Professional Documents
Culture Documents
(Asce) 1090-0268 (2006) 10 3
(Asce) 1090-0268 (2006) 10 3
Abstract: This paper deals with an innovative technique for strengthening reinforced concrete 共RC兲 structures using steel-reinforced
polymer 共SRP兲 materials. The results of an experimental campaign using RC beams strengthened in flexure with carbon fiber-reinforced
polymer or SRP laminates are summarized, and the experimental outcomes are compared to the predictions provided by analytical models
and code formulations in terms of flexural strength, curvature of the cross section, deflections, and crack widths. Under ultimate
conditions, the ACI 440.2R-02 approach provided conservative flexural strength, and a modified expression for the bond coefficient km
was proposed. Under serviceability conditions, good agreement was obtained between experimental results and a theoretical model
developed by the writers. Comparisons of code models in terms of both crack width and deflections highlighted the need for a calibration
of code formulas to account for effects due to externally bonded reinforcement.
DOI: 10.1061/共ASCE兲1090-0268共2006兲10:3共195兲
CE Database subject headings: Concrete beams; Concrete, reinforced; Curvature; Deflection; Stiffening; Cracking; Ultimate
strength.
B-3a Steel 12⫻ C 200 1 equal 1. The calculations considered the above-mentioned
B-4a Steel 12⫻ C 200 2 material properties.
C-1 CFRP E 300/500 2 To determine the failure mode according to the guidelines
C-2 CFRP E 300 3 共ACI 2002兲, one should check whether the effective strain of
Note: E⫽epoxy and C⫽cementitious. the external reinforcement, fe, is no larger than the product of
a the ultimate reinforcement strain, fu, times the bond-dependent
With nails.
coefficient, km. If this condition is verified, the failure mode is
concrete crushing, and the effective strain in the composite rein-
parameters should be recalibrated to improve the reliability of forcement is obtained from the linearity of the strain diagram. If it
code formulations 共Eurocode 2 1992; ACI 1995兲. is not verified, the failure is due to FRP/SRP rupture, and the
effective strain is computed as
fe = km fu 共1兲
Experimental Program
The document ACI 共2002兲 provides two expressions for
The experimental program was aimed at evaluating the effect computing km, depending on the value of the product n f · E f · t f
of various types of external flexural reinforcements on RC 共n f is the number of plies and E f and t f are the Young’s modulus
beams. A total of 10 RC shallow beams with a cross section of and nominal thickness of one ply of the external reinforcement,
400⫻ 200 mm were tested. The stirrups were ⌽8 mm steel respectively兲. For beams discussed in the present paper, this
bars spaced at 100 mm center to center. For all specimens, five product is always less than the threshold of 180 MPa, and the
⌽10 and two ⌽8 steel bars were used as tensile and compressive following equation should be used:
internal reinforcement, respectively. All beams were tested as
simply supported members over a clear span of 3.40 m and
loaded to failure under a four-point configuration, with a constant-
km =
1
冉
60 fu
1−
nE f t f
360,000
冊
艋 0.90 共2兲
moment region of 1.0 m across the midspan. One of these beams Note that by means of km, the maximum usable strain in
was not strengthened 共specimen D兲, while the other nine were the external reinforcement is reduced by ACI 共2002兲, and thus
externally reinforced using steel tapes 共Beam Types A and B兲 and SRP/FRP rupture represents a conventional way to also account
carbon fiber-reinforced polymer 共CFRP兲 laminates 共Type C兲 for premature failures.
共Prota et al. 2004兲. Type A and B beams were strengthened using For all tested beams, the ultimate neutral axis position can be
3 ⫻ 2 and 12⫻ cord SRP laminates, respectively. The former is a computed by a trial-and-error procedure from the equilibrium
medium-density laminate, which the latter is characterized by a equation in which the summation of horizontal compressive
low density of steel fibers. 共i.e., concrete and compressive steel reinforcement兲 and tensile
Seven beams were strengthened with steel tapes impregnated 共i.e., tensile steel and FRP reinforcement兲 forces is equal to zero.
