Composite Structures 88 (2009) 497508

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Flexural behaviour of reinforced concrete beams strengthened with prestressed carbon composites
Dong-Suk Yang a, Sun-Kyu Park b, Kenneth W. Neale a,*
a b

Department of Civil Engineering, University of Sherbrooke, 2500 boul. de IUniversit, Sherbrooke, Quebec, Canada, 1K 2R1 J Department of Civil Engineering, Sungkyunkwan University, 300 Chunchun-Dong, Jangan-Gu, Suwon, Gyeonggi-Do 440-746, South Korea

a r t i c l e

i n f o

a b s t r a c t
In this study, a total of 13 FRP-strengthened reinforced concrete beams were tested in exure and analyzed using the nite element method. The various variables included bonding or no bonding of the FRP, the anchorage system, the amount of prestressing, and the span length. The experiments consisted of one control beam, two non-prestressed FRP-bonded beams, four prestressed FRP-unbonded beams, four prestressed FRP-bonded beams, and two prestressed FRP-unbonded beams with different span lengths. All the beams were subjected to three-point and four-point bending tests under deection control, with the loading, deection and failure modes recorded to the point of failure. A nonlinear nite element analysis of the tested beams was also performed using the DIANA software; this analysis accounted for the nonlinear concrete material behaviour, the reinforcement, and an interfacial bond-slip model between the concrete and CFRP plates. The aim of this investigation was to study the exural performance of reinforced concrete members strengthened using CFRP plates, employing different FRP bonding and prestressing methods. The failure mode of the prestressed CFRP-plated beams was not debonding, but FRP rupture. For the reinforced concrete members strengthened with externally bonded prestressed CFRP plates, debonding of the composite laminates occurred in two stages. After the debonding of the CFRP plates that occurred in the bonded cases, the behaviour of the bonded CFRP-plated beams changed to that of the unbonded CFRP-plated beams due to the effect of the anchorage system. The exural test results and analytical predictions for the CFRP-strengthened beams were compared and showed very good agreement in terms of the debonding load, yield load, and ultimate load. The ductility of the beams strengthened with CFRP plates having the anchorage system was considered high if the ductility index was above 3. 2008 Elsevier Ltd. All rights reserved.

Article history: Available online 7 July 2008 Keywords: Anchorage system CFRP plates Ductility, Finite element analysis Flexural tests Prestressing

1. Introduction The bonding of steel plates for the strengthening and rehabilitation of reinforced concrete structures was a popular strengthening method in the past. In recent years, there has been extensive research on the use of bre reinforced polymer (FRP) composites for the replacement of steel plates in plate bonding. FRPs have also been widely used as external wrapping for column strengthening and prestressed bonding for reinforced concrete members. Indeed, there is considerable interest worldwide in the eld of FRPs for civil engineering infrastructure and much research has been carried out on a variety of important topics [118]. The widely used strengthening method of simply bonding FRPs, such as carbon bre, aramid and glass bre, for strengthening a structure generally produces a debonding failure prior to the attainment of the tensile strength of the FRP being used [6,11,
* Corresponding author. Tel.: +1 819 821 7752; fax: +1 819 821 7974. E-mail address: Kenneth.Neale@USherbrooke.ca (K.W. Neale). 0263-8223/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.compstruct.2008.05.016

13]. In fact, the debonding strain of the composite differs according to the material used, and this debonding occurs at about 3050% of the FRP tensile strength [8]. A reinforced concrete member strengthened with prestressed FRPs is a combination of the FRP bonding and external prestressing methods. This leads to a more efcient use of the FRP; the method can be used to compensate for prestressing losses of existing reinforced concrete or prestressed concrete members, to enhance the strengthening performance, and to improve the serviceability of cracked or deected members. The focus of this paper is to study the exural performance of reinforced concrete members strengthened by CFRP plates, using FRP bonding and prestressing methods. Flexural tests were performed with respect to the bonding method, the anchorage system, the amount of prestressing, and the span length as experimental variables. A total of 13 beams, including a control beam, were subjected to exural tests. A nonlinear nite element analysis of the beams was also performed for each exural test using the DIANA software package, incorporating the plasticity of concrete and

