Professional Documents
Culture Documents
Michael E. Kreger, FACI, is the Dewitt C. Greer Centennial Professor at the University
of Texas at Austin. He is a member of ACI Committees 215, Fatigue of Concrete; 318-H,
Seismic Provisions; 374, Performance-Based Seismic Design of Concrete Building; and
Joint ACI-ASCE Committees 352, Joints and Connections in Monolithic Concrete
Structures, and 441, Reinforced Concrete Columns.
Laboratory specimens
Eighteen beams were strengthened using composite
materials from four different manufacturers. The remaining
two specimens were used as control specimens to compare
the response of the bare reinforced concrete beams with the
strengthened specimens. All the reinforced concrete beams
had rectangular cross sections (Fig. 1). The test specimens
were designed assuming that strains varied linearly with
depth and that perfect bond existed between the concrete
surface and the composite materials. Each specimen was
proportioned such that the composites would rupture before Fig. 2—Experimental setup.
the concrete crushed at the extreme compression fiber. To
achieve the design goal, two different sizes of beams were reinforcement. As summarized in Table 1, the actual material
required because of the differences in strength of the composite properties exceeded the values assumed for design.
systems used in this study. Ten of the beams were 8 in. wide x
14 in. deep (203 x 356 mm) and ten beams were 8 in. wide CFRP Composites
x 16 in. deep (203 x 406 mm). The beams with the smaller The four composite material systems used in this study
cross sections had spans of 106 in. (2.69 m) and the beams with were selected because of their immediate availability for the
the larger cross sections had spans of 118 in. (3.0 m). construction industry. Three of these materials are manu-
The test setup is shown in Fig. 2. Two vertical loads were factured in the form of nonimpregnated carbon fiber sheets.
applied symmetrically about the midspan of the beams to Two used unidirectional carbon fibers (Composite Systems A
generate a 22 in. (560 mm) region of constant moment. The and B) and one comprised a woven fabric (Composite
beams were supported on 36 in. (915 mm) tall concrete pedestals System C). These dry-fiber systems were bonded to the
so that the researchers could safely observe the bottom face of reinforced concrete beams using a wet-layup process. The
the beams during the tests. Elastomeric bearing pads were used fourth product (Composite System D) was fabricated in the
to support the ends of the beams to avoid crushing of the concrete form of stiff pultruded plates that were bonded to the surface of
and to permit unrestrained rotation. The shear span to depth the concrete using an epoxy-based paste. Material properties
ratio was approximately the same for both sets of beams. for the four composite systems are reported in Table 2.
The reinforcement details are shown in Fig. 1. All beams Detailed information about each of the composite systems is
were reinforced with two No. 5 bars in tension and No. 6 available in the technical literature provided by the manu-
gage wire stirrups spaced at 4 in. (102 mm) on center within facturers (Fyfe 1997; Master Builders 1998; Mitsubishi
the shear span. Two No. 3 bars were added as longitudinal Chemical 1999; Sika Corp. 1997 and 1999).
reinforcement in the compression zone to facilitate fabrication The results from previous experimental studies indicated
of the reinforcing cages. that debonding of the CFRP from the surface of the reinforced
All beams were designed using a nominal 28-day concrete concrete beams is the likely mode of failure (Arduini, Di
compressive strength equal to 4500 psi (31 MPa), an assumed Tommaso, and Nanni 1997; GangaRao and Vijay 1998;
yield stress for the longitudinal reinforcement of 60 ksi Garden and Hollaway 1998). Therefore, the objective of the
(420 MPa), and an assumed yield stress of 75 ksi (520 MPa) first series of tests was to determine if debonding could be
for the smooth wire that was used to fabricate the transverse prevented by increasing the bonded area between the CFRP
composites and the concrete. In this phase of the investigation, Two in. (50 mm) wide transverse straps were added along
the CFRP composites were attached to the bottom face of the the shear span in Configuration II. The straps were wrapped
beams (Configuration I in Fig. 3). As will be discussed in detail around the bottom of the cross section and extended vertically
later in the paper, the CFRP materials debonded from the to within 3 in. (75 mm) of the compression face. The center-
to-center spacing of the straps was half the depth of the beam
surface of the concrete for all of these test specimens, even
for all specimens. As testing progressed, the number of
when the composites extended nearly the entire length of the transverse straps and the bonded length of the longitudinal
shear span. Therefore, three additional configurations of the composites were reduced.
