Flexural behavior of aged prestressed concrete girders strengthened with various FRP systems

Transcription

1 Construction and Building Materials 21 (2007) Construction and Building MATERIALS Flexural behavior of aged prestressed concrete girders strengthened with various FRP systems Owen Rosenboom, Tare K. Hassan, Sami Rizkalla * Department of Civil Engineering, North Carolina State University, Centennial Campus, Raleigh, NC , United States Received 20 January 2005; received in revised form 24 May 2006; accepted 19 June 2006 Available online 7 September 2006 Abstract Many prestressed concrete bridges are in need of upgrading in order to increase their posted capacities. Departments of transportation across the country have been faced with large financial burdens on the maintenance budget, negative psychological effects on highway users, long traffic delays during maintenance, potential safety hazards, and reduced service life as a result of the deficiencies. In response to considerable consultation with the North Carolina Department of Transportation (NCDOT), a research project with practical goals was initiated to evaluate the cost-effectiveness and value engineering of Carbon Fiber Reinforced Polymer (CFRP) repair and strengthening systems for prestressed concrete bridge girders. This paper presents the first phase of the research program, involving the testing under static loading conditions of eight prestressed concrete bridge girders, six strengthened with various CFRP systems. Results show that the ultimate capacity of prestressed concrete bridge girders can be increased by as much as 73% using CFRP without sacrificing the ductility of the original member. Transverse CFRP U-wrap reinforcements are recommended along the length of the girder to control debonding type failures. The second phase of the research will examine the fatigue behavior of the strengthened girders, and provide analysis under service loading conditions. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Prestressed; Strengthening; Fiber reinforced polymers; Bridge girder; Near surface mounted; Externally bonded; Flexural behavior 1. Introduction 1.1. Research objectives Many prestressed concrete bridges are in need of upgrading in order to increase their posted capacities. Departments of transportation across the country have been faced with large financial burdens on the maintenance budget, negative psychological effects on highway users, long traffic delays during maintenance, potential safety hazards, and reduced service life as a result of the deficiencies. In response to considerable consultation with the North Carolina Department of Transportation (NCDOT), * Corresponding author. Tel.: ; fax: addresses: oarosenb@ncsu.edu (O. Rosenboom), (T.K. Hassan), sami_rizkalla@ncsu.edu (S. Rizkalla). a research project with practical goals was initiated to evaluate the cost-effectiveness and value engineering of Carbon Fiber Reinforced Polymer (CFRP) repair and strengthening systems for prestressed concrete bridge girders. The use of CFRP began after World War II in military high-performance applications. Use within the civil infrastructure, either as a retrofit material or in construction of new structures, gained popularity in the 1980s. Today, CFRP pre-cured bars, strips, and tendons, as well as wet lay-up sheets are commonly used in the repair and retrofit of concrete structures. Some of the major benefits of CFRP include its high strength to weight ratio, high fatigue endurance and the ease of fabrication, manufacturing, handling and installation. Codification of the use of CFRP in civil infrastructure is quickly catching up with the rapid growth in installations, with many code writing agencies around the world concerned with this task [1 3] /$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi: /j.conbuildmat

2 O. Rosenboom et al. / Construction and Building Materials 21 (2007) Included in this paper are results and analysis from the first phase of the research, concerned with the flexural behavior under static loading conditions of prestressed concrete girders strengthened with various CFRP systems. After a brief review of the literature on the topic, test results are presented on the static testing to failure of eight girders. Six were strengthened with various CFRP systems including externally bonded sheets and strips and near surface mounted bars and strips while two were control girders. Two different analysis procedures are included which can determine accurately the ultimate capacity of a CFRP strengthened prestressed concrete girder. Details of an additional phase of the research program concerning the fatigue behavior and design under service loading conditions of such girders is provided elsewhere [4] Background One of the most common uses of CFRP materials is to externally bond the material directly to the concrete surface. Externally bonding CFRP systems to concrete can be achieved by bonding pre-cured laminates, or applying wet lay-up sheets. The systems are typically bonded to the tension side of the members to increase the flexural strength and/or on the sides of the member to improve the shear capacity. Installation of this technique is relatively simple and can be achieved in a very short time. The system must be designed to avoid premature failure due to possible delamination or debonding of the