFRP alone for shear was suf®cient
to offset the lack of steel stirrups and allow conventional RC beam failure by yielding of the tension steel. This allowed ultimate de¯ections to
be 200% higher than the pre-existing shear de®cient beam. If bridge beams were retro®t with only the designed CFRP failure would still
result from diagonal tension cracks, albeit at a 31% greater load. Beams retro®t with only the designed shear GFRP would fail in ¯exure at the
mid-span at an equivalent 31% gain over the control specimen, failing mechanism in this case being yielding of the tension steel. Successful
monitoring of strain using ®ber optics was achieved. However, careful planning tempered by engineering judgement is necessary as the
location and gauge length of the ®ber optic gauge will determine the usefulness of the collected data. q 2000 Elsevier Science Ltd. All rights
reserved.
Keywords: Fiber reinforced polymers

1. Signi®cance and scope of research loads. Since the beams were cast without any steel stirrups,
the concrete provided the entire shear resistance. Two of the
Upto 40 percent of the bridges in both the United States and longitudinal and all cross beams were also 50 percent de®-
Canada are structurally de®cient [1]. Structural elements cient in ¯exural capacity. The bridge, noted to be a historic
composed of concrete and reinforcing steel are frequently structure, needed a retro®t scheme that maintained its origi-
determined inadequate to sustain current or new load levels nal appearance. After careful considerations, FRPs were
imposed on bridges. Recently, a new method of retro®t has selected to strengthen the beams. To verify the strengthen-
been studied that uses the addition of externally bonded ®ber ing design and to investigate the structural behavior of FRP
reinforced polymers (FRPs) to increase load capacity. reinforced full size members, full-scale replicate reinforced
Horsetail Creek Bridge is located in Columbia Gorge, concrete (RC) beams were fabricated in Oregon State
Oregon. The bridge was built in 1912, and consists of University laboratories. Experimental beams were repli-
three 610 cm spans. The two cross-beams were found to cated to match geometry and ultimate strength of the exist-
be 95 percent de®cient in shear, relative to the current traf®c ing beams. In order to avoid any variables between the real
structure and the laboratory replicates, the beams were
* Corresponding author. Tel.: 11-541-737-6154; fax: 11-541-737-3052/
retro®tted by the contractor who performed the strengthen-
6154. ing of the bridge. Concrete and steel reinforcement proper-
E-mail address: [email protected] (D. Kachlakev). ties and locations were matched as close as possible.
1359-8368/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1359-836 8(00)00023-8
446 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452

