FRP-strengthened RC Slabs Anchored With FRP Anchors

Engineering Structures 33 (2011) 1075–1087Contents lists available at ScienceDirect Engineering Structures journal homepage: www.elsevier.com/locate/engstruct FRP-strengthened RC slabs anchored with FRP anchors Scott T. Smith a,∗ , Shenghua Hu a , Seo Jin Kim a , Rudolf Seracino b a Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China b Department of Civil, Construction, and Environmental Engineering, North Carolina State University, USA article info Article history: Received 15 June 2010 Received in revised form 28 October 2010 Accepted 4 November 2010 Available online 3 February 2011 Keywords: Debonding Deformability Fibre-reinforced polymer composites Flexural strengthening FRP anchors Slabs abstract An abundance of tests over the last two decades has shown the bending capacity of flexural members such as reinforced concrete (RC) beams and slabs to be enhanced by the bonding of fibre-reinforced polymer (FRP) composites to their tension face. The propensity of the FRP to debond, however, limits its effectiveness. Different types of anchorages have therefore been investigated in order to delay or even prevent debonding. The so-called FRP anchor, which is made from rolled fibre sheets or bundles of lose fibres, is particularly suitable for anchoring FRP composites to a variety of structural element shapes. Studies that assess the effectiveness of FRP anchors in anchoring FRP strengthening in flexural members is, however, limited. This paper in turn reports a series of tests on one-way spanning simply supported RC slabs which have been strengthened in flexure with tension face bonded FRP composites and anchored with different arrangements of FRP anchors. The load–deflection responses of all slab tests are plotted, in addition to selected strain results. The behaviours of the specimens including the failure modes are also discussed. The greatest enhancement in load and deflection experienced by the six slabs strengthened with FRP plates and anchored with FRP anchors was 30% and 110%, respectively, over the unanchored FRP-strengthened control slab. The paper also discusses the strategic placement of FRP anchors for optimal strength and deflection enhancement in FRP-strengthened RC slabs. © 2010 Elsevier Ltd. All rights reserved. 1. Introduction Numerous experimental investigations have proven the ability of fibre-reinforced polymer (FRP) composites to increase the flexural capacities of beams and slabs when bonded to their tension faces [1,2]. Numerous studies have also observed the FRP to debond at strains well below its rupture strain. Such premature failure, which has been observed to initiate at the base of flexural and flexural-shear cracks along the length of the member (e.g. IC debonding, [2]) or at the FRP plate end (e.g. concrete cover separation, [2]), can occur in a relatively sudden manner and constitutes an under-utilisation of the strength and strain capacity of the FRP. Mechanical anchorage of the FRP offers a real solution to the debonding problem and several different systems have been trialed to date. They include, but are not limited to, embedded metal threads [3], nailed plates (also known as hybrid bonding [4]), U-jackets [5], near-surface mounted rods [6], and anchors made with FRP [7] (also known as spike anchors but herein referred to as FRP anchors or anchors). FRP anchors are versatile as they are non-corrosive and can be applied to wide dimensioned elements such as slabs and walls. A recent review of FRP anchors is provided ∗ Corresponding author. Tel.: +852 2241 5699; fax: +852 2559 5337. E-mail address: [email protected] (S.T. Smith). 0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2010.11.018 in [7] while a review of other anchorage methods (including FRP anchors) is presented in [8]. The anchorage of steel strengthening plates using metallic bolts is a related field of research (e.g. [9]), however, it is outside the scope of this paper and is therefore not considered further. Fig. 1. is a schematic representation of the face of a concrete member which has been strengthened with an externally bonded FRP plate and anchored with an FRP anchor. Such an anchor is essentially made from glass or carbon fibres in which fibre sheets are folded or rolled, or lose fibres are bundled together. One end of the anchor (herein anchor dowel) is inserted into an epoxy filled hole in the concrete substrate (Fig. 1(b)) and the other end of the anchor is passed through the externally bonded FRP strengthening plate (herein FRP plate or plate). The free ends of the fibres (herein anchor fan) are splayed and epoxied onto the surface of the plate in order to disperse local stress concentrations. The double anchor fan arrangement (herein bow-tie) shown in Fig. 1 has been tailor made for the test slabs reported herein. As a precursor to the bow-tie anchor fan form, Smith [10] reported FRP anchors with a single fan component to increase the shear strength and slip capacity of FRP-to-concrete joints by up to approximately 70% and 800%, respectively, over unanchored control joints. The relative difference between the