CFRP material from the concrete surface, a failure mode first noted in reinforced concrete beams strengthened with mild steel plates [5]. Many researchers today are concerned with the task of controlling and predicting debonding type failures: terminology and methodology of the FRP to reinforced concrete debonding failure proposed by Teng et al. [6] and Oehlers et al. [7] has been generally accepted and slowly makes it way into codification. ACI Committee 440F [1] deals with the debonding mechanism by applying a bond reduction coefficient, but its use is overly conservative for CFRP strengthened prestressed concrete due to initial strains present in the concrete due to prestressing. In response to premature debonding failures, the use of near surface mounted (NSM) CFRP systems was introduced by several researchers [8 10]. By inserting a CFRP bar or strip into a pre-cut groove and filling with epoxy, NSM systems can reduce the propensity for debonding failures. In addition, the embedded nature of the technique provides superior environmental performance in comparison to externally bonded techniques, as well as protection against possible accidental and intentional damage. There has been little research on the strengthening of prestressed concrete with CFRP materials. There have been some field applications using CFRP to repair prestressed concrete [11], yet few full scale specimens have been tested to failure. Takacs and Kanstad [12] showed that prestressed concrete girders could be strengthened with externally bonded CFRP to increase their ultimate flexural capacity. Two 11.3 m (37 ft) specimens plated with precured CFRP laminates achieved an increase in flexural moment capacity of 28% and 37%, respectively. They accurately predicted the behavior using a finite element model based on the smeared crack methodology. Hassan and Rizkalla [13] examined the flexural behavior of prestressed concrete bridge slabs strengthened with various CFRP systems. The flexural capacity of the slabs could be increased by as much as 50% using the CFRP strengthening, with the most cost effective solution being the CFRP sheets. 2. Experimental program 2.1. Test girders As part of an extensive research program sponsored by the North Carolina Department of Transportation eight prestressed concrete C-Channel bridge girders were statically tested to failure. The C-Channel prestressed concrete girder is a superstructure member commonly used for short span bridges in rural areas, typically used to span small streams or estuaries. Bridges using this type of girder were built between the late 1950s to the mid 1970s. Of the eight girders tested under static loading conditions, five had a Type I prestressing configuration with ten 1723 MPa (250 ksi) prestressing strands and came from a decommissioned bridge erected in The other three girders were Type II girders, prestressed with eight 1862 MPa (270 ksi) strands and also came from a decommissioned bridge. The two types of prestressing steel configurations along with the cross-sectional dimensions and other reinforcement details of the girders are shown in Fig. 1. The measured camber for all the girders in both prestressing configurations was 32 mm (1.25 in.). The Ramberg Osgood function [14] was used to match the stress versus strain behavior of the prestressing strands. " # 1 A f p ¼ E p e ps A þ ð1þ ½1 þðbe ps Þ C Š 1=C where f p is the prestressing strand stress, E p is the modulus of elasticity of the prestressing strand, e ps is the strain in the prestressing strand, and A, B, C are material constants. The constants A, B and C were determined from the stress strain relationships generated from tension tests. For the Type I girders with 1724 MPa (250 ksi) strands, the average values for A, B and C were 0.025, and 6, respectively. For the Type II girders with 1862 MPa (270 ksi) strands the average values for A, B and C were calculated to be 0.017, and 7. The nominal concrete strength specified for the C-Channel prestressed concrete girders was 34.5 MPa (5000 psi) at 28 days and 27.6 MPa (4000 psi) at transfer of prestressing force. During the design stage for strengthening of these girders, the concrete compressive strength was estimated

3 766 O. Rosenboom et al. / Construction and Building Materials 21 (2007) U-Wrap for externally bonded FRP systems only A mm 4572 mm A @ End strand spacing NSM1S Prestressing strand 1 Ø10 /web CFRP bar Prestressing strand Sec A-A 40 Two-part epoxy 20 Prestressing strand Two-part epoxy Thickness=1.5 mm 40 1 CFRP strip/web (1.2x50 mm) EB1S Ø10/180 Ø12/330 Ø/180 Ø/330 TYPE I 2@ CL strand spacing End strand spacing TYPE II NSM2S EB2S 2@ CL strand spacing Prestressing strand 2 CFRP strips/web 2 x (2x16mm) Two-part epoxy Prestressing strand Two-part epoxy 4 Plies/web (100 mm wide) EB CFRP sheets Sheet thickness = 2.4 mm Prestressing strand 40 Two-part epoxy 40 Two-part epoxy 3 Plies/web (100 mm wide) 60 EB CFRP sheets Sheet thickness = 1.0 mm 60 EB3S EB4S 5 Plies/web (125 mm wide) EB CFRP sheets Sheet thickness = 1.0 mm Fig. 1. C-Channel elevation, prestressing configurations, and strengthening details. using the following equation which accounts for the strength increases due to aging [15]. f 0 0 cðtþ ¼fc ð28þ t ð2þ 4 þ 0:85ðtÞ where fc 0 ðtþ is the concrete compressive strength as a function of time and t is the time in days. Three concrete core samples were obtained from each of the tested girders. The average concrete strength was 69.5 MPa (10,100 psi) and 56.3 MPa (8200 psi) for the Type I and Type II girders, respectively Design of the strengthened girders The design of the strengthened girders proceeded after testing the control girder. Three different levels of strengthening were examined, the design criteria being to achieve a 20%, 40% or 60% increase in the ultimate load carrying capacity with respect to the control girder. The design of each strengthened girder was based on a cracked section analysis program, Response 2000 [16]. The base curve of the concrete compression model used in the cracked section analysis program was the Popovics curve, compression