composites were developed (Configurations II thorough IV) The longitudinal composites were positioned on the sides
to delay debonding. The goal was to develop the rupture of the beams in Configuration III, rather than on the bottom
strength of the composites, although this was only achieved surface. This arrangement of the composites reduced the prying
in four of the 18 test specimens. action within the longitudinal composites in the vicinity of
Instrumentation
Applied load, vertical displacements, and strains were
monitored throughout the tests. Vertical loads were applied
monotonically to the beams. Displacements were measured
at midspan, at the locations of the applied loads, and at the
centerlines of the supports. The displacements measured at
the supports were caused by deformations within the elastomeric
pads. Therefore, these deformations were subtracted from
the displacements measured along the span to determine the
relative deflection of the test specimens.
All of the beams were fabricated with a crack initiator: a
0.015 in. (0.4 mm) thick piece of sheet metal that was placed
in the forms before casting the concrete. The sheet metal
extended the entire width of the beam and was approximately
0.25 in. (6 mm) deep. The crack initiator was positioned 42 in.
(1070 mm) from the centerline of one of the supports for the Fig. 3—Layouts of CFRP materials.
smaller beams and 50 in. (1270 mm) from the centerline of
one support for the larger beams. Strain gages were attached to
the reinforcing bars, the surface of the concrete, and the sur-
face of the CFRP materials at the location of the crack ini-
tiator (Fig. 4). The crack initiator was located under one of
the load points for the smaller beams and was located within
the constant moment region for the larger beams. The crack
initiator served three purposes: it fixed the location of the
first flexural crack in each beam, it ensured that the steel
strains were measured at the location of a crack, and for cases
where the bonded length of the CFRP materials was not
symmetric at the two ends of the beam, the shorter bonded
length was always positioned near the crack initiator. Fig. 4—Locations of strain gages at instrumented section.
Fig. 6—Measured response of beams strengthened using Fig. 8—Measured response of beams strengthened using
Composite System B. Composite System D.
Strain gages were attached to the top surface of both of the of the longitudinal reinforcement, and the capacity of each of
bottom longitudinal bars in all beams and to the top surface the beams are summarized in Table 3 and 4. The load and
of one of the top longitudinal bars in all specimens except displacement levels corresponding to cracking and yielding
B4, B5, D4, and D5. Strain gages were attached to both sides were estimated from changes in the slope of the measured
of the concrete at middepth of the top layer of reinforcement. load-deflection curves.
For beams strengthened using Configurations I and II, a The response of the strengthened specimens was essentially
strain gage was attached to the bottom surface of the CFRP the same as the response of the control specimens before the
material. One strain gage was attached to each side at mid- concrete cracked. After cracking, the strengthened specimens
depth of the CFRP material for beams strengthened using tended to be stiffer than the control specimens. The load at
Configurations III and IV. yielding of the longitudinal reinforcement increased between
5 and 10% for specimens strengthened using Composite
MEASURED RESPONSE OF BEAMS Systems A and B, between 15 and 25% for specimens
Representative load-deformation and load-strain data are strengthened using Composite System C, and between 20
discussed in this section. Complete information about the and 40% for specimens strengthened using Composite
response of the test specimens is presented in the report by System D. Numerous flexural cracks formed in beams during
Breña et al. (2001). this phase of the testing, but the CFRP composites remained
bonded to the surface of the concrete until the longitudinal
Load-deformation response reinforcement yielded.