strength and the behaviour of single fan and bow-tie anchors in FRP-to-concrete joint tests has also been summarised in [10]. While Smith [10] reported both types of the deflection at failure for both anchored slabs was 76% of the control. namely (i) first cracking of the tensile concrete. Such an extensive slip capacity is a desirable feature of an FRP anchor especially when large slips are expected between the FRP strengthening and concrete substrate in structural members. The low tensile strength of the GFRP (i. The anchors were not successful in delaying the occurrence of concrete cover separation and as a result they did not enhance load or deformation capacity of the beam. [24] strengthened ten beams of varying soffit curvature with carbon FRP (CFRP) tension face plates. holes were drilled through the initially bonded (and cured) FRP plate at regular intervals along the whole length of the plate and then metal bolts were inserted.18]. (b) cut-away view. [7. FRP anchor and plate: (a) overall view. Micelli et al. and the failure mode was IC debonding. The span of the beams was 6 m. The generic response consisted of several key features.11–14]) and more work is clearly required. Brunckhorst et al. there is no rational methodology for the design and placement of the anchors. [25] did not consider FRP anchorage. Two of the strengthened slabs were anchored with FRP anchors positioned 150 mm and 300 mm from the fixed end.e. The second anchored slab test utilised an extra layer of GFRP. Lam and Teng [17] then reported an additional five RC cantilever slabs tests of 700 mm span in which fours slabs were strengthened in flexure with wet lay-up GFRP and anchored with FRP anchors positioned in the same locations as Teng et al. In both cases. [25] presented a diagram of a generic moment–displacement (analogous to load–deflection) response of an RC beam strengthened in flexure with a tension face bonded CFRP pultruded plate comprising of multi-directional fibres. and (v) residual strength provided by the bolts (after complete plate debonding). FRP anchors were generally shown to be effective in enhancing the strength and deformability of the strengthened members. 428 MPa) made it susceptible to rupture failure.g. In the first case. however. Such work is.1076 S. Ten of these beams were strengthened in flexure with tension face GFRP plates formed in a wet lay-up manner.2 m. The beams were susceptible to concrete cover separation failure. the GFRP plate ruptured after the debonding crack had propagated to the second anchor. The anchored beam appeared to fail by complete debonding of the FRP followed by anchor rupture. however. slabs [17. The unanchored slabs were found to fail by IC debonding with debonding initiating at the fixed end of the slab. In some cases. the effectiveness of the FRP anchor in enhancing load and deflection had been proven. Such is the motivation for the experimental program reported in this . Another benefit of the bow-tie anchor alternative is that slip may be in the other direction for members where the applied load can move. (iii) a sharp drop of moment upon initiation of debonding. Smith et al. a clear understanding surprisingly still does not exist about the exact role the FRP anchors play when used in structural members. [23] reported seven cantilever RC slabs tests of 700 mm span of which six slabs were strengthened in flexure with glass FRP (GFRP) composites formed in a wet lay-up manner. debonding was halted by the first anchor and in other cases no debonding was observed. so two FRP anchors were positioned at 500 mm centres at the end of one beam specimen. the FRP anchors were generally not the focus of these studies. Also. Eshwar et al. however. In light of the overall success of FRP anchors in delaying or suppressing IC debonding failures. The strengthened beams ultimately failed by IC debonding after which the behaviour of the beams resorted to that of the unstrengthened (and unanchored) control beam. the length of FRP was 5. the FRP anchors were observed to reduce the rate of debonding crack propagation. two were strengthened with identical configurations of wet lay-up FRP and one of these beams was additionally installed with FRP anchors at 500 mm centres. and concrete and masonry walls [21. anchors to exhibit similar load–slip characteristics over most of the responses.T. respectively. Teng et al. Oh and Sim [15] reported tests on eleven simply supported beams each of 2 m span. Of the three beams with greatest curvatures. Brief reviews of some of the literature of FRP-anchored FRP flexurally strengthened RC beams and slabs are provided as follows. Also. their research is still applicable and is therefore reviewed here. The plate was also anchored with regularly spaced metal screw-bolts.2 m spanning beams to increase the load carrying capacity of the FRP-strengthened beams by 13% above the strengthened but unanchored control beams. 