4 O. Rosenboom et al. / Construction and Building Materials 21 (2007) softening was determined using the Vecchio Collins model and the tension stiffening was determined from the Bentz model [16]. In the preliminary design, the compression strength of the concrete was based on the compression strength specified in the original NCDOT drawings modified according to Eq. (2) to account for the age of concrete at the time of testing. The compression strength used in the design was 41.4 MPa (6000 psi). The prestressing strands were all assumed to be 1724 MPa (250 ksi) for Type I prestressing configuration modeled using the Ramberg Osgood equation with 0.030, 121 and 6 for the coefficients A, B and C, respectively. The CFRP systems used in the design were modeled as linear elastic up to failure using material properties provided by the manufacturer. Flexural failure, defined as rupture of the FRP or crushing of the concrete in compression, was the desired mode of failure. It was recognized that externally bonded systems are more prone to debonding failures than near surface mounted systems. According to Malek et al. [17], shear stresses developed at the FRP cut-off point for the externally bonded systems were significantly lower than the shear strength of the concrete. Therefore, plate-end debonding was not expected to occur. However, in order to delay FRP delamination-type failures along the length of the girder, 150 mm (6 in.) wide U-wraps were installed at 900 mm (3 ft) spacing for all externally bonded strengthened girders. This arrangement was selected to simulate typical anchorage details commonly used by the construction industry for reinforced concrete members strengthened with FRP. Three types of CFRP systems were used in this research: externally bonded wet lay-up type systems, externally bonded pre-cured laminates, and near surface mounted (NSM) systems. The shape, type and amount of CFRP applied to the soffits of the C-Channel girder are shown in Fig. 1. The strengthening systems used for girders NSM1S and NSM2S consisted of one CFRP bar or two CFRP strips placed in near surface mounted grooves in each soffit. The strengthening system used for girder EB1S consisted of one CFRP laminate per soffit. Girders EB2S, EB3S and EB4S were strengthened with various configurations of either normal or high modulus CFRP wet lay-up sheets. The material properties of each CFRP system was determined through constitutive testing, the results of which are shown in Table Test setup and instrumentation All girders were tested in three-point loading using a 490 kn (110 k) hydraulic actuator mounted to a steel frame at midspan as shown in Fig. 2. To simulate field loading conditions, a set of truck tires filled with silicon rubber were used as a contact surface while applying the load. The girder was supported at both ends on a 64 mm (2.5 in.) thick neoprene pad which in turn rested on a 25 mm (1 in.) steel plate. The behavior during testing was measured using a combination of string potentiometers placed along the girder span, and a combination of PI gauges (a strain gauge mounted to a spring plate) and Fig. 2. Typical test setup. Table 1 CFRP tension test results Girder strengthened Specimen Thickness (mm) Tensile strength (MPa) E F (MPa) e fu (%) NSM1S Manufacturer , NSM2S Experimental average , Manufacturer , EB1S Experimental average , Manufacturer , EB2S Experimental average , Manufacturer ,765 1 EB3S Experimental average , Manufacturer ,976 1 EB4S Experimental average , Manufacturer ,

5 768 O. Rosenboom et al. / Construction and Building Materials 21 (2007) electric resistance strain gauges located around the midspan section to determine the strain profile along the depth of the girder. The loading sequence began by increasing the applied load up to a load level slightly higher than the cracking load. The girder was then unloaded, and reloaded again at a rate of 2.5 mm/min (0.1 in/min) up to the load level equivalent to yielding of the prestressing strands. This loading sequence was selected to determine the effective prestressing force in the girders by observing the re-opening of the flexural cracks. From the measured load at reopening of the flexural crack at midspan, P ro, the average effective prestressing strands, P eff, can be determined according to the following equation: 0 ¼ M D S b þ L P ro 10 P eff X d i P eff 4 S b A c S b where M D is the moment due to the dead load, S b is the bottom section modulus, L is the span, A c is the area of the concrete section, and d i are the locations of the different layers, i, of prestressing strand measured from the neutral axis of the