with epoxy resin or cementitious grout 共A and B beams兲; the By means of the equilibrium condition, the nominal flexural
remaining two beams 共C-1 and C-2兲 were strengthened with moment M n, strains in the concrete, steel reinforcement, and FRP
CFRP laminates impregnated with epoxy resin. Two of the can be determined. For Beams A-2, B-4, C-1, and C-2, a failure
beams 共B-3 and B-4兲 were strengthened with steel tape using due to concrete crushing was predicted, while for all the other five
cementitious grout and mechanically anchored with nail anchors. beams, Eq. 共1兲 governed and the expected failure mode involved
Table 1 shows the research program, summarizing for each beam FRP/SRP rupture.
the type, matrix, bond width, b f , and number of plies of the The results of such theoretical analysis are reported in Table 2,
externally bonded reinforcement, n f . where a comparison with experimental outcomes is also pro-
The experimental concrete mean cube strength was 29.7 MPa. posed. Tests showed that the strength increases provided by SRP
Based on experimental tests, the yield stress, ultimate stress, and bonded with cementitious grout were smaller than those obtained
ultimate strain for ⌽10 bars were 500 MPa, 600 MPa, and 0.12, using epoxy. CFRP was more effective than epoxy-bonded SRP in
respectively. The CFRP laminates had a unidirectional fiber tex- terms of strength, and the epoxy resin was more effective than the
ture with a density of 300 g / m2 and an equivalent fiber thickness cementitious grout in engaging the concrete substrate. It was also
of 0.167 mm 共Mapei 2000兲. The Young’s modulus was 230 GPa, found that the nail anchors did not affect the performance of the
and the ultimate tensile strain was 0.015. The experimental char- SRP impregnated with cementitious grout because the lack of a
acterization of the SRP showed that the material can be consid- transverse link in the steel tape did not allow distributing the local
ered linear elastic up to failure, with no yielding of the steel; tests stress concentration at the anchor location. A detailed description
provided mean ultimate tensile stress of 3,070 MPa, ultimate of experimental results can be found in Prota et al. 共2004兲.
strain of 0.017, and a Young’s modulus of 184 GPa. The compres- As far as the failure mode is concerned, the theoretical predic-
sive and splitting tensile strengths of the cementitious grout, at tion 共concrete crushing兲 was not confirmed by experiments for
28 days of curing, were equal to 41.4 and 5.2 MPa, respectively. one beam of both Series A and B 共A-2 and B-4兲 and for the
B-1 0.0126 0.0108 0.0108 SR 80.4 48.22 SRP delamination 88.6 53.16
B-2 0.0126 0.0108 0.0108 SR 80.4 48.22 SRP delamination 72.7 43.62
B-3 0.0126 0.0108 0.0108 SR 80.4 48.22 SRP delamination 71.5 42.90
B-4 0.0095 0.0108 0.0095 CC 105.7 63.44 SRP delamination 86.7 52.02
C-1 0.0105 0.0131 0.0105 CC 96.1 57.65 FRP delamination 96.5 57.90
C-2 0.0080 0.0114 0.0080 CC 121.0 72.58 FRP delamination 134.8 80.88
Note: CC= concrete crushing and SR= SRP failure.
two beams of Series C 共Table 2兲. The same amount of external grout 共B-2, B-3, and B-4兲. To calculate effective strains in the
reinforcement arranged in different configurations determines that SRP that provide theoretical moments equal to those experimen-
the ACI formulas are able to capture the failure mode of A-3 共two tally measured for the three beams, values of about 0.58 共B-2 and
plies and width equal to 37.5% that of the beam兲, whereas they do B-3兲 and 0.47 共B-4兲 of fu are obtained. Based on these consid-
not perform in the same way if the SRP is arranged in one ply, but erations, the km equation could be safely extended to external
on a width equal to 75% of that of the beam 共A-2兲. Since the reinforcements bonded with cementitious grout as follows:
冉 冊
strain that the FRP would attain at concrete crushing by linearity
共second column of Table 2兲 is almost equal for both 共0.090 versus 1 nfEftf
0.092兲, the difference is due to values of km equal to 0.90 共which km = 1− 共3兲
120 fu 360,000
would be 1.06, but the limit controls兲 and 0.71 for A-2 and A-3,
respectively. The strength prediction that would be obtained for the three
For Beams B-1, B-2, and B-3, the km value would be 1.17, beams according to Eq. 共3兲 are summarized in Table 3; recall that
whereas it would be 0.95 for B-4; therefore, the limit of 0.90 Beams B-2 and B-3 are identical from a strength point of view, as
controls them all. This allows for a good prediction of the failure the presence of nails is not accounted for in the adopted formulas.