498

D.-S. Yang et al. / Composite Structures 88 (2009) 497508 Table 1 Details of the tested beams Beams Control NFCB1 NFCBW2 PFCU1-0R PFCU1-2R PFCU1-4R PFCU1-6R PFCB1-0R PFCB1-2R PFCB1-4R PFCB1-6R PFCU1-4L PFCU1-6L Variables of experiment No anchorage system Prestressing No bond Content Not strengthened 1 laminate 2 laminate of width direction Prestressing 0% + anchorage Prestressing 20% + anchorage Prestressing 40% + anchorage Prestressing 60% + anchorage Bond Prestressing 0% + anchorage Prestressing 20% + anchorage Prestressing 40% + anchorage Prestressing 60% + anchorage Unbond + prestressing 60% + anchorage Unbond + prestressing 60% + anchorage Span length 240 cm

interface elements between the FRP and concrete. The analytical results for the CFRP-strengthened reinforced concrete are compared to the exural test results. The strengthening effect, failure mode, and load-deection behaviour were considered for each experiment, and the ductility was evaluated as the ratio of the ultimate load for each tested beam to the deection at yielding. 2. Experimental program for reinforced concrete beams strengthened with CFRP plates 2.1. Variables of the experiments and beams In this investigation, exural tests were performed with the experimental variables being bonding or no bonding of the FRPs, the anchorage system, the amount of prestressing, and the span length (240 cm, 450 cm and 600 cm). Flexural tests were conducted on a control beam without strengthening, on beams strengthened with bonded CFRP plates with one and two laminates in the width direction, and on beams strengthened with CFRP plates prestressed with 0%, 20%, 40%, and 60% of the ultimate tensile strength of the CFRP plates. A total of 13 beams were subjected to exural tests. Rectangular normal-weight concrete beams were cast with dimensions of 200 mm (b) 300 mm (h). The characteristics of all the beams, as well as their steel reinforcement details, are shown in Fig. 1 and Table 1. 2.2. Materials The type of concrete used in the exural tests was a readymixed concrete, which had been aged for 28 days, with a specied

L/hf

450 cm 600 cm

concrete strength of 18.0 MPa, measured compressive strength of 16.4 MPa, and slump of 120 mm. All the beams were reinforced with three D13 bars (diameter of 13 mm) and three D10 bars (diameter of 10 mm) in the compression and tension zone, respectively. The beams were provided with 10 mm diameter shear reinforcements, with a 100 mm spacing; they were all designed to prevent shear failure. The design yield stress of the 10 mm and 13 mm diameter reinforcing bars were 475.2 MPa and 466.2 MPa, respectively. The composite material used in this test program consisted of a three-layer component with a bi-directional CFRP sandwiched between two unidirectional CFRP plates. This was done during the general moulding process so as to prevent cracking in the unidirectional layers due to the prestressing load. The material properties of the CFRP used in the exural tests are given in Table 2.

2,700
HD10@100

150

2,400 3-HD13

150
3-HD13 300 3-HD10

CFRP Plate 145 1,900

3-HD10

HD10@100

200

(a) Beams of span length 240cm

4, 800 150 1, 025 2, 450 1, 025 3- HD 13 150

CFRP Plate 4, 000

3- HD 10

HD 10 @ 100

(b) Beams of span length 450cm

6,300 150 1, 500 3, 000 1, 500 3-HD13 150

CFRP Plate 5,500

3-HD10

HD10@100

(c) Beams of span length 600cm

Fig. 1. Beam details (units: mm).

D.-S. Yang et al. / Composite Structures 88 (2009) 497508 Table 2 Material properties of the CFRP plates Tensile strength (MPa) CFRP Plates Epoxy resin 2350 33.5 Modulus of elasticity (MPa) 1.73 105 15003500 Remarks Width 5 cm thickness 0.13 cm Bond strength 4.3 MPa

499

system is necessary for xing the prestressed CFRP plates. The rough surface of the anchorage system is processed, and then xed to the beam with anchor bolts. In order to prevent a load concentration, the anchorage system of the CFRP plate was attached to a GFRP tab. The device for prestressing the beams with the CFRP plates is shown in Fig. 3. 2.4. Loading equipment and arrangement of strain gauges During loading, a strain gauge was attached to the CFRP plates to measure the strain. The exural members strengthened with prestressed CFRP plates were tested under three and four-point loading, over spans of 240 cm, 450 cm and 600 cm. As shown in Fig. 4, the strain gauges were attached to the bottom edge and the mid-depth point of the CFRP plates at the middle

2.3. Anchorage system and prestressing The reinforced concrete beams were provided with prestressed CFRP plates anchored to the tension face in order to obtain the required strengthening capacity. As shown in Fig. 2, an anchorage

End section Concrete

Center section

M12 anchor bolt Base Plate CFRP GFRP Tab Rough surface

Bearing Plate M 16 bolt

Fig. 2. Details of the anchorage system.