section. When the girders were loaded beyond yielding of the prestressing strands, the rate of the applied load was increased to 5 mm/min (0.2 in/min) up to failure. The average prestress force for Type I and Type II girders was 70.5 kn (15.9 k) and 82.7 kn (18.6 k), respectively. ð3þ Five girders with a Type I prestressing configuration were tested statically to failure. One of these was a control girder, and four were strengthened with various CFRP systems. Summarized test results are given in Table 2, including the measured cracking load, crack reopening load and ultimate load for each tested girder. Brief test descriptions are provided below, while a discussion of the results is presented in the next section. No visible flexural cracks were observed in the control girder (girder CS) upon delivery to the laboratory. The girder was loaded statically up to failure. Yielding of the lower prestressing strands took place at a load level of 116 kn (26 k) according to readings of the PI gauges. Failure occurred due to crushing of concrete at a load level of 148 kn (33.2 k) at an ultimate deflection of 228 mm (9 in.) as shown in Fig. 3. Due to the confining effect induced by the loading tires, crushing of the concrete occurred first at the edge of the girder at midspan before it extended underneath the loading area Type I girders Fig. 3. Crushing of concrete failure in control girder, CS. Table 2 Summarized test results Specimen designation CS NSM1S NSM2S EB1S EB2S CF2 EB3S EB4S Strengthening NSM bars NSM strips EB strips EB sheets EB sheet EB HM sheets Prestressing configuration Type I Type I Type I Type I Type I Type II Type II Type II P cr (kn) P ro (kn) P e (kn) Losses (%) P ult (kn) S (%) increase in capacity e c, ultimate compressive strain in concrete (experimental) (%) e t, ultimate tensile strain in CFRP (experimental) (%) Experimental/manufacturer tensile strain in CFRP (%) Failure mode a C C C D R C C R Initial stiffness (kn/m) b Secondary stiffness (kn/m) c Structural efficiency (%/kn) d a C = crushing of concrete, R = rupture of CFRP, D = debonding of CFRP. b Defined from 9 to 44.5 kn (2 11 k). c Defined from to kn (28 33 k). d Defined as (% increase in capacity)/(e FRP A FRP ).

6 O. Rosenboom et al. / Construction and Building Materials 21 (2007) Behavior of the two girders, NSM1S and NSM2S, strengthened with a NSM system using either CFRP bars or strips, respectively, was similar during testing including the failure modes. No cracks were observed in either girder at the time of delivery. The initial stiffness of both girders was similar to that of the control girder, as was the post-cracking stiffness. After yielding of the prestressing strands, the presence of the CFRP reinforcement constrained opening of the cracks and consequently reduced the midspan deflection compared to the control girder. Failure of both girders strengthened with NSM CFRP reinforcement was due to crushing of the concrete at the extreme compression zone followed by debonding of the NSM CFRP reinforcement at a load of 181 kn (40.8 k) for the NSM bars and 180 kn (40.7 k) for the NSM strips. Test results showed that strengthening of the prestressed girders using NSM CFRP bars and strips increased the ultimate load carrying capacity of the girder by 22.9% and 22.6%, respectively compared to the control girder. The behavior of the prestressed concrete girder strengthened with externally bonded CFRP strips (EB1S) matched that of the NSM strengthened girders before and after cracking. Failure occurred due to debonding of the CFRP strips at a load level of 176 kn (39.6 k) as shown in Fig. 4. The maximum recorded strain in the CFRP prior to debonding was 72% of the rupture strain measured during tension coupon tests. Compared to the control girder, EB1S achieved an increase in the ultimate load carrying capacity of 19%. The girder strengthened with externally bonded CFRP sheets (EB2S) was designed to achieve an increase of 60% in the ultimate load carrying capacity compared with the control girder. The initial stiffness of this girder was similar to that of the control. Failure was due to rupture of CFRP sheets at a load of 236 kn (53.1 k) providing an increase of 60% over the ultimate load of the control girder Type II girders Three girders with a Type II prestressing configuration were tested under static loading conditions. Since it was not known prior to testing that the girders had different prestressing strand configurations, no control girder was tested for the Type II configuration. However, a Type II specimen was tested as a control girder under fatigue loading conditions [4] and showed very little degradation after 2 million cycles. It is the data from the monotonic load to failure which was applied after completion of the fatigue loading which is being used in this context as a control for the girders with a Type II prestressing strand configuration. The control girder prestressed with a Type II configuration, CF2, behaved similar to the control girder with a Type I prestressing configuration, CS, before and after cracking. Failure occurred due to crushing of the concrete at a load of 142 kn (32.0 k) which was 3.6% less than the ultimate strength achieved in the static test of the control girder prestressed by a Type I C-Channel prestressing configuration (CS). Flexural cracking of girder EB3S, strengthened with