mode on B-1, B-2, and B-3 beams, while the effective strain Note also that the use of Eq. 共3兲 so much reduces the effective
in the SRP reinforcement is overestimated when two plies are strain in the SRP of the B-4 beam that the strain in its tensile steel
installed and concrete crushing is expected 共B-4兲. The same hap- reinforcement at ultimate is lower than 0.005 共0.0048兲. This
pens for the two CFRP-strengthened beams. Finally, for the nine would have an important repercussion from a design standpoint
strengthened beams, the analysis according to ACI 共2002兲 would because it would determine the use of a strength-reduction factor
provide ultimate strains in the tensile steel reinforcement, s, lower than 0.90.
much larger than 0.005 共ranging between 0.0061 of C-2 and
0.0084 of B-1兲 and then all largely fall into the region where the
strength reduction factor, , is equal to 0.90. For this reason, Moment–Curvature Relationships
no changes to the formulas appear necessary, considering that a
correction aiming at fitting all experimental failure modes would The behavior of tested beams was also analyzed in terms of
not affect the overall safety of the strengthening design. moment–curvature relationships. For the control beam, the
Regarding the nominal flexural moments, the ACI approach experimental mean curvature was determined from readings of
is conservative for both A-1 and A-2 beams and overestimates linear variable differential transducers 共LVDTs兲 mounted over a
the strength of A-3 by 7.4% 共Table 2兲. With respect to Series B, gauge length of 350 mm across midspan at the heights of both
it is conservative for the beam impregnated with epoxy 共B-1兲, compressive concrete and internal steel reinforcement 关Fig. 1共a兲兴.
whereas it overestimates the strength of those impregnated For the strengthened beams, the experimental mean curvature was
with cementitious grout by about 11.4% 共B-2 and B-3兲 and 22% obtained from readings of the LVDT transducer mounted on the
共B-4兲; it is always conservative for the CFRP-strengthened beams compressive concrete and of strain gauges installed at midspan on
共Table 2兲. the external reinforcement 关Fig. 1共b兲兴.
The main conclusion of such comparison is that the approach
of ACI 共2002兲 overestimates by more than 10% the strength of
the beams whose SRP reinforcement is bonded with cementitious
Fig. 8. Comparison between experimental and theoretical values of crack width: 共a兲 comparison between experimental and model results;
共b兲 comparison between experimental and code results
Deflections
the model and experimental results for the unstrengthened Eurocode 2 共1992兲 formulates the deflection by combining the
Beam D and Beams A2 and C2 with the same width 共300 mm兲 deflection calculated for the gross, f 1, and the cracked section, f 2,
and adhesive 共epoxy兲 and similar equivalent reinforcement through a tension-stiffening coefficient, , depending on loading
percentage 共0.74 for Beam A2 and 0.79 for Beam C2兲. The type, bond properties, and ratio of cracking to service bending
experimental values of crack width are given by a mechanical moment according to the following expression:
deformometer and have not been registered until the service load
for all beams. f = 共1 − 兲 · f 1 + · f 2 共5兲
Figs. 8共a兲 compare the experimental values of the crack width,
wexp, at the service loads with those given by the model, while On the other hand, the ACI approach is based on a combination of
Fig. 8共b兲 compares the values, wcode, given by Eurocode 2 共1992兲 gross, Ig, and cracked, Icr, inertias using the ratio of cracking to
and ACI 共1995兲 to show the reliability of different theoretical service bending moment, according to the following expression:
冉 冊 冋 冉 冊册
approaches. Fig. 8共a兲 confirms the effectiveness of the model. In
3 3
the EC2 formulation the crack spacing was calculated according M cr M cr
to the expression suggested by the code. The writers noticed that Ie = · Ig + 1 − · Icr 共6兲
M M
by introducing the experimental crack spacing at the service load,
a better agreement could be obtained 共Ceroni et al. 2004b兲. This value of equivalent inertia is used to calculate the deflection
ACI 共1995兲 estimates the effective concrete area in tension according to the load pattern. Both formulations take FRP into
surrounding each bar to be equal to the value of Act,eff used for account in the calculation of inertia of the cracked section Icr.