Table 1 studies involving ¯exural bending of FRP reinforced RC

Experimental beam description beams has been reported [6,11±14]. Load capacity increases
BEAM Description are usually approximately 1.5±5.0 times that of the
unstrengthened beam, depending upon FRP system type,
Control Typical beam with no shear thickness of the composite reinforcement, ®ber angle orien-
stirrups and bottom ¯exural steel
tation, geometry and condition of the original beam. Many
Flexure-only Control w/ added ¯exural carbon
FRP (CFRP) reinforcing of these studies do not involve full-scale specimens.
Shear-only Control w/ added shear glass Many applications of ®ber optic monitoring are available
FRP (GFRP) reinforcing now or likely to be in the future [15]. One of the main
Shear&Flexure Control w/ added shear GFRP advantages for using ®ber optic strain gauges is that they
and ¯exural CFRP reinforcing
are less sensitive to environmental effects than conventional
resistive gauges. Strains can potentially be monitored over
This study investigates the load carrying capacity and kilometers. After the completed retro®t, Horsetail Creek
bending characteristics as the result of FRP application to Bridge was classi®ed as a smart structure in that strains
inadequate RC beams by examining de¯ection and strain as can be monitored in real-time [2]. Fiber-optic technology
a function of the external load. Fiber optic data compli- is useful for structural monitoring and the technology is
mented conventional resistive strain gauges. Fiber optic increasing in accuracy and reliability.
sensors were also installed on the actual bridge to monitor Long gauge lengths (1000±700 mm) collected strain data
static, dynamic and long-term load response [2]. Data has for the experimental beams, exactly replicating the ®ber
been collected already for almost one year and will be optics installed on the bridge. However, for most experi-
compared to ®nite element models of the bridge [3]. mental applications, small strains at discrete locations result
in failure of structural members. Thus, the data gathered by
the ®ber optics dif®cult in this study was supplemented by
2. Preview of previous work
reading provided by regular strain gages.
Fiber reinforced polymer materials present signi®cant
potential in the ®eld of Civil Engineering. FRPs are not 3. Experimental description
chronologically new materials, but have been used in the
aerospace industry for a number of decades. Limited All four beams were constructed of the same geometry
demand and manufacturing of FRP material has tradition- (6096 £ 305 £ 762 mm 3) and placement of steel and were
ally resulted in high material costs. The ®rst structures built closely matched to Horsetail Creek bridge beams. Concrete
utilizing these materials were in the late 1950s by Prof. H. used was Type I, 20.7 MPa (28 day) strength with 152 mm
Isler [4]. The ®rst bridges to use FRP materials in the USA slump. Beams were cured in a moist condition until
dated from the late 1980s. removed from the forms fourteen days after pouring. Four
The need to prevent shear failure in older RC bridges has different FRP reinforcing schemes were used to compare the
been so signi®cant that numerous studies have focused on effects of each. A description of each beam is given in Table
shear strengthening [5±9]. Full-scale and scaled experi- 1, hereafter referred to as the Control, F-only, S-only and
mental studies have shown that proper quantity and place- S&F beams for discussion.
ment of FRP for shear strengthening is needed in order to
ensure adequate strength [7,6]. Although not a common 3.1. Material properties
practice, ®eld applications of FRP for shear strengthening
Structural properties used in the experiments are given in
are also recorded [10]. The number of FRP retro®t projects
Table 2. Note that the properties given for the CFRP and
involving shear strengthening is likely to grow dramatically,
GFRP are composite properties and not properties of the
since many RC structures constructed in the ®rst half of the
®ber. All properties are provided by the manufacturer [16].
twentieth century were inadequately reinforced for today's
traf®c demands. 3.2. Data collection
Flexural behavior is better understood, in contrast to
shear, torsion and combined loading. Basic, yet essential Three locations of de¯ection were collected to ensure an
Table 2
Design material properties

Material Limiting stress (MPa) Limiting strain Limit state Elastic modulus (GPa)

Concrete 20.7 0.003 Crushing 21.5

Steel 414 0.002 Yielding 200
Glass FRP 414 0.02 Rupture 20.7
Carbon FRP 760 0.012 Rupture 62
 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452 447

Fig. 1. De¯ection and horizontal strain data collection (internal steel reinforcing shown as dashed lines).

understanding of the beam deformation. Critical strains at and strain data. As visible in Fig. 4 the shear FRP reinforc-
the mid-span section and two sections in high shear regions ing successfully stopped the shear cracks from growing. All
were collected as shown in Fig. 1 Gauges were placed on the tested beams exhibited unique failure mechanics. The
concrete surface, FRP surface and inside the beam on the control beam failed in diagonal tension, the F-only beam
steel, and were protected appropriately before casting of the displayed debonding over the support point and the S-only
beams. Fiber optic gauges were utilized only on the FRP beam failed by yielding of the tension reinforcement. The
reinforced beams (i.e. no ®ber optics were collected on the S&F beam would likely fail due to crushing of the concrete
Control beam). These gauges were placed according to the in compression. Shear failures are considered those that
geometry shown in Fig. 2. occur from diagonal tension cracking in high shear spans
and ¯exural failures occurred at mid-span. A brief explana-
tion of each mechanism follows.
4. Experimental results A summary of the experimental load and de¯ection
results is presented in Table 4. From this table, it is clear
FRP reinforced beams exhibited signi®cantly higher
that the addition of FRP not only increases the sustainable
load-carrying capacity and stiffness that the control beam.
load, but also changes the de¯ection characteristics. All
Failure modes are presented in Table 3. The S&F beam was
beams revealed higher ¯exural stiffness. Since the control
tested beyond the design truck loads for Horsetail Creek
beam was de®cient in shear de®cient it failed failing long
bridge, but did not exhibit failure due to limitations of the
before ¯exural capacity was achieved. Therefore, the ulti-
equipment maximum load. Even after an adjusted load
mate de¯ections for all FRP-strengthened beams were
con®guration to increase the applied moment to 1.3 times
increased. It is important to realize that the experimental
the ultimate design value (658 kN m), the S&F beam did not
gains resulting from using the FRP would not be as signi®-
show critical strain values. De¯ections were visible, but the
cant if the beam was only de®cient in ¯exure.
beam did not exhibit signs of failure. The load was held for
30 min to observe real time strains, but no indication of the 4.1. Fiber optic strain data
beam achieving critical strains was noticed. De¯ections
were more evident than the ®rst testing, but did not increase Fiber optic gauges in the shear region collected data over
under a constant moment of 868 kN m. This fully reinforced a 700 mm gauge length, which included a shear crack near
beam was then preserved for future cyclic loading to gain the beam bottom and near-zero strains near the top of the
understanding of fatigue behavior. beam. Strains near the bottom of the beam and near the
The behavior of all four beams is presented in Fig. 3. support were very high due to diagonal tension cracking.
Beam cracking was recorded to compliment the de¯ection In contrast, strains near the top of the beam over the supports