three layers of CFRP sheets, occurred at a load of 62 kn (13.9 k). At an applied load of 201 kn (45 k), interfacial debonding was initiated at the location of the flexural cracks at midspan spread to cause debonding between the U-wraps located on either side of midspan. The debonding did not cause failure, however, but occurred due to crushing of the concrete at a load of 246 kn (55.3 k). The energy released at failure due to crushing of concrete led to peeling of the U-wraps from the webs of the girder, but did not cause the CFRP to rupture. Girder EB3S achieved an increase of strength of 72.8% compared to the ultimate load of the Type II control girder (CF2). The measured performance of the girder exceeded the design values due to a high rupture strain measured in the CFRP during the test, which was 33.3% higher than the value reported by the manufacturer. Flexural cracking of girder EB4S strengthened with high modulus CFRP sheets occurred at a load of 64 kn (14.3 k). At a load of 150 kn (33.7 k) failure occurred due to rupture of the CFRP sheets at midspan, which provides an increase of only 5.3% compared to girder CF2 during the final static test. The tensile strain in the CFRP achieved during this test was 13.3% lower than the value reported by the manufacturer. After rupture of the CFRP, the test was continued until ultimate failure occurred due to a combination of progressive CFRP rupture and concrete crushing close to the values of ultimate load and displacement of the control girder. 3. Analytical modeling Fig. 4. Debonding failure in girder EB1S. In order to present, the test results along with their analytical predictions, the analytical procedures are presented

7 770 O. Rosenboom et al. / Construction and Building Materials 21 (2007) in this section. For the two types of analysis, the material properties used within were determined from constitutive material testing. The prestressing steel was modeled using Eq. (1) with coefficients determined from material testing. Characteristics of the CFRP material were also based on material testing and presented as linear elastic to failure. Concrete compressive strength was determined from core samples and the stress strain characteristics described below Non-linear finite element simulation ANACAP Ó is a non-linear finite element program for analysis of plain, reinforced and prestressed concrete members and structures and was developed by the ANATECH Corporation. The program was used to run finite element simulations on all the tested girders with the various strengthening configurations. ANACAP Ó uses the smeared cracking methodology for modeling of concrete where cracking is assumed to be distributed over an entire element. This mechanics-based philosophy uses plasticity theory that incorporates cracking and other concrete properties. The effect of concrete confinement at different stress levels was incorporated into the analysis as well as an elastic modulus allowing for changes between the three distinct zones of the stress strain curve of concrete the initial linear region, strain hardening region and strain softening region. The concrete cross-section of the girder was modeled with 20 node elements using quadratic isoparametric displacement interpolation with mesh shown in Fig. 5. Experimental verification of the accuracy of the program can be found elsewhere [9,10] Cracked section analysis In addition to the non-linear finite element simulation, a cracked section analysis was performed for the strengthened girders using Response 2000 Ó software. Verification of this program can be found elsewhere [16]. The base curve of the concrete model used in the cracked section analysis program was the Popovics curve including compression softening which was based on the Vecchio Collins model and the tension stiffening was based on the Bentz model [16]. The results from the cracked section analysis showed good agreement with the non-linear finite element analysis and the experimental results presented in the following section. Based on good agreement achieved between the cracked section analysis and the experimental results, the need for a finite element simulation to predict the behavior of such girders is not warranted. 4. Test results and discussion 4.1. Crack development Initiation of the flexural cracks was determined either by visual inspection or by analysis of the test data. Typically, cracking occurred between the loads kn ( k) for both the Type I and Type II girders. Flexural cracks were located at the bottom of the C- Channel soffit near midspan, with a distance from midspan equal to the depth of the girder from the edge of the loading area. Spacing of the cracks was approximately 330 mm (13 in.), which corresponds to the distance between the transverse stirrups used for the C-Channels. The CFRP strengthening reduced crack spacing, crack width and crack growth for all of the strengthened girders with respect to the control girder. PI gauges were mounted at the level of the lower prestressing strand on both sides of the C-Channel soffit to measure the tensile strain in the concrete at various load levels. The average crack width at midspan can be calculated using the measured strains at any applied load level from the following equation: x360= Supports 6x127=762 Y Z X Load Axes of Symmetry Fig. 5. Mesh used for finite element modeling.