Fig. 3. Prestressing device.

Fig. 4. Location of CFRP plate strain gauges.

500

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

of each shear span, as well as at the middle of the beam span in the longitudinal direction. To obtain an accurate deection reading, a Linear Variable Differential Transducer (LVDT) was also mounted at the mid-span and connected to a data logger. Crack initiation and propagation were monitored by visual inspection during testing, and the crack patterns were marked. 3. Nonlinear nite element analysis 3.1. General The numerical analysis of the experimentally tested FRP-plated concrete beams was conducted with a full 2D nonlinear nite element analysis of the experimental set-up using the commercial program DIANA. Fig. 5 shows the mesh geometry for the control and prestressed FRP-plated beams. The DruckerPrager criterion was used for the concrete in compression, and the combined smeared crack model for the tensile behaviour of the concrete. The general nite element model was as follows: the bond-slip between the concrete and steel rebars was not considered in this paper. The concrete and CFRP plates were modelled using plane stress elements, with smeared elements for the steel reinforcement, and interface elements between the concrete and bonded CFRP laminates. 3.2. Material models As shown in Fig. 6, the commonly used concrete analysis model consisted of the traditional plasticity model for the compression concrete, while the tensile behaviour consisted of a smeared crack model, which assumes micro-cracking. As shown in Fig. 7a, the reinforcing rebars had an elasto-plastic behaviour, dened by its yield strength, with a typical elastic modulus of 200 GPa. A yield plateau, whose range depends on the class
concrete model compression bar model
Fig. 6. Concrete model for the nite element analysis.

f2 ft

cr fnn

constant shear retention

constant tension cut - off

ft

f2

Multi linear tension softening

cr nn

of steel, is followed by a strain-hardening behaviour up to failure. In this model the reinforcing bars are modelled as steel layers of equivalent thickness. The CFRP plates have a very high unidirectional tensile strength, but with a stiffness close to that of steel. The behaviour was
Stirrup model

Tension bar model

(a) Reinforced concrete beams without CFRP plates

concrete model compression bar model Stirrup model

Bond -Slip model Anchorage system model

CFRP Plates model Tension bar model

(b) Reinforced concrete beams strengthened with CFRP plates

Fig. 5. Mesh geometry for the nite element analysis.

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

501

fs
fu fy

Eu Es

fu

(a) Reinforcement
y

(b) CFRP Plates

2' 3 1 1' 2

(c) Interface element (4-nodes)

Fig. 7. Reinforcement and CFRP plate models for the nite element analysis.

Table 3 Comparison of results for concrete beams strengthened with CFRP plates Beams Cracking load (kN) Yield load Test (a) (kN) Control NFCB1 NFCBW2 PFCU1-0R PFCU1-2R PFCU1-4R PFCU1-6R PFCB1-0R PFCB1-2R PFCB1-4R PFCB1-6R PFCU1-6L2 PFCU1-6L3 Mean C.O.V (%) 18.2 13.7 18.9 33.5 47.0 54.1 24.5 26.4 42.4 51.8 25.7 33.6 40.4 56.3 98.4 43.0 56.4 75.9 83.6 55.4 71.6 85.2 100.5 101.3 62.4 Anal. (b) (kN) 41.0 59.0 99.5 47.4 61.0 83.9 92.6 56.4 75.5 91.6 105.9 98.4 65.0 (b)/(a) 1.01 1.05 1.01 1.10 1.08 1.11 1.11 1.02 1.05 1.08 1.05 0.97 1.04 1.05 3.91 Ultimate load Test (a) (kN) 49.4 77.0 98.4 115.0 119.8 120.7 122.5 121.5 123.0 125.2 122.8 121.4 71.8 Anal. (b) (kN) 49.5 83.4 104.0 114.2 112.6 124.2 123.6 117.8 108.0 114.8 121.7 125.5 76.8 (b)/(a) 1.00 1.08 1.06 0.99 0.94 1.03 1.01 0.97 0.88 0.92 0.99 1.03 1.07 1.00 5.71 Flexural Debonding Delamination FRP rupture FRP rupture FRP rupture FRP rupture Debondingrupture Debondingrupture Debondingrupture Debondingrupture FRP rupture FRP rupture Failure mode