formulation of Eurocode 2 共1992兲 divided by the number of bars. In both code formulations the effect of the beam weight was
Both code formulations generally overestimate the experimental added to the applied load. Fig. 9 compares the experimental
values, except for two beams 共B2 and B3兲, which resulted in a deflections at the service load and the values given by EC2 and
crack width more deformable than the Reference Beam D. The ACI formulations, showing that the ACI provisions underestimate
code formulations for strengthened beams with epoxy resin show the experimental values in all cases. On the other hand, EC2
an approximation with respect to the experimental values gener- mainly shows a better agreement. Note that in some cases the
ally no higher than for the unstrengthened beam. experimental deflections are close to deflection f 2, corresponding
to the condition of cracked inertia; beams strengthened with
cementitious mortar in some cases show deflections larger than f 2.
Fig. 10. Comparison between experimental and code values of Fig. 12. Comparison between experimental and code values of mean
deflections for Beam A1 curvature for Beam A1
c ⫽ concrete cover;
where the tension-stiffening coefficient is the same used for Ec ⫽ Young’s modulus of concrete;
deflection. E f ⫽ Young’s modulus of FRP;
Fig. 12 compares the code value of mean curvature, code, Es ⫽ Young’s modulus of steel bar;
using Eq. 共7兲 with the experimental value, exp, given by LVDT F ⫽ vertical applied load;
measures: the theoretical values show a good agreement with the Fserv ⫽ service load;
experimental ones. Fu,th ⫽ theoretical ultimate load;
Fu,exp ⫽ experimental ultimate load;
f ⫽ deflection;
Conclusions f code ⫽ deflection calculated by codes;
f exp ⫽ experimental deflection;
The results of a theoretical-experimental analysis are presented f fu ⫽ tensile strength of FRP;
for RC beams strengthened in flexure with FRP and SRP lami- f ref ⫽ deflection;
nates. Regarding the ultimate behavior, the comparison allows f 1 ⫽ deflection at condition of gross inertia;
assessing the strength gains provided by FRP and SRP systems. f 2 ⫽ deflection at condition of cracked inertia;
For the latter, the performance of epoxy resin and cementitious Ig ⫽ gross inertia;
mortar are compared. Icr ⫽ cracked inertia;
The paper confirms that the approach of ACI 共2002兲 provides H ⫽ height of cross section;
conservative strength estimates for both FRP and SRP systems, if km ⫽ bond-dependent coefficient;
the external reinforcement is bonded with epoxy. The different M ⫽ bending moment;
bond behavior of a cementitious mortar is not predicted by the M cr ⫽ cracking bending moment;
current equations given by ACI 共2002兲; a modified expression for M n ⫽ nominal bending moment;
the bond coefficient km has been proposed herein to extend these M u,exp ⫽ experimental ultimate bending moment;
equations to systems impregnated with mortar. Further tests will n ⫽ number of cracks;
need to be performed to confirm the reliability of the modified n f ⫽ number of FRP plies;
equation. sfrp ⫽ slip at concrete-laminate interfaces;
Regarding the behavior at service conditions, the following smax ⫽ slip at concrete-laminate interface corresponding to
conclusions are based on the experimental results: shear stress equal to zero;
• At similar serviceability load levels, crack width becomes srm ⫽ mean crack spacing;
smaller if the width of the laminate increases, while the num- srm,exp ⫽ mean experimental crack spacing;
ber of fiber layers is less important; s1 ⫽ slip at concrete-laminate interface corresponding to
• Steel cords and carbon fibers, both impregnated with epoxy, maximum shear stress;
gave very similar results when the reinforcement percentage t f ⫽ thickness of FRP sheet;
was the same; wexp ⫽ experimental crack width;
• Low-density steel cords bonded with cementitious grout gave wm ⫽ mean crack width;
a low tension-stiffening effect; wmod ⫽ mean crack width;
• Comparisons between code formulations and experimental wcode ⫽ mean crack width;
values of crack width and deflections for beams strengthened xc ⫽ neutral axis;
with epoxy resin show a scatter generally comparable to that fe ⫽ effective FRP strain;
observed for unstrengthened beam. When the cementitious fu ⫽ ultimate FRP strain;
mortar is used, experimental values are much larger than code sm ⫽ mean strain of steel bar;
values; and 2 ⫽ steel strain at cracked section;
• Beams strengthened with cementitious mortar showed a less ⫽ tension-stiffening coefficient;
stiff behavior in serviceability conditions. ⫽ ratio between experimental ultimate load over
service load;
⫽ ratio between internal reinforcement over Act,eff
s ⫽ percentage of internal reinforcement;
Acknowledgments eq ⫽ equivalent reinforcement percentage;
frp ⫽ percentage of external reinforcement;
The writers would like to thank Hardwire LLC, Pocomoke max ⫽ maximum bond stress of FRP at concrete interface;
City, Maryland; Mapei Spa, Milan, Italy; and Sika Italia, Milan, ⫽ steel bar diameter;
Italy, for donating the steel tapes, FRP system, and epoxy resin/ ⫽ mean curvature;
cementitious grout used to bond SRP, respectively. exp ⫽ experimental curvature;
for structural concrete.” ACI 318, ACI, Farmington Mills, Mich. A. Nanni, eds., ASCE, Reston, Va., 269–277.