8 O. Rosenboom et al. / Construction and Building Materials 21 (2007) Fig. 6. Average crack width at midspan for Type I girders. Fig. 7. Average crack width at midspan for Type II girders. CW ave ¼ ðe ci e ccr Þl PI n where e ci is the measured strain at a certain applied load level, e ccr is the measured strain in the concrete at the flexural cracking load, l PI is the length of the PI gauge, and n is the number of cracks observed within the PI gauge length. The load versus the average crack width at midspan for the Type I and Type II girders are shown in Figs. 6 and 7, respectively. The figures indicate that the presence of the strengthening system restrained crack opening and growth with respect to the control girders. At ultimate, the crack widths of the strengthened girders were as much as 400% less than the control girder. ð4þ 4.2. Stiffness The initial stiffness and secondary stiffness of the Type I and Type II girders are given in Table 2. Comparing the initial stiffness of the strengthened girders which achieved a 20% increase in ultimate load capacity, with the initial stiffness of the control girder, it was obvious that the strengthening system has very little effect on the initial stiffness. However, using a strengthening system to achieve a 60% increase in ultimate capacity, modest increases in initial stiffness can be obtained. The initial stiffness of the strengthened Type II girders was 9% and 11% higher than the control girder for girders EB3S and EB4S, respectively. The girder strengthened with the high modulus material

9 772 O. Rosenboom et al. / Construction and Building Materials 21 (2007) has a secondary stiffness nearly double the girder with normal modulus material Structural efficiency The structural efficiency, SE, of a CFRP strengthening system was evaluated using the following expression: SE ¼ S ; %=k ð%=mnþ ð5þ E f A f where S is the percent increase in ultimate flexural capacity achieved using a CFRP system compared to the control girder and E f and A f are the elastic modulus and area of CFRP material, respectively. Material properties of the CFRP, including thickness of the laminates, were measured for each case. The structural efficiency as defined represents how efficient the CFRP strengthening system is with respect to the amount of material and its stiffness. The results, given in Table 2, indicate that NSM systems had the highest structural efficiency, around 1.4%/kN (6%/k). The NSM systems performed well under this definition as a result of the high rupture strains that were achieved during the test due to the superior bond characteristics that can be achieved for this type of system. The Type II girder strengthened with normal modulus CFRP sheets (EB4S) had a structural efficiency similar to the NSM strengthened girders as a result of the high level of strengthening achieved Ultimate load and displacement Test results indicate that the addition of a brittle material such as CFRP, to a ductile structural member such as prestressed concrete does not reduce the overall ductility of the member. As shown in Figs. 8 and 9, the ultimate Fig. 8. Experimental load versus midspan displacement for Type I girders. 250 EB3S 200 (kn) 150 EB4S CF2 Load CF2: Control EB3S: EB Sheets EB4S: EB Sheets Midspan deflection (mm) Fig. 9. Experimental load versus midspan displacement for Type II girders.

10 O. Rosenboom et al. / Construction and Building Materials 21 (2007) (kn) Load Applied Experimental Cracked Section Analysis Finite Element Simulation Midspan deflection (mm) Fig. 10. Analysis versus experimental for girder NSM1S. loads of a prestressed concrete girder can be substantially increased using CFRP materials without sacrificing the ductility of the section. Girder EB2S, strengthened with four layers of normal modulus CFRP sheets, failed due to rupture of the CFRP material. The measured concrete strain at ultimate was 0.3%. The measured tensile strain in CFRP at ultimate was 1.17% which is 117% of the manufacturer s specified rupture strain. Two of the strengthened Type I girders (NSM1S and NSM2S) experienced failure at ultimate due to concrete crushing which was the same failure mode as the control girder (CS). The maximum measured compressive strain in concrete at failure for both girders was 0.36%. The ductility of the prestressed concrete section was maintained in girder EB3S with the installation of normal modulus CFRP sheets. The ultimate load was 72.8% higher than the control girder, and the ultimate displacement was only 16% less