Fig. 2. Fiber optic gauge locations.

448 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452

Table 3 capacity of an FRP strengthened concrete beam. To acquire

Beam failure modes a good bond to the beam surface, proper surface preparation
Beam Failure mode of the concrete beam requires a number of steps, including
mechanical abrasion. This was done with experimental and
Control Diagonal tension crack (shear failure) Horsetail Creek bridge beams. In addition, FRP systems
F-only Diagonal tension crack (shear failure)
must be approved for use. The system used in strengthening
S-only Yielding of tension steel followed by crushing of
compression concrete after extended de¯ections experimental and bridge beams was an approved system,
S&F Not observed. Believed to be yielding of tension steel commonly used in projects worldwide. This phenomenon
followed by crushing of the concrete. should be investigated further.

were very small (Fig. 5). Thus, providing a gauge over this 5.2. Shear capacity due to ¯exural reinforcement
distance provided useful information about the average
As can be seen from Table 4, the F-beam sustained
strains.
substantially more load than the control beam. However,
the failure of the F-beam was related to the same de®ciency
5. Discussion that the control beam exhibited. Since shear reinforcing was
absent, the addition of CFRP for ¯exure should not be
5.1. Lag in FRP strain expected to add shear strength. In the classical sense, the
CFRP added no additional shear strength. Diagonal tension
Nearly every calculation of the moment capacity of an cracks were visible at nearly the same load levels as the
FRP reinforced beam assumes that FRPs are adhered control beam. However, since the CFRP was wrapped up
perfectly to the beam surface. Although it is understood the sides (see Fig. 4), the CFRP was able to equilibrate
that a perfect bond does not exist, it would be desirable to forces across the diagonal tension crack. In conventional
quantify how imperfect the bond might be. In this experi- calculations, horizontal structural components are not
mental study, a deviation from the plane-sections-remain- allowed to be used to resist diagonal tension cracking.
plane assumption was observed. This was evident for the F- However, the layers of CFRP bridging the critical diagonal
only and S&F beam in particular as visible from the strain tension crack on both ends of the beam displayed the signi®-
pro®les in Fig. 6. This plot suggests that the strain observed cant capability that these materials posses.
in the FRP was near that in the main reinforcing steel. This The CFRP ®bers were able to maintain integrity of the
is obviously not in agreement with common bending beam in the presence of the shear crack. The shear failure of
assumptions. This behavior is likely because the ®bers the beam was accompanied by transverse rupture of the
were not initially taught/straight, since the FRP was applied composite, noted to be shear failure of the matrix in the
by hand without pre-tensioning. A more detailed under- direction parallel to the carbon ®bers. Since the composite
standing of this behavior is needed to truly predict the was intended to be horizontally unidirectional in strength,

Fig. 3. Load vs. de¯ection for experimental beams; Control Beam; Flexure-only beam (F-Only); Shear-only beam (S-only); Shear & ¯exure beam (S&F).
 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452 449

Fig. 4. Experimental beam cracking.