essentially linearly elastic up to the tensile strength limit. Once the tensile strength has been reached, it is assumed to suddenly fail in a perfectly brittle mode (Fig. 7b). The interface model between the CFRP plates and concrete used 4 node elements, in the same location as the perpendicular 2 nodes, as shown in Fig. 7c. As an interface model between CFRP plates and concrete, a bond-slip model was applied. 4. Results and discussion 4.1. Load-carrying capacities A total of 13 beams were subjected to exural tests, and later analyzed using the nite element approach described above. Bond-

ing or no bonding, the anchorage system, the amount of prestressing and the span length were the experimental variables. Table 3 summarizes the results of the exural behaviour of all the beams in terms of their exural loading capacities. From the exural tests and analytical results, the load capacity of the beam strengthened with one laminate (5 cm of width) was 4060% higher than that of the control beam, and the beam strengthened with two laminates in the width direction (10 cm in width) was 100% or greater than that of the control beam. In comparison to the control beam, as the stiffness of the beams strengthened with bonded CFRP plates increased, so did the cracking load of the beams increase. Whether bonded or not, the ultimate load of the beams strengthened with prestressed CFRPs remained almost constant.

502

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

It was seen that the failure mode of the prestressed CFRP plated-beams was not debonding, but FRP rupture. However, the cracking and yield loads of the bonded prestressed CFRP platedbeams were greater than those of the unbonded prestressed CFRP plated-beams, as a result of the increased stiffness due to the composite bonding. As shown in Table 3, the average ratio of the test and analytical values for the yield load was 1.05, with a coefcient of variation of 3.91%. As for the ultimate load, the average ratio was 1.00, with a coefcient of variation of 5.71%. Thus, the analytical results for the reinforced concrete beams strengthened with CFRP plates show a very good agreement with those obtained experimentally for the yield and ultimate loads. 4.2. Cracking modes for the beams Fig. 8 shows the cracking modes of the beams. The failure mode of the bonded non-prestressed beam (NFCB1) strengthened with only one laminate was induced by brittle intermediate crack-induced debonding from the mid- to end-span when the strain of the CFRP plates reached about 65007000 l. However, that of beam (NFCBW1) strengthened with two laminates in the width direction was a result of plate-end debonding when the CFRP plate strain reached about 5200 l. Therefore, it can be seen that the strengthening efciency of the member strengthened with one

laminate was better than that of the member strengthened with two or more laminates. The crack width of the control beam under ultimate load was also observed to be wider than that of the prestressed beams. The crack propagation of the prestressed beams with bonded CFRP plates was similar to that of the control beam. However, as the prestressing of the beams strengthened with CFRP plates increased, the crack width of the beams decreased. Also, the cracking mode of the prestressed beams with bonded CFRP plates was concentrated within the constant moment region. Thus, it can be seen that beams strengthened with prestressed CFRP plates had better performance in terms of crack control than those strengthened with non-prestressed CFRP plates. 4.3. Beams strengthened with bonded CFRP plates Fig. 9a shows the load-deection relationships for the control and strengthened beams (NFCB1, NFCBW2). With regard to the strengthening efciency, the beam strengthened with one laminate was 4060% higher than that of the control beam, and the beam (NFCBW1) strengthened with two laminates in the width direction was 100% or greater than that of the control beam. In comparing the test and analytical results, we observe that both gave almost identical results for the control beams, but the beams strengthened with CFRP plates (NFCB1, NFCBW2) displayed about a 510% greater stiffness than their analytical values. However, the analytical and test values for the deection were almost identical when the beams strengthened with one and two laminates in the width direction experienced debonding. Fig. 9b displays the CFRP plate strain of the NFCB1 beam obtained from the exural and analytical results. The initial strain of the CFRP plate, as determined by the analysis, was greater than the test value, but the stiffness after cracking load was almost identical. In addition, the strain of the CFRP plate rapidly increased after about 50 kN, the reinforcement yielding point, in both the test and analytical results. As mentioned above, the bonded non-prestressed beam (NFCB1) strengthened with only one laminate failed due to debonding when the strain of the CFRP plate reached about 6500 7000 l. Therefore, it can be seen that the prestressing amount of the CFRP plate was an extra strain excluding the debonding strain (=65007000 l), to maximize the strengthening efciency of the CFRP plate. 4.4. Beams strengthened with unbonded prestressed CFRP plates As the prestressing of the CFRPs increased, the load capacity of the beams strengthened with unbonded prestressed CFRP plates also increased. Also, because the FRP was not bonded, failure was induced by rupture of the CFRP plates. Even after the steel bars yielded, the load capacity of the beams continuously increased, and the ductility of the prestressed beams increased more than that of the non-prestressed beams. With regard to the cracking load, the values for the control beam and non-prestressed beam (PFCU1-0R) were similar, but those of the prestressed beams were a maximum of twice that of the control beam. The ultimate load of the prestressed beams was about 130150% greater than that of the control beam, but those of the prestressed beams were similar regardless of the amount of prestressing. Fig. 10a shows that the beam (PFCU1-0R) strengthened with unbonded non-prestressed CFRP plates, when under the initial load, had approximately 10% stiffness differences between the analytical and test values; however, there was no stiffness difference after the cracking load. With regard to specimens PFCU1-4R and PFCU1-6R, the analytical values for the stiffness were generally