1. Here the minimum criterion is to position the anchor fan on the side of the anchor in the direction of load.’s [23] test slabs. The majority of the research conducted to date on FRP anchors has been on the anchorage of flexurally strengthened RC beams [15. More recently. slab–column connections [19]. the FRP was not observed to fail and as a result the limits of the anchors were not established. In such research. The two anchored slabs experienced a 24% and 61% increase over the unanchored but strengthened control slab respectively. (iv) residual strength (above the plain unstrengthened RC beam) provided by the remaining bonded FRP. outside the scope of this paper. (ii) initiation of debonding of the FRP plate via ‘gliding fracture’ (this translation appears to be consistent with IC debonding). The main test variables were preloading as well as internal tension steel ratio and position. however. / Engineering Structures 33 (2011) 1075–1087 Fig. While Brunckhorst et al. in many cases. In order to install the bolts. [16] showed FRP anchors spaced at 250 mm centres in 2. the bow-tie anchors were ultimately able to resist much greater slips (while maintaining limited strength) before failure.16]. confined columns [20].22]. There has been limited research though conducted to date on characterising the fundamental behaviour of FRP anchors (e. The increase in strength and mid-span deflection of this anchored beam to its unanchored counterpart was 34% and 74%. In all strengthened slab tests the FRP was observed to rupture. In both anchored slab cases the slope of the load–deflection curve clearly decreased as debonding propagated. In this case the anchors failed after the debonding crack had propagated along the plate. . The type and positioning of the FRP anchorage were the key variables in this study and the following comments are offered in support of the layouts presented in Fig. In addition. One slab served as an unstrengthened and unanchored control while the remaining seven slabs were strengthened in flexure with CFRP (herein FRP) formed in a wet lay-up manner. LVDT. Type 1 anchors contained twice the amount of fibre as Type 2 anchors. 3). 2(a). / Engineering Structures 33 (2011) 1075–1087 (a) Elevation. Table 1 provides a summary of the key variables for all the eight slabs. CMR constant moment region}. The remaining six slabs were all strengthened with the same flexural strengthening arrangement as Slab S2 in addition to being anchored with different arrangements and types of FRP anchors. At the initial design stage of the project. 2. the anchorage arrangements for Slabs S3–S7 were intuitively selected.S. The anchorage scheme for Slab S3 (Fig. although. 3. Fig. the aims of the study reported herein are to (i) quantify the load and deflection enhancement that FRP anchors can provide to FRP flexurally strengthened RC slabs failing by IC debonding. Fig. no anchors were installed at cross-bar locations because (i) the embedment depth of the anchor was greater than the cover to the cross-bar. (b) FRP-strengthened and unanchored { tension and cross-bar steel.33%. it was not deemed logical to position an anchor where flexural cracking is likely to occur. two were tested as control slabs S1 and S2. support line. Smith et al.. 3 and 4 provide a summary of the FRP flexural strengthening and FRP anchorage arrangements for control Slabs S1 and S2. 2(b). The slabs were reinforced in flexure with two 10 mm diameter hot-rolled steel reinforcing bars positioned at an effective depth of 120 mm as shown in Fig. The anchorage arrangement for Slab S8 was decided upon after observing the test results of Slabs S3–S7. Slab S2 was strengthened in flexure with three layers of carbon fibre sheet in a wet lay-up procedure but not anchored. 4. and (iii) manipulate the load–deflection response in beneficial ways with the strategic use of FRP anchors. loading and instrumentation details { 90 mm strain gauge. Experimental setup 2. Control slab details (tension face) and strain gauge layout: (a) unstrengthened and unanchored. (ii) observe the behaviour and failure modes of the anchored slabs. and (ii) as cross-bars can act as crack initiators. To achieve these objectives.- Of the eight identical slabs cast. 2.T. Slab S7 was anchored with Type 2 anchors and Slab S8 was anchored with a combination of Type 1 and Type 2 anchors. Slabs S3–S6 were anchored with Type 1 anchors. (b) S2. 4(a)) entailed the anchor dowel being positioned mid-distance between the cross-bars along the . All slabs were rectangular in crosssection of nominally 150 mm depth and 400 mm width with a clear span of 2400 mm as shown in Fig. eight simply supported one-way spanning RC slabs were constructed and tested to failure. (a) S1. The same steel reinforcement type was also placed on the top of the longitudinal reinforcement at 200 mm centres as cross-bars (Fig. 1077 (b) Section. and anchored Slabs S3–S8. RC slab and geometry. 10 mm strain gauge}. More specifically. respectively. Slab S1 was a plain RC slab which contained no FRP flexural strengthening and no FRP anchorage. Details of test slabs The experimental program consisted of eight simply supported one-way spanning RC slab tests. Six of these seven strengthened slabs were anchored with different FRP anchor types and layouts. Fig. The span-to-effective depth ratio of each slab was 20 and the steel reinforcement ratio was 0.1. All slabs were cast from the same batch of ready-mix concrete. both anchors were made by hand in an identical manner. paper. 