straining of the composite in the vertical direction would achieve the shear or moment capacity of the fully FRP
undoubtedly result in premature failure. Cracking in reinforced beam, reasonable assumptions can be made
concrete is the result of tension, whether diagonal or hori- from the available data. Calculations suggest that the ulti-
zontal, as understood from Mohr's circle of stress. mate load of the beam would be limited by crushing of the
concrete. Concrete strains at the top-mid-span location were
5.3. Failure of the shear & ¯exure beam approaching 0.0015 at the maximum applied load. Strains in
the CFRP reinforcing at mid-span were approaching 0.003
Although the experimental testing machine was unable to and strains in the main tension reinforcing steel were
450 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452

Table 4
Summary of experimental load and de¯ection

Item Control value Comparison to control beam (%)

F-only S-only S&F

Midspan de¯ection at 67 kN 1.18 mm 103 105 93.5

Max observed de¯ection 24 mm 124 144 104
Midspan de¯ection at failure 24 mm 124 208 a Not observed
Load at failure 476 kN 145 145 Not observed b
Load at ®rst signi®cant cracking 78.3 kN 123 112 123
Calculated moment capacity 473 kN m 188 100 188
Experimental maximum applied 436 kN m 145 145 200
moment
a
Extrapolated from data (device failed).
b
152% was achieved during loading.

slightly in excess of 0.002. This is clear evidence to the This should be understood and over-strengthening the
projected failing sequence of the beam, where the steel beam with FRP might be good practice to account for uncer-
yields, extended de¯ections result, the concrete crushes tainties and the fact that the beam is very likely to fail by
and the FRP ruptures from substantial de¯ections. What is crushing of the concrete.
resulted is an ªover-reinforcedº beam in the classical sense.
This behavior should be anticipated for any FRP reinforced
beam that has adequately adhesion and anchoring to the 6. Conclusions
beam and was designed originally as an under-reinforced
beam. These are observations with adequate evidence to be
A very important point every designer should realize is considered conclusive.
that the added ¯exural capacity of FRP to most RC beams is
not an amazing structural accomplishment. Most concrete 1. FRP retro®t for structural strengthening can increase
beams were intended to fail by yielding of the tension steel static load capacity upwards of 150% of the original
before the concrete was able to crush. This results in a beam capacity, depending on the mode of failure, geo-
ªductileº beam, which is able to undergo visible de¯ections metry and material properties.
before ultimately losing load carrying capacity. Most retro®t 2. Horsetail Creek bridge beams retro®tted only with the
projects where FRP is used to enhance a de®cient member ¯exural CFRP would still result in diagonal tension
will involve under-RC beams. The engineer is then taking failure albeit at a 31% greater load. Since the CFRP
that additional capacity to a level of over-reinforcement. was intended to provide ¯exural reinforcing, it was

Fig. 5. Experimental load vs. strain at mid-span.

 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452 451

Fig. 6. FRP horizontal strain ªlagº observed from F-only beam test.

horizontally unidirectional. The CFRP was wrapped conducted. Many studies published to-date do not involve
up the sides a suf®cient amount to provide resistance full-scale specimens. Few publications report large sample
across the diagonal tension crack. In addition, the size, full-scale testing, undoubtedly because of economic
increased stiffness provided by the CFRP decreased reasons. For increased con®dence of beam response and to
the deformation and offset cracking by reducing assure proper use of FRP materials, full-scale specimens
strain in the beam. should be tested. It is also recommended that investigation
3. The addition of GFRP for shear was suf®cient to be conducted into the strain lag of FRP (slightly imperfect
offset the lack of stirrups and cause conventional bond observed). This phenomenon may cause unacceptable
RC beam failure by steel yielding at the mid-span. inaccuracies in calculations. In addition to experimental
This allowed ultimate de¯ections to be 200% higher studies, analytical investigation should be made into the
than the shear de®cient control beam, which prema- strain lag of FRP. Shear deformations of the epoxy resin
turely failed due to a signi®cant diagonal tension can be estimated with theoretical calculations. Such studies
crack. are likely to help develop more accurate assumptions to be
4. Horsetail Creek bridge beams retro®tted with both used in design calculations.
the GFRP for shear and CFRP for ¯exure should
well exceed the static demand imposed by the new
traf®c loads of 658 kN m sustaining up to 868 kN m Acknowledgements
applied moment.
5. Load at ®rst crack was increased, primarily due to The study was funded by the US Department of Trans-
the added stiffness of the ¯exural CFRP, by approxi- portation, Federal Highway Administration and by the
mately 23%. This added stiffness reduces the de¯ec- Oregon Department of Transportation. The author thanks
tions, which in turn reduced the strains and stresses Fyfe Corp., and Contech Services Inc. for supplying the
in the cross-section. materials and completing the FRP installation. My apprecia-
6. Addition of ¯exural CFRP offset the load at yielding tion also goes to the Department of Civil, Construction and
of tension steel beyond 33%. Environmental Engineering Department at Oregon State
7. An imperfect bond was evidenced from the strain lag University for providing the necessary facilities for this
of the CFRP from the expected plane-sections- study.
remain-plane assumption. Due to the extremely
small sample size of the experiment (i.e. only four References
specimens each with different reinforcing), further
investigation is necessary to determine if this effect [1] Kachlakev D. Strengthening Bridges Using Composite Materials.
is of signi®cance in structural analysis and safety. Report FHWA-OR-RD-98-08, US Department of Transportation
Federal Highway Administration and Oregon Department of Trans-
portation, Washington, DC, 1998.
7. Recommendations
[2] Udd E. Fiber Optic Grading Sensors for Static and Dynamic Strain
Measurements on Bridges to Support Health and Vehicle Monitoring
To fully quantify the behavior of FRP reinforced beams it System, Northwest Transportation Conference 2000, Oregon State
is recommended that statistical full-scale studies be University, Corvallis, Oregon, 2000.
452 D. Kachlakev, D.D. McCurry / Composites: Part B 31 (2000) 445±452