Fig. 8. Beam cracking modes.

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

503

120
Control (test) Control (analysis) NFCB1 (analysis) NFCBW2 (analysis)

100

NFCB1 (test) NFCBW2 (test)

80

Load(kN)

60

40

20

0 0 10 20 30 40 50 60 70 80

Deflection(mm)

(a) Loaddeflection relations

90 80 70 60

Load(kN)

50 40 30 20
Test

10
Analysis

0 0 1000 2000 3000 4000 5000 6000 7000 8000

Microstrain

(b) LoadCFRP plate strain relations

Fig. 9. Comparison of results for reinforced concrete beams strengthened with bonded CFRP plates.

greater than the test values; for the PFCU1-2R beam the analytical values for the stiffness became smaller than the test value under the ultimate load. Fig. 10b displays the load-strain relationships for the unbonded prestressed CFRP plates (0%, 60%). The beams PFCU1-0R and PFCU1-6R showed approximately 10% stiffness differences between the analytical and test values during the initial stiffness, but after cracking, the analytical values approached those obtained in the tests. 4.5. Beams strengthened with bonded prestressed CFRP plates From the experimental observations, for the beams with bonded prestressed CFRP plates, debonding occurred in two stages up to the anchorages and then, due to the anchorage system, the CFRP plates ultimately ruptured. Also, due to external bonding of the CFRP plates, the beams with bonded prestressed CFRP plates maintained their stiffness from the initial to the debonding loads. After the two debonding phases, the load capacity of the beams temporarily decreased. However, the behaviour of the bonded CFRP-plated beams changed into that of the unbonded CFRP-plated beams due to the presence of the anchorage system. Thus, the CFRP plates of the bonded prestressed beams ruptured in a similar

fashion to the unbonded prestressed beams under the ultimate load. As shown in Table 3, with regard to the cracking load for the bonded non-prestressed beam (PFCB1-0R) with the anchorage system, it increased by about 35%, 30%, respectively, compared to the control beam and the unbonded non-prestressed beam (PFCU1-0R) with the anchorage system. The cracking load of the bonded prestressed CFRP-plated beams increased by a maximum of 180% compared to the control beam. The yield loads of beams with the anchorage system and bonded CFRP plates were 35150% higher than that of the control beam. In Fig. 11a, the analytical values obtained for the beams PFCB10R were generally similar to the test values, with the exception of the debonding load; however, the analytical values for the load and deection under the debonding load were about 20% different from those of the test values. The ultimate load of all the beams, with the exception of the 20% prestressed beam (PFCB1-2R), were almost the same, 120 kN. Thus, the analytical results for the reinforced concrete beams with bonded prestressed CFRP plates show good agreement with those obtained experimentally for both the yield and ultimate loads. Fig. 11b shows the relationship between the load and CFRP plate strain for the externally bonded prestressed beams. Both

504

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

140
PFCU1-0R (test) PFCU1-0R (analy.) PFCU1-2R (analy.) PFCU1-4R (analy.) PFCU1-6R (analy.) CONTROL (analy.)

120 100

PFCU1-2R (test) PFCU1-4R (test) PFCU1-6R (test) CONTROL (test)

Load(kN)

80 60 40 20 0 0 20 40 60 80 100 120

Deflection(mm)

(a) Loaddeflection relations

140 120 100

Load(kN)

80 60 40
PFCU1-0R (test) PFCU1-0R (analy.)