than the control. The failure mode of girder EB3S was due to crushing of concrete, similar to the control girder. The maximum measured compressive strain in concrete at failure was 0.29%, slightly lower than the observed value for the control girder of 0.32%. The maximum tensile strain measured during testing of girder EB3S was 133% of the manufacturer recommended rupture strain. Girder EB4S failed due to rupture of CFRP material at a load of 150 kn (33.7 k), which represents an increase of 5.3% over the ultimate load measured in the testing of the control girder. The maximum measured tensile strain in the CFRP at the rupture event was 0.26%, which is 86.7% of the rupture strain used in design. The high modulus material was very sensitive to fiber orientation and this could have played a role in the difference between the design and the measured rupture strain. The failure was due to concrete crushing, after numerous CFRP rupture events along the length of the girder, at a measured concrete compressive strain of 0.25% Predicted versus experimental All girders strengthened with CFRP were analyzed using a cracked section analysis and finite element model. For the girder strengthened with NSM bars (NSM1S), the measured ultimate load was 181 kn (40.8 k), whereas the ultimate load determined from cracked section analysis and the finite element model was kn and kn (40.3 and 41.6 k), respectively, a difference of 1.2% and 1.9% as shown in Fig. 10. The finite element analysis, cracked section analysis and experimental load versus deflection behavior for girder EB3S is shown in Fig. 11. For comparison, the analyses shown for this girder were conducted using design material properties provided by the manufacturer, not values obtained during material testing. The strengthening scheme of girder EB3S was designed to increase the load carrying by 40% compared to the control girder. The measured ultimate capacity was increased by 72.8% because the stiffness and ultimate strength values of the CFRP were larger than the specified values used in design. 5. Cost-effectiveness analysis To closely resemble field conditions during strengthening, the girders were placed side by side, as they would be on a bridge, on top of a steel substructure approximately 2.5 m (8 ft) off the ground. To determine the costeffectiveness of each strengthening technique, the following items were considered: (1) labor costs of the professional FRP applicators, (2) time taken to complete all tasks, (3) material costs, and (4) equipment used for strengthening. The material costs include all primers, adhesives and CFRP required for field application for each technique. Equipment items include the rental of sandblasting pot, compressor, sand, and other items used in surface preparation. Also included in the equipment cost are items such as latex

11 774 O. Rosenboom et al. / Construction and Building Materials 21 (2007) (kn) Load Experimental Cracked Section Analysis Finite Element Simulation Midspan deflection (mm) Fig. 11. Analysis versus experimental for girder EB4S. gloves, plastic mixing buckets and items used in the CFRP installation. Other equipment (such as grinders, mixers, safety equipment, etc.) is either assumed to be provided by the contractor or used equally in each of the strengthening systems. A labor summary of the CFRP strengthening and the results of the cost-effectiveness analysis are given in Tables 3 and 4, respectively. The labor costs were determined using a wage of $45 per hour. Consultation with several FRP installers produced this value which is commonly used for cost estimation in FRP strengthening work. Table 3 Labor summary for CFRP strengthening systems (h) System designation NSM1 NSM2 EB1 EB2 EB4 EB5 EB6 Strengthening NSM bars NSM strips EB strips EB sheets EB sheets EB sheets EB HM sheets Gluing strips 1 Concrete repair Groove cutting Grinding/chipping Sandblasting Cutting of fiber CFRP lay-up Total (h/girder) Total (h/m) a a Length of strengthening = 8.23 m (27 ft). Table 4 Cost-effectiveness analysis for C-Channel CFRP strengthening systems (in US dollars) System designation NSM1 NSM2 EB1 EB2 EB4 EB5 EB6 Strengthening NSM bars NSM strips EB strips EB sheets EB sheets EB sheets EB HM sheets Main CFRP (m) CFRP U-wraps (m) Main adhesive (m) Included U-wrap adhesive (m) above Equipment (m) Labor (m) Total cost (m) % increase in strength a Cost-effectiveness b a For girders tested under static loading condition. b Based upon (percent increase in strength)/(total cost per meter).