[3] Kachlakev D. Finite Element Modeling and Experimental Analysis of [9] Al-Sulaimani GJ, Sharif A, Basunbul I, Baluch M, Ghaleb B. Shear
FRP-Retro®tted Bridges, Northwest Transportation Conference 2000, repair for reinforced concrete by ®berglass plate bonding. Struct J,
Oregon State University, Corvallis, Oregon, 2000. ACI 1994;91(3):458±64.
[4] Rizkalla S, Labossiere P. Planning for a New Generation of Infra- [10] Crasto A, Kim R. Rehabilitation of concrete bridge beams with ®ber-
structure: structural engineering with FRPÐin Canada. Concrete Int reinforced composites. 42nd International SAMPE Symposium, 1997,
1999:25±8. p. 77±83.
[5] Meier U, Deuring M, Meier H, Schwegler G. Strengthening of struc- [11] GangaRao HVS, Vijay P. Bending behavior of concrete beams
tures with CFRP laminates: research and applications in Switzerland. wrapped with carbon fabric. J Struct Engng, ASCE 1998;124(1):3±
Advanced Composite Materials in Bridges and Structures. Canadian 10.
Society for Civil Engineering, 1992. p. 243±51. [12] Rostasy FS, Hankers C, Ranisch EH. Strengthening of R/C- and P/C-
[6] Kachlakev D, Barnes W. Flexural and shear performance of concrete Structures with Bonded FRP Plates. Advanced Composite Materials
beams strengthened with ®ber reinforced polymer laminates. Fourth in Bridges and Structures. Canadian Society for Civil Engineering,
International Symposium on Fiber Reinforced Polymer Reinforce- 1992. p. 255±63.
ment for Reinforced Concrete Structures. The American Concrete [13] Ritchie PA. External Reinforcement of Concrete Beams Using Fiber
Institute, SP-188, 1999. p. 959±72. Reinforced Plastics. Struct J, ACI 1991;88(4):490±6.
[7] Cheng RJJ, Hutchinson R, Rizkalla S. Rehabilitation of concrete [14] Saadatmanesh H, Ehsani M. RC beams strengthened with GFRP
bridges for shear de®ciency using CFRP sheets. 42nd International plates I: experimental study. J Struct Engng, ASCE
SAMPE Symposium, 1997. p. 325±35. 1991;117(11):3417±33.
[8] Arduini M, D'Ambrisi A, Di Tommaso A. Shear failure of concrete [15] Udd E. Fiber Optic Smart Structures. Proc IEEE 1996;84(1):60±66.
beams reinforced with FRP plates. Infrastruct Repair Meth 1999:123± [16] Fyfe Corporation, LLC. Application Manual, San Diego, California,
30. 1998.