20
PFCU1-6R (test) PFCU1-6R (analy.)

0 0 2000 4000 6000 8000 10000 12000 14000

Microstrain

(b) Load CFRP plate strain relations

Fig. 10. Comparison of results for reinforced concrete beams strengthened with unbonded prestressed CFRP plates.

beams show about 10% strain differences between the analytical and test values for the initial and debonding loads, but after the two stages of FRP debonding, the analytical values are similar to those obtained in the tests. 4.6. Beams strengthened with prestressed CFRP Plates in terms of L/hf effect The failure mode, according to the length-to-depth (L/hf) effect, of beams strengthened with externally unbonded prestressed CFRP plates was the same as for those strengthened using the bonded prestressed system. That is, the stiffness until the initial cracking load was the same regardless of prestressing or bonding. In comparison to the bonded system beams, those with unbonded systems showed a signicantly increased ductile behaviour from the yield to the ultimate loads. The exural behaviour was observed to be the same, even if the span length of the member was increased. Fig. 12a shows the relationships between the loads and deections according to the beam span length. According to the analytical values for the 60% prestressed unbonded beam (PFCU1-6L2) with a 450 cm span length, the analytical values for the deection

under ultimate load differed by about 10 mm compared to the test values, but in the cracking and yield loads they were similar. Additionally, the load results obtained for the beams PFCU1-6L3 were generally 10% higher than the test values, but were similar in relation to the deection. Fig. 12b compares the analytical and test values obtained for the CFRP plate strains. Beams strengthened by external unbonded prestressed CFRP plates, according to the L/hf value, showed similar failure modes to the beams mentioned: the analytical results obtained for the reinforced concrete beams with bonded prestressed CFRP plates showed good agreement with those obtained experimentally in relation to the load capacities and CFRP plate strains. 4.7. Behaviour of bonded and unbonded systems As shown in Fig. 13a, the CFRP plate strains for all spans show a linear strain distribution in the bonded system and, after steel yielding, the gradient of the CFRP plate strains increased with the increasing portion of the load on the CFRP plates. Therefore, when the CFRP plate strain in the mid-span exceeds 6500 l, this will be the debonding strain.

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

505

160
PFCB1-0R (test) PFCB1-0R (analy.) PFCB1-2R (analy.) PFCB1-4R (analy.) PFCB1-6R (analy.) CONTROL (analy.)

140 120 100

PFCB1-2R (test) PFCB1-4R (test) PFCB1-6R (test)

Load (kN)

CONTROL (test)

80 60 40 20 0 0 20 40 60 80 100 120

Deflection (mm)

(a) Loaddeflection relations

160
PFCB1-0R (test) PFCB1-0R (analy.) PFCB1-6R (analy.)

140 120 100 80 60 40 20 0 0 2000 4000 6000

PFCB1-6R (test)

Load (kN)

8000

10000

12000

14000

Microstrain

(b) LoadCFRP plate strain relations

Fig. 11. Comparison of results for reinforced concrete beams strengthened with bonded prestressed CFRP plates.

After debonding of the CFRP plates occurred in a bonded system, the behaviour of the bonded CFRP-plated beams changed to that of the unbonded CFRP-plated beams due to the effect of the anchorage system. Therefore, it was predicted that a redistribution of stress affects in the whole span of the CFRP plate, which is made uniform by debonding, as shown in Fig. 13b. However, the CFRP plate strain after debonding is not uniform over the whole span. Because of the friction with the epoxy resin bonded to the CFRP plate surface and concrete paste, it seems that uniform strain distributions cannot occur. As shown in Fig. 13c, the CFRP plate strain distribution in the unbonded system was uniform over the whole span, and maintained an identical pattern with increasing load. In other words, the CFRP plate strain of the mid-span in the bonded system was similar to that of the steel bar until the yield load. However, the increment of the CFRP plate strain in the unbonded system was very small until the yield load. The strain of the CFRP plates in the bonded system increased linearly. However, the CFRP plate strains in the unbonded system were less than those in the member due to the distribution of the CFRP plate stress. Moreover, the rupture strain for the CFRP

plates in the unbonded system decreased compared to that in the bonded system. 4.8. Estimation of ductility In this paper, the safety of the members is assessed by using ductility indices, which are expressed as the ratio of the deections when a member is subject to the ultimate load to that when it yields. The ductility index for each beam is shown in Table 4. Beams with bonded CFRP plates without the anchorage system showed a lower ductility index than the control beam, due to brittle failure caused by debonding failure between the concrete and CFRP plates. Disregarding the 60% prestressed CFRP-plated beams (PFCB1-6R, PFCU1-6R), the ductility index of the prestressed CFRP-plated beams was greater than that of the control beam. Therefore, the ductility index was increased with a decreased prestressing. However, the ductility of beams strengthened by CFRP plates with an anchorage system was considered high when the ductility index was above 3. Generally, the ductility indices of the beams in a bonded system were less than that of those in an