2(b). Figs. . (d) S6. (c) S5.cross-bar (tension steel omitted). 4. (e) S7. (f) S8. . Smith et al. FRP anchor layout (tension face) and strain gauge layout { 10 mm strain gauge}.. / Engineering Structures 33 (2011) 1075–1087 (a) S3. . Type 1 FRP anchor. Fig. support line.1078 S. (b) S4.T. Type 2 FRP anchor. In order to avoid collision with the cross-bars.2. the anchors in Slab S7 were offset 50 mm to those of Slab S5. considers both load and deflection. the shear span anchors of Slab S2 were only retained. a 2. The 90 mm length was to form a 50 mm long fan with a 40 mm embedded portion which included a small allowance for the 90 degree bend portion. 4(e) Fig. 5. The results and discussions presented in this paper are therefore void of the variation inherent in experimental testing of like specimens. Type 2 FRP anchor. 5(a)). 4(c) Fig. 4(a) Fig. (b) schematic of completed FRP anchor.S. 4(c)). In preparing the anchors. All bow-tie anchors were made from 90 mm long carbon fibre sheets. Type 1 anchors were made from a 250 mm wide sheet of fibre which was twice the width as that used for Type 2 anchors. Both exterior edges of the sheet were then rolled towards the centre of the sheet until the two rolled portions b Fig. FRP anchor construction c The same roll of carbon fibre sheet of 0.166 mm nominal thickness and the same tins of two-part epoxy were used to make all the FRP anchors and all the FRP plates in this study. The following explanation is offered in support of the making of the anchors. of approximate circular crosssection. only the ends of the FRP plate were anchored. For Slab S6 (Fig. It should be noted that only one specimen was tested for each configuration of FRP strengthening and FRP anchorage. 4(b) Fig. The understanding gained from this study on FRP anchors can be extended to other RC structural elements. (c) actual FRP anchor.T. 4(c)). the fibre sheet was spread out on a flat surface and a 25 mm long region at one end of the sheet was impregnated with epoxy across the whole width of the sheet. slabs have been selected in this study in order to ensure the occurrence of IC debonding failure of the FRP flexural strengthening. The hybrid arrangement of anchors in Slab S8 utilised both Type 1 and Type 2 anchors. b whole length of the FRP flexural strengthening while in Slab S4 (Fig. In reality. met (Fig. such as beams. 4(b)) every second anchor was removed. however. Smith et al. In addition. Construction of bow-tie FRP anchors: (a) schematic of fibre sheet rolling (epoxy impregnated end fibres shown in foreground). were then inserted into preformed holes of 14 mm (for Type 1 anchors) or 10 mm diameter (for Type 2 anchors) in a . / Engineering Structures 33 (2011) 1075–1087 1079 Table 1 FRP strengthening and FRP anchorage details. The reasons for positioning the Type 1 anchors closer to the mid-span region and the Type 2 anchors closer to the free end of the FRP plate are discussed in detail in Section 3 of this paper. c Total cross-sectional area of FRP anchor crossing the FRP-to-concrete interface in one shear span (Area = cross-sectional area of the fibre sheet in single anchor dowel times number of anchors in one shear span). Slab S1 S2 S3 S4 S5 S6 S7 S8 FRP strengthening Nil 3-layers 3-layers 3-layers 3-layers 3-layers 3-layers 3-layers FRP anchor details Comments Arrangement Fibre content (mm) Shear connection fibre (mm2 )c Nil Nil Fig. The anchored regions as well as the amount of fibre used to construct the anchors in Slabs S7 (Fig. 4(f)) were identical to Slab S5 (Fig. The current experimental program. In Slab S5 (Fig. Also. 1 and 5. 4(d) Fig. The method of manufacturing the FRP anchors was essentially the same as that reported in detail in [26]. The rolled sheets. The difference though was the use of a single anchor fan in [26] and a bow-tie arrangement in the present study. 4(e)) and S8 (Fig. 4(d)). all anchors were made from the same bow-tie form as shown in Figs. The Type 2 anchors used in Slab S7 were double in number and half the spacing as the Type 1 anchors used in Slab S5. Such a debonding failure mode is most common in FRP-strengthened RC slabs and can provide a useful framework for evaluation of the effectiveness of FRP anchors. serviceability rather than strength commonly governs the design of slabs. 4(f) Nil Nil 250a 250 250 250 125b 125 and 250 Nil Nil 166 83 166 83 166 166 Unstrengthened control Unanchored control Type 1 anchor Type 1 anchor Type 1 anchor Type 1 anchor Type 2 anchor Hybrid Type 1 and Type 2 anchors a Type 1 FRP anchor. 2. 3. Curve the dry fan fibres of the anchor (without kinking). 2(a) (only the results at mid-span are reported in this paper.26]). splay into the bow-tie form. Flush anchor dowel holes with compressed air for cleaning and then inject epoxy. Recent work by this research group [26] has identified the importance of not impregnating the anchor fibres in the bend region with epoxy upon installation of the FRP anchor. Smith et al. and then epoxy onto the outer surface of the outermost plate layer. Such large slippage will enable greater deflection of the FRPstrengthened and FRP-anchored slabs considered herein and help prevent brittle FRP anchor rupture failure. Fig. The mass of the spreader beam (170 kg) was included in the load. FRP strengthening may need to be applied to damaged or deteriorated RC members containing corroded reinforcement. Fig. 4.g. 5. 