The strength and deformation capacity of the test specimens After yielding, the strengthened specimens continued to
varied significantly depending on the CFRP system applied be stronger and stiffer than the control beams. Although
and the layout of the composites. The measured response of most of the strengthened beams failed at loads that exceeded
the specimens strengthened with each of the CFRP materials the capacity of the control beams, all the strengthened
is shown in Fig. 5 through 8. Each of these figures compares specimens failed at displacement levels that were considerably
the response of the strengthened specimens with the highest less than the capacity of the control beams. Most of the
and lowest capacity for a given CFRP material with the strengthened beams failed when the longitudinal composites
response of the corresponding control specimen. In order to debonded from the surface of the concrete. This was a sudden
facilitate comparisons, the deformation data at midspan have mode of failure, and there were few visual indicators that
been normalized by the yield displacement of the companion debonding was imminent.
control specimen and the load data have been normalized by Secondary cracks formed in the vicinity of the flexural
the yield load of the control specimen. The loads and displace- cracks after yielding of the longitudinal reinforcement, and
ment levels corresponding to cracking of the concrete, yielding these cracks tended to propagate along the sides of the CFRP
(a) (b)
Fig. 9—Photographs of typical cracks.
A s f s + A f ffe
a = -------------------------
- (2) Fig. 17—Determination of strength of control specimen at
0.85f c′ b displacement level corresponding to flexural capacity of
strengthened specimen.
where Mtest = flexural strength of beam strengthened using
CFRP composites; Msteel = contribution of reinforcement to reinforced concrete section at the displacement level equal to
flexural strength of the beam; MCFRP = contribution of the measured displacement at the flexural capacity of each of
composites to flexural strength of the beam; a = depth of the strengthened specimens (Fig. 17). The applied moment
equivalent stress block; d = effective depth of the steel reinforce- was calculated from the measured load using statics. The
ment; dCFRP = effective depth of the composite laminate; strength of the control specimen at the displacement corre-
As = area of reinforcement in tension; fs = stress in the reinforce- sponding to the peak load was assumed to be equal to the
ment at the flexural capacity of the beam; Af = area of composite internal moment contribution of the reinforcement in the
laminate in tension; and ffe = effective stress in the composite strengthened section (Msteel), which included the effects of
at the flexural capacity of the beam. steel strain hardening. The contribution of the composite
Several assumptions implicit in Eq. (1) and (2) should be laminates to the total internal moment of the strengthened
discussed. First, the Whitney stress block was used to calculate section (MCFRP) was obtained by subtracting the contribution
the compression force in the beams although the maximum of the reinforcement from the measured moment of the
measured strain in the concrete at failure was lower than the strengthened section. Equation (1) and (2) were then solved
maximum attainable strain εcu = 0.003 that is commonly simultaneously to obtain the effective stress ffe in the composites
used for design. This assumption leads to a slightly larger at failure of the beams. The results of these calculations are
moment arm between the compressive and tensile force listed in Table 6. The ratio of the effective composite stress
components than the actual moment arm at CFRP debonding, to the rupture stress gives an indication of the efficiency of
but the flexural strength of the beams is not affected signifi- the composite system for the different strengthening
cantly. Second, the contribution of the composites to the schemes. The higher stress ratios correspond to specimens
flexural strength was obtained by using an effective stress ffe where transverse straps or side placement controlled premature
instead of using the rupture strength of the composite ffu. As debonding of the composites. Additionally, the stress ratios
discussed previously, the composites in the specimens tested are consistently lower than the strain ratio values reported in
in this research project often debonded from the surface of Table 5. The magnitude of the measured strains depends on
the concrete before developing the tensile strength of the the location of the strain gages relative to the cracks in the
composites. Consequently, the tensile force in the composite beams. The strain in the CFRP composite at a crack is expected
at failure of the specimens was lower than the tensile to be higher than the strain measured between cracks. Using
strength of the laminates. Therefore, the effective stress in the measured composite strains to compute the effective
the composite laminates needs to be estimated in order to stresses in the composites can result in artificially high values
calculate the flexural capacity of strengthened sections. of stress resulting in high flexural strength values.
The effective composite stresses were determined from More tests are required to develop design provisions that
the test results by first determining the strength of the bare include effective composite stresses as a design parameter