American Concrete Institute 共ACI兲. 共2002兲. “Guide for the design and Pecce, M., and Ceroni, F. 共2004兲. “Modeling of tension stiffening
construction of externally bonded FRP systems for strengthening con- behavior of reinforced concrete ties strengthened with fiber-reinforced
crete structures.” ACI 440, ACI, Farmington Hills, Mich. plastic sheets.” J. Compos. Constr., 8共6兲, 510–518.
Bakis, C. E., et al. 共2002兲. “Fiber-reinforced polymer composites Prota, A., Tan, K., Nanni, A., Pecce, M., and Manfredi, G. 共2004兲.
for construction—State-of-the-art review.” J. Compos. Constr., 6共2兲, “Performance of RC shallow beams externally bonded with steel
73–87. reinforced polymer.” ACI Struct. J. 共in press兲.
Ceroni, F., Pecce, M., and Matthys, S. 共2004a兲. “Tension stiffening of RC Sato, Y., Ueda, T., and Shoji, K. 共2002兲. “Tension stiffening effect of
ties strengthened with externally bonded FRP sheets.” J. Compos. reinforced concrete member strengthened by carbon fiber sheet.”
Constr., 8共1兲, 22–32. Proc., Int. Symp. on Bond in Concrete, 606–613.
Ceroni, F., Pecce, M., Prota, A., and Manfredi, G. 共2004b兲. “Flexural Smith, S. T., and Teng, J. G. 共2002a兲. “FRP-strengthened RC beams.
strengthening of RC beams using emerging materials: Cracking I: Review of debonding strength models.” Eng. Struct., 24共4兲,
behavior.” Proc. Int. Conf. CICE 2004, Taylor & Francis, London, 385–395.
171–178. Smith, S. T., and Teng, J. G. 共2002b兲. “FRP-strengthened RC beams.
Comité Euro-International du Béton 共CEB兲. 共1985兲. “Manual on cracking II: Assessment of debonding strength models.” Eng. Struct., 24共4兲,
deformation.” Bulletin d’Information No. 158-E, CEB, Paris. 397–417.
Cosenza, E., Manfredi, G., and Nanni, A., eds. 共2001兲. Composites in Teng, J. G., Chen, J. F., Smith, S. T., and Lam, L. 共2001兲. FRP strength-
construction: A reality, ASCE, Reston, Va. ened RC structures, Wiley, Chichester, U.K.
Eligehausen, R., Popov, E. P., and Bertero, V. V. 共1983兲. “Local bond Ueda, T., Yamaguchi, R., Shoji, K., and Sato, Y. 共2002兲. “Study on
stress-slip relationships of deformed bars under generalized behavior in tension of reinforced concrete members strengthened by
excitations.” Rep. No. 83/23, Environmental Engineering Research carbon fiber sheet.” J. Compos. Constr., 6共3兲, 168–174.
Council, Univ. of California at Berkeley, Berkeley, Calif. Zhang, Y., Toutanji, H., and Balagrou, P. 共2003兲. “Crack widths in
Eurocode 2. 共1992兲. “Design of concrete structures. 1-1: General rules RC beams externally bonded with CFRP sheets.” Proc., Int. Conf.
and rules for buildings.” ENV 1992-1-1. FRPRCS-6.