2. three layers of fibre sheet were used to form the FRP plate in this study). 2.T. Application of FRP strengthening and FRP anchors The FRP plates and FRP anchors were installed in accordance with the procedures described in [26]. exposes aggregate and also roughens the concrete surface (Fig. The specimens considered in this study are undamaged specimens. In reality though. It has been found to be successful on numerous occasions by the first author of this paper (e. The mechanical properties of the concrete. however. it is not anticipated that the behaviour of the FRP-strengthened system in question would significantly change. Extensive arrays of electric strain gauges of 10 mm gauge length were mounted onto the FRP plates for Slabs S2–S8 as shown in Figs. Such results are also not reported herein. 6(a)). also. Prepare surface of concrete to be strengthened using a pneumatic needle scaler. Dry fibres in the bend region will enable slips of the plate relative to the concrete surface to be achieved that are generally greater than 5 mm. This method of surface preparation removes the top surface of weakened concrete.28]). it would be necessary to repair and prepare the concrete substrate onto which the FRP system will be applied according to the FRP manufacturer’s recommendations and design guidelines (e. [27. (b) Wet lay-up application of fibre sheet and anchor fans. The following summary is. Instrumentation and test procedure Two electric strain gauges of 90 mm gauge length were mounted onto the top compressive face of each of the eight RC slabs at mid-span as shown in Fig. 3 and 4. 6. support deflections were negligible). FRP anchor and plate installation. / Engineering Structures 33 (2011) 1075–1087 (a) Installed FRP anchor dowel. In such cases. . provided for completeness: 1. Material properties All the eight slabs were poured in one batch and tested over a ten week period.3. polystyrene mould which was pre-filled with epoxy in order to properly form the anchor dowel.5. The anchors were then removed from the mould once the epoxy had cured for at least one day. Fig. 7. or spalled concrete cover. cracked concrete. Fig. In this situation.g. 6(b)) (i. suitable comments are provided in Section 3 of this paper. the self-weight of the slab was not. 2. Allow the plate and anchor fan fibres to cure for a period of 7 days prior to testing. however. 6(a)) and allow to cure for at least half a day. In the case of this experimental program.4. linear variable displacement transducers (LVDTs) were positioned along the length of the slab as well as above the supports in order to record the vertical deflection as shown in Fig. 7 shows a typical test in progress. 2. strain and LVDT readings.1080 S. 5(b) and (c) show completed anchors. In addition. 6. Test setup.e. however. Drill anchor dowel holes into concrete slabs at appropriate locations. All slabs were tested in a stiff reaction frame and load was applied through a 1000 kN capacity servo-controlled actuator by displacing the ram of the actuator at a rate of 1 mm/min. insert the preformed anchor dowel (Fig. all slabs were left in a controlled laboratory environment. LVDTs were also placed at each end of the FRP and were reacted off the adjacent slab soffit in order to measure plate end slip for Slabs S5–S8. Then. [7. Slip parted carbon fibre sheet fibres over the anchor fan fibres and then form the plate from the sheet fibres in a wet lay-up manner (Fig. 8(a). ER = modulus of rupture. Experimental results 3. are reported in Table 2. Slab S8 offers the highest increase in load and Slab S7 offers the highest increase in central deflection. strengthening as well as the influence of the FRP anchors. which were determined in accordance with the BS 1881 suite of standards [29–32]. in addition to the age of the slabs upon testing.7 28 389 3. fct splitting strength. The two main turning points.8 GPa). The mechanical properties of the epoxy were tested in accordance with the 527 series of documents published by BS EN ISO 527:1996 [35]. The results of seven specimens were averaged to produce an elongation at rupture of 6716 µε (0. were predominantly of a trilinear nature [36].1. Load–deflection responses and load–deflection–strain summary The total load (P) versus mid-span deflection responses for all the eight slab tests are shown in Fig. Most slabs then sustained a post-peak reserve of strength.8 29 234 4. = 391 µε). The dramatic drop in load upon the peak load being reached was due to complete debonding of the FRP plate on one side of the slab.9 a b fcu = cube compressive strength. / Engineering Structures 33 (2011) 1075–1087 1081 Table 2 Concrete mechanical properties.2% proof stress) and elastic modulus of the steel reinforcing bars. = 1. = 402 µε). 