506

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

160
PFCU1-6L2 (test) PFCU1-6L2 (analy.)

140
PFCU1-6L3 (test) PFCU1-6L2 (analy.)

120 100 80 60 40 20 0 0 20 40 60 80 100 120 140 160 180 200

Load(kN)

Deflection(mm)

(a) Loaddeflection relations

140

120

100

Load (kN)

80

60

40
PFCU1-6L2 (test) PFCU1-6L2 (analy.)

20
PFCU1-6L3 (test) PFCU1-6L3 (analy.)

0 0 1000 2000 3000 4000 5000 6000 7000

Microstrain

(b) LoadCFRP Plate strain relations

Fig. 12. Comparison of results for reinforced concrete beams strengthened with prestressed CFRP plates in terms of the L/hf effect.

unbonded system. That is, the load and the deection at the yield load of beams in a bonded system, which is a composite section with CFRP plates, were more than those of beams in an unbonded system. The average ratio of the test and analytical deection values for the yield and ultimate loads were 0.96 and 1.06, with a coefcient of variation of 12.91% and 11.81%, respectively. Thus, the analytical results for reinforced concrete beams strengthened by CFRP plates show good agreement with those obtained experimentally for the deection under the yield and ultimate loads. 5. Conclusions In this paper, exural tests and a nite element method analysis, using the DIANA program, were performed for reinforced concrete beams strengthened by prestressed CFRP plates. The main conclusions drawn from the study were as follows: (1) For the exural tests performed on reinforced concrete beams strengthened with prestressed CFRP plates, whether bonded or not, the ultimate load of the beams strengthened

with CFRP was reasonably constant. The observed failure mode of the prestressed CFRP plated-beams was not debonding, but FRP rupture. (2) For reinforced concrete members strengthened with externally bonded prestressed CFRP plates, two stages of FRP debonding occurred. After the debonding of the CFRP plates in the bonded system, the behaviour of the bonded CFRP-plated beams changed to that of unbonded CFRP-plated beams due to the effect of the anchorage system. (3) A nite element analysis of the reinforced concrete beams strengthened with CFRP plates was performed using the DIANA program, which considered nonlinear concrete material behaviour, the steel reinforcement, and an interfacial bond-slip model between the concrete and CFRP plates. The analytical results showed good agreement with those obtained experimentally for the debonding load, yield load, and ultimate load. (4) Beams with bonded CFRP plates without an anchorage system showed far lower ductility indices than the control beam, due to brittle failure caused by debonding failure between the concrete and CFRP plates. However, the

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

507

7000
20kN

6000

40kN 60kN 80kN

2nd debonding

5000

1st debonding

Microstrain

4000

Yield load

100kN

3000
120kN

2000
123kN

1000

0 0 15 30 45 60 75 90 105 120 135

Distance from supporting point (cm)

(a) CFRP plate strains in a bonded system

Strain in CFRP Plates (a) Before debonding failure

Strain in CFRP Plates (b) after debonding failure

(b) CFRPplate strain of a bonded system

6000
122kN

5000

100kN 83.6kN

CFRP plate strain (*10-6)

4000
80kN

CFRP plates rupture

3000

60kN 40kN

2000
20kN
Yield load

1000

0 0 15 30 45 60 75 90 105 120 135

Distance from supporting point (cm)

(c) LoadCFRP plate strains of an unbonded system

Fig. 13. Comparisons of bonded and unbonded system.

ductility of beams strengthened by CFRP plates with an anchorage system was considered high when the ductility

index was above 3. The ductility indices of the beams in the bonded system were less than those of the unbonded

508 Table 4 Estimates of ductility Beams Deection of yield load Test (a) (mm) Control NFCB1 NFCBW2 PFCU1-0R PFCU1-2R PFCU1-4R PFCU1-6R PFCB1-0R PFCB1-2R PFCB1-4R PFCB1-6R PFCU1-6L2 PFCU1-6L3 Mean C.O.V (%) 5.7 7.2 7.8 5.8 4.9 5.8 5.3 6.5 7.2 5.9 8.3 9.8 12.8