12 O. Rosenboom et al. / Construction and Building Materials 21 (2007) The cost analysis indicates that the most cost-effective system, when comparing the variables described previously and combined with the percent increase in strength, were the girders which were strengthened using normal modulus CFRP sheets (girders EB2S and EB3S). The NSM systems also performed well using these criteria. When CFRP was used to increase the ultimate flexural capacity by 20% (NSM1S, NSM2S and EB1S), the labor required for installation of a NSM system was substantially higher than for installation of an externally bonded system at the same strengthening level. The difference in labor costs between EB3S (EB Sheets) and EB4S (EB HM Sheets) should be noted. Although the girders strengthened with the high modulus sheets had more material, it was not only the amount of material that caused in increased labor costs. The contractors also said that the material itself was harder to work with and install than equivalent normal modulus CFRP sheets. 6. Conclusions Eight 9.14 m (30 ft) prestressed concrete C-Channel girders were tested under static loading conditions, six strengthened with various CFRP systems and two tested as control girders. Based on the experimental program and analysis of the test results, the following conclusions can be made: 1. Proper design and installation of a CFRP strengthening system can lead to the failure mode of crushing of concrete in the compression zone, preserving the ductile structural response of an unstrengthened girder. 2. Externally bonded CFRP sheets are the most cost-effective strengthening technique and are the most applicable technique for these types of girders. 3. The most structurally efficient strengthening technique used is the near surface mounted (NSM) CFRP bars or strips system. 4. The ultimate flexural capacity of prestressed concrete can be increased substantially using CFRP materials. The flexural capacity of the C-Channel girders tested in this research program could be increased by as much as 73% with the use of externally bonded CFRP sheets. 5. The use of transverse CFRP U-wraps most likely delayed debonding failures in externally bonded CFRP systems. 6. For externally bonded CFRP wet lay-up systems, the experimental tensile strain in the CFRP outperformed the manufacturer s provided value. Therefore, the use of a bond reduction coefficient, j m, in ACI 440F 2002, would be conservative. 7. The crack spacing and crack widths at ultimate can be substantially reduced using CFRP strengthening. Crack widths observed during the testing of the C-Channels were reduced by as much as 400% using CFRP materials in comparison to the unstrengthened girder. 8. The initial and secondary stiffness of C-Channel girders can be increased by the use of high modulus CFRP materials. Using strictly a serviceability criterion, high modulus CFRP materials outperformed the normal modulus CFRP materials. Acknowledgements The authors acknowledge the support of the North Carolina Department of Transportation through Project Valuable help was provided on this project by Ronaldson Carneiro, and Dr. Amir Mirmiran. Several industry members made much appreciated donations: David White of the Sika Corporation, Doug Gremel of Hughes Brothers, Akira Nakagoshi of Mitsubishi Chemical America, Peter Emmons of Structural Preservation Systems and Ed Fyfe of Fyfe Corporation. Special thanks to Structural Preservation Systems and Fyfe Corporation for carrying out the strengthening and repair work. Thanks are also extended to the personnel at the NCDOT Bridge Maintenance Department, especially Dallie Bagwell and Tracy Stephenson for providing Bridge Maintenance facilities and constructing steel substructures for the repair and strengthening work. The authors thank Jerry Atkinson and Bill Dunleavy, technicians at the Constructed Facilities Laboratory at North Carolina State University, for their invaluable help. References [1] ACI Committee 440F. Design and construction of externally bonded frp systems for strengthening concrete structures. American Concrete Institute: ACI Manual of Concrete Practice, [2] Concrete Society. The design guidance for strengthening concrete structures using fibre composite materials. Technical Report 55. 2nd ed. United Kingdom: The Concrete Society; [3] Japan Society of Civil Engineers. Recommendations for upgrading of concrete structures with use of continuous fiber sheets. Japan: Concrete Engineering Series 41; [4] Rosenboom OA, Rizkalla S. Fatigue behavior of prestressed concrete bridge girders strengthened with various CFRP systems, ASCE J Compos Constr, accepted for publication. [5] Oehlers DJ. Reinforced concrete beams with plates glued to their soffits. J Struct Eng 1990;118(1). [6] Teng JG, Chen JF, Smith ST, Lam L. FRP strengthened RC structures. London: Wiley; [7] Oehlers DJ, Seracino R. Design of FRP and steel plated RC structures. London: Elsevier Press; [8] De Lorenzis L, Nanni A. Characterization of NSM FRP rods as nearsurface mounted reinforcement. J Compos Constr 2001;5(2). [9] Hassan T, Rizkalla S. Investigation of bond in concrete structures strengthened with near surface mounted CFRP strips. ASCE J Compos Constr 2003;7(3). [10] Hassan T, Rizkalla S. Bond mechanism of NSM FRP bars for flexural strengthening of concrete structures. ACI Struct J 2004;101(6). [11] Schiebel S, Parretti R, Nanni A. Repair and strengthening of impacted PC girders on bridge A4845. Missouri Department of Transportation Report RDT01-017/RI [12] Takács PF, Kanstad T. Strengthening Prestressed Concrete beams with carbon fiber reinforced polymer plates. Norway: NTNU Report R-9-00, 2002.

13 776 O. Rosenboom et al. / Construction and Building Materials 21 (2007) [13] Hassan T, Rizkalla S. Flexural strengthening of prestressed bridge slabs with FRP systems. PCI J 2002;47(1). [14] Collins MP, Mitchell D. Prestressed concrete structures. New Jersey: Prentice Hall; [15] MacGregor JG. Reinforced concrete: mechanics and design. New Jersey: Prentice Hall; [16] Bentz EC. Section Analysis of Reinforced Concrete Members, Ph.D. Thesis, University of Toronto, [17] 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).