8. are due to the concrete cracking and yield of the internal tension steel reinforcement.5%) (sd. The response of the third portion was a direct consequence of the FRP Fig. are also reported in Table 2.1 S4 S5 74 79 77 55. The yield stress (0. Load–deflection responses of anchored slabs relative to control Slabs S1 and S2: (a) all anchored slabs. In order to further enhance the clarity of these results. Table 3 provides a summary of the enhancement in peak load and corresponding peak deflection of all slab specimens as well as the maximum measured compressive strain in the concrete and measured tensile strain on the FRP.1 S6 S7 S8 92 93 109 94 56. Table 3 also provides a summary of the post-peak results which refer to the behaviour of the slabs. prior to complete debonding of the FRP strengthening. tensile strength at rupture of 3163 MPa (sd. = 206 MPa) and elastic modulus of 239 GPa (sd.4 MPa) and elastic modulus of 4273 MPa (sd. 8(b) while the slabs which experienced a predominant increase in deflection only are shown in Fig.3R-04 [34] to produce an elongation at rupture of 14.8 GPa) respectively. = 5 MPa) and 198 GPa (sd. The 10% increase in concrete compressive strength over the duration of the experimental program was expected to have a minor influence on the test results as none of the test slabs failed in compression. This reserve was due to frictional resistance from the intact anchors clamping the debonded plate to the rough concrete substrate failure plane. Smith et al.1 29 299 4. Two layers of fibre sheet were used to form flat FRP coupons of 30 mm width. = 126 MPa). tensile strength at rupture of 28. Slab Age (days) Propertiesa Age (days) fcu (MPa) Ec (MPa) fct (MPa) ER (MPa) S1 S2 S3 31 43 44 36 51. the slabs which experienced an increase in both load and deflection are shown in Fig. Five coupons were prepared and then tested to failure in accordance with ACI 440. (b) anchored slabs with increasing strength and deflection (c) anchored slabs with predominantly increasing deflection.3 MPa (sd. 3. were found to be 566 MPa (standard deviation. which were averaged from three specimens tested in accordance with BS EN 1000201:2001 [33].T.2b 5. The load–deflection responses of Slabs S2–S8.2 6. = 3. .3 4.7%) (sd. sd. = 6. The age of the concrete when the mechanical property tests were conducted. 8(c). which define the ends of the first two linear portions.674 µε (1. Ec = elastic modulus. once the large and sudden drop in load had occurred due to complete plate debonding. The influence on strength and deflection of the FRP-strengthened RC slabs by the FRP anchorage can be clearly observed. Based on one test result.S. e. Slab S1 S2 S3 S4 S5 S6 S7 S8 Load (peak) Deflection (peak) Load (post-peak) Peak.e. The gradient of the third portion of the load–deflection response then started to decrease as the debonding crack propagated. In addition. (ii) partial rupture at the bend region.59 51. 8). 8 and 9 and Table 3. 3. [38] (using their best-fit factor of 0. The stepped descent of the load–deflection response was believed to be due to the residual restraint offered by the intact FRP anchors. More specifically. Slab S7) upon the addition of FRP anchors.38]. the spacing between the anchors in Slab S4 was twice that over Slab S3. The level of utilisation of the tensile strain capacity of the FRP was 45% for the unanchored control slab (Slab S2). The slab then sustained a post-peak reserve of strength 32% in excess of the strength of control Slab S1.65 48. Slab S4 The anchorage scheme of Slab S4 involved the removal of every second anchor from the Slab S3 anchorage scheme. This translated into Slab S4 having 55% of the number of anchors as Slab S3.90 26. the corresponding strain in the tension steel at concrete crushing was 4. 3. namely (i) complete rupture at the bend region.1. Smith et al.4. first cracking occurred at a load of 13 kN with tension steel reinforcement yielding at a load of 34 kN. The FRP plate increased the load capacity of the slab. P (kN) P /PS1 a (%) P /PS2 b (%) Peak. The anchors were found to fail in one of three different modes. compared to Slab S1.g. 3. concrete strain to reach 3000 µε ) at 95 mm of mid-span deflection. .2. c % of flat coupon capacity (flat coupon capacity = 14.47 19. δ (mm) δ/δS2 b (%) Peak. 8. (P − Ps1 )/Ps1 × 100%. Upon complete plate debonding. This level was increased by up to 79% (i. by 105%. dependent on the number of anchors) with the stiffer slabs experiencing slightly higher cracking and steel yielding loads. decreased the rate of propagation of debonding compared to Slab S2.2.78 22. and (iii) pull-out from the concrete substrate. (P − Ps2 )/Ps2 × 100%.3%) 8 025 (54. Theoretical calculations [36] estimated the concrete to crush (i. The post-peak reserve of strength was increased by 17% in excess of the capacity of Slab S1. As it served no purpose in this study to then displace the slabs until the concrete crushed. 9 provides a schematic representation of the debonded regions for all anchored slab tests in addition to the condition of the anchors immediately following complete FRP plate debonding. Slab S1 The unanchored–unstrengthened control Slab S1 behaved in a manner befitting that of an under-reinforced ductile member. the central deflection had increased by 46% above Slab S2. 