D.-S. Yang et al. / Composite Structures 88 (2009) 497508

Deection of ultimate load Anal. (b) (mm) 5.4 6.5 7.4 4.7 4.8 4.5 5.5 6.0 7.0 7.5 8.9 8.0 12.6 (b)/(a) 0.95 0.90 0.95 0.81 0.98 0.78 1.04 0.92 0.97 1.27 1.07 0.82 0.98 0.96 12.91 Test (a) (mm) 33.6 15.4 16.2 55.7 43.3 34.4 29.5 58.0 49.3 32.4 32.0 73.8 102.0 Anal. (b) (mm) 34.7 15.5 15.6 53.7 43.1 34.5 24.8 55.5 31.1 34.4 30.4 82.0 102.1 (b)/(a) 1.03 1.01 0.96 0.96 1.00 1.00 0.84 0.96 0.63 1.06 0.95 1.11 1.00 0.96 11.81

Ductility index Test 5.89 2.14 2.08 9.60 8.84 5.93 5.57 8.92 6.85 5.49 3.86 7.53 7.97 Anal. 6.43 2.38 2.11 11.43 8.98 7.67 4.51 9.25 4.44 4.59 3.42 10.25 8.10

system. That is, the load and the deection under the yield load of the beams in a bonded system, a composite section with CFRP plates, were more than those of the beams in a bonded system. References
[1] Pendhari SS, Kant T, Desai YM. Application of polymer composites in civil construction: a general review. Compos Struct 2008;84:11424. [2] Wang YC, Lee MG, Chen BC. Experimental study of FRP-strengthened RC bridge girders subjected to fatigue loading. Compos Struct 2007;81:4918. [3] Silva MAG, Biscia H. Degradation of bond between FRP and RC beams. Compos Struct 2008;85:16474. [4] Ji G, Li G, Li X, Pang SS, Jones R. Experimental study of FRP tube encased concrete cylinders exposed to re. Compos Struct 2008;85:14954. [5] Deniaud C, Neale KW. An assessment of constitutive models for concrete columns conned with bre composite sheets. Compos Struct 2006;73: 31830. [6] Alfarabi SA, Al-Sulaaimani GJ, Basunbul IA. Strengthening of initially loaded reinforced concrete beams using FRP plates. ACI Struct J 1994;91:1608. [7] Bizindavyi L, Neale KW. Transfer lengths and bond strengths for composites bonded to concrete. J Compos Constr ASCE 1999;3(4):15360. [8] Chen JF, Teng JG. Anchorage strength models for FRP and steel plates bonded to concrete. J Struct Eng 2001;127(7):78491.

[9] Fanning PJ, Kelly O. Ultimate response of reinforced concrete beams strengthened with CFRP plates. J Compos Constr ASCE 2001;5(2):1227. [10] KICT. Development of new technology on strengthening reinforced concrete structures with externally post-tensioning CFRP strips (2nd Year). Korea Institute of Construction Technology; 2005 [in Korean]. [11] Malek AM, Saadatmanesh H, Ehsani MR. Prediction of failure load of R/C beams strengthened with FRP plate due to stress concentration at the plate end. ACI Struct J 1998;95(1):14252. [12] Michel J, Chajes TF, Januszka DR, Mertz TA, Thomson JR, Willam WF. Shear strengthening of reinforced concrete beams using externally applied composite fabrics. ACI Struct J 1995:295303. [13] Nguyen DM, Chan TK, Cheong HK. Brittle failure and bond development length of CFRP-concrete beams. J Compos Constr, ASCE 2001;5(1):127. [14] Park SK, Yang DS. Flexural behaviour of reinforced concrete beams with cementitious repair materials. Mater Struct 2005;38:32934. [15] Rahimi H, Hutchinson A. Concrete beams strengthened with externally bonded FRP plates. J Compos Constr, ASCE 2001;5(1):4456. [16] Ritchie PA, Thomas DA, Lu LW, Connely GM. External reinforcement of concrete beams using bre reinforced plastics. ACI Struct J 1991;88(4): 490500. [17] Sebastian WM. Signicance of mid-span debonding failure in FRP-plated reinforced concrete beams. J Struct Eng, ASCE 2001;127(7):7928. [18] Teng JG, Smith ST, Yao J, Chen Jf. Intermediate crack-induced debonding in reinforced concrete beams and slabs. Constr Build Mater 2003:44762.