10(b) shows a pulled-out anchor. The relative enhancement in stiffness was a function of the amount of fan fibre (i. The three anchors adjacent to the debonded plate end had partially ruptured (Fig. Ppp (kN) 20.6%) 11 566 (78.2% which fits within the ACI 318M-05 [37] definition of a tension-controlled failure (i. 3.3%) 7 676 (52.66 51.58 37. Behaviour and failure modes Fig. the load dropped down to the capacity of the plain control Slab S1 and at this stage no anchors had failed (Fig. This strain is identical to the predicted debonding strain of 6650 µε from Teng et al. In addition. 10(a) shows the condition of a typical completely ruptured anchor while Fig.27 0 105 152 116 155 100 153 167 NA NA 25. 3.T. Slab S3 Localised debonding of the FRP initiated at a load of approximately 39 kN (upon observation of the first drop in the load–deflection response) at one of the applied load positions once the third linear portion of the load–deflection response had been reached (Fig.2.7%) 8 884 (60. however. The maximum measured tensile strain on the FRP was about 6650 µε (Table 3) upon complete debonding of the FRP.e. In the Slab S2 test. Fig.32 41. The effect of such an arrangement of anchors in Slab S4 reduced the slope of the third portion of the load–deflection response to virtually zero once the load and deflection corresponding to failure of the control Slab S2 was reached.e. All anchored slabs (i. d compressive (−’ve) strain.2. the load suddenly decreased although at this stage none of the anchors were found to have completely failed.2R-08 [28].S1 a (%) 0 3 32 17 29 11 −3 10 εconc (µε) (concrete)d εfrp (%c ) (µε) (FRP) −1872 −1237 −1566 −1543 −1300 −1141 −2564 −1528 NA 6 649 (45. the FRP was no longer effective.78 53. 9(b)).32 20.1082 S.e. a % increase over unstrengthened control (S1).82 26.99 31.674 µε).3%) NA = not applicable. b % increase over unanchored control (S2).3. First cracking of the tensile concrete occurred at a load of 11 kN and then the internal tension steel yielded at a load of 19 kN. ductile section).35 0 23 5 24 −3 24 30 Strain (peak) Ppp /Ppp. The plate then completely debonded at a load of 23% and deflection of 63% in excess of the respective load and deflection at debonding in Slab S2.2. The FRP anchors.87 23. Slabs S3 to S8) except Slab S2 were on the whole stiffer on account of the anchor fan fibres increasing the thickness of the FRP plate.2.22 43. 9(a)) while the fourth anchor from the plate end had pulled-out due to a major crack passing right through the anchor dowel region (Fig. e.753) and close to 7650 µε as calculated from ACI 440. The maximum midspan deflection of all anchored slabs prior to complete FRP plate debonding was about 54 mm. When the debonding crack had propagated to the plate end. and Slab S2 then behaved in a very similar manner to the control Slab S1. e. The following comments are offered in support of the behaviour and failure of all test slabs in the context of the results contained in Figs. / Engineering Structures 33 (2011) 1075–1087 Table 3 Summary of load and deflection (mid-span) enhancement and peak strains.5%) 6 696 (45. 9(b).2. The slab then deflected extensively until the test was stopped at 80 mm of mid-span deflection.47 54.g.20 22. the loading was stopped and released once a mid-span deflection of 80 mm had been reached.34 41.80 40. Upon complete debonding. At this stage the strain in the compressive concrete was on average about 1600 µε and well below its crushing strain (except Slab S7 due to high deflection). Slab S2 The flexurally strengthened but unanchored control Slab S2 failed by IC debonding of the FRP plate. 10(b)). debonding first initiated at a load of 40 kN (deflection 23 mm) as observed from the large drop in the load in Fig.53 41. Once the plate completely debonded. Such a failure mode has been well documented in the open literature [1.76 NA 0 63 46 64 24 110 91 20.8%) 11 348 (77. The shear span anchors in the debonded plate portion were observed to be partially ruptured when the load was stopped at a mid-span deflection of 80 mm as shown in Fig.90 51. 10. / Engineering Structures 33 (2011) 1075–1087 1083 Fig. Typical FRP anchor failure modes: (a) completely ruptured anchor (Slab S8. As a result. 4). Both slabs therefore had exactly the same number and same type of anchors in the shear span.5. { debonding crack. Smith et al. 3. 9. anchor no. Slab S5 followed virtually the same load–deflection response as Slab S3. a b Fig. completely ruptured anchor. The peak load and deflection of Slab S5 was about 1% to that of Slab S3.. 1). Type 2 FRP anchor. partially ruptured anchor. The similarity in slab behaviour between the anchorage layouts for Slabs S3 and S5 suggests that anchors located in the shear span are the most effective. pulled-out anchor. FRP plate and FRP anchor conditions post-plate debonding.load point. The large release of energy . Slab S5 Slabs S5 and S3 were identical apart from the omission of three anchors in the constant moment region in the former. (b) pulled-out anchor (Slab S3. CMR = constant moment region. Type 1 FRP anchor. .2.T. anchor no. undamaged anchor}..S. however. the former was on the whole slightly less stiff than the latter due to the difference of anchors in the mid-span region. The effect of anchorage in the constant moment region can be speculated to have enabled the stepped unloading response observed in Slab S3. the crack then rapidly propagated to the inner anchor. Smith et al. contributed to strength gains at greater deflections (i. The closer spaced anchors of Slab S7. / Engineering Structures 33 (2011) 1075–1087 a b c Fig. Once the debonding crack had become unstable. Slab S6 then sustained a small post-peak reserve strength of at most 11% above Slab S1. upon which the debonding crack passed the FRP anchors and the plate subsequently completely debonded.2. 9(d)). Upon complete plate debonding. associated with complete debonding of the plate (and the lack of a stepped descent) caused the four anchors in the debonded plate end to partially rupture (Fig. Slabs S3 and S5 demonstrated the effective strength gains caused by Type 1 anchors positioned in the shear span region. Finally. however. Just prior to the sharp drop in load in the load–deflection response of Slab S6 at 40 kN. although the closer spaced anchors were successful in slowing down the propagation of the debonding crack enough for a 28% enhancement in central deflection to be achieved over Slab S5. This was a direct result of not impregnating the bend region fibres with epoxy as discussed in Section 2. The anchor at the debonded plate end completely ruptured. This was due to the Type 2 anchors of Slab S5 containing less anchor fan fibre than the Type 1 anchors of Slab S5. the plate had been progressively debonding from the concrete substrate in a similar manner to that of Slab S2. a softer load–deflection response) over Slabs S3 and S5. 3. Slab S8 Slab S8 was anchored with a combination of anchor types (i. The resulting arrangement of anchors in Slab S8 produced the early strength gains in the third linear portion of the load–deflection response associated with Type 1 anchors near the peak moment . (b) Type 2 anchors (Slab S7).1084 S. the sudden release of energy caused all the anchors in the critical shear span to completely rupture as shown in Figs. the partially ruptured anchors and the remaining intact anchors were able to contribute to the 29% post-peak reserve in strength over Slab S1. Unfortunately the LVDT measuring plate end slip at the debonded plate end detached due to the large release of energy upon plate debonding.2. 3. 11(b) also shows that debonding occurred at the FRP-to-concrete interface in the concrete. The load was then steadily increased again until the capacity of Slab S2 was reached approximately. The reduced amount of fibre used in the Type 2 anchors caused them all to rupture and the failure of Slab S7 was the most catastrophic one out of all the anchored slab tests reported herein.8. The results of Slab S7 showed the effectiveness of closer spaced anchors in enhancing deflection.e. 3.T. (c) Hybrid Type 1 and Type 2 anchors (Slab S8).2. Slab S7 therefore contained anchorage spread over the same shear span region as Slab S5. the second anchor from the plate end had partially ruptured (Fig. all FRP-strengthened slabs (S2–S8) failed at the same desired FRP-to-concrete interface in the concrete. 11(a) shows the test specimen immediately after plate debonding. No post-peak reserve of strength was able to be maintained on account of all the anchors failing. The ability of the anchors to cater for such plate slip was due to the ability of the fibres in the bend region to deform. The anchorage was designed after observing the behaviour of all preceding anchored slab tests S3–S7. Typical FRP anchor conditions after complete plate debonding: (a) Type 1 anchors (Slab S5). Slabs S6 and S2 were identical apart from the end anchorages in the former.3. Measurements made during the test with a hand-held scale revealed the slip at the plate immediately post-plate debonding to be in excess of 5 mm. Interestingly.7. Slab S7 The logic behind this anchorage scheme was to have anchors spaced closer together in a bid to further slow down the rate of propagation of the debonding crack. Type 1 and Type 2) and anchor spacings. the total amount of fibre used to make all the anchors in Slab S8 was identical to the amount of fibre used to make all the anchors in Slabs S5 and S7. however. 9(e) and 11(b). As a result. Fig. 9(c)) and Fig. Slab S6 Slab S6 contained the least amount of anchorage and the two anchors positioned at each end of the FRP proved to be least effective in enhancing the load and deflection. The anchorage scheme of Slab S7 caused the third linear portion of the load–deflection response to considerably soften compared to that of Slab S5. After all. The load–deflection response of Slab S7 was softer than Slab S5. however.e. the results of Slabs S4 and S6 showed that anchors spaced too far apart did not lead to significant strength gains but did lead to deflection gains. 11. Finally. In fact. the load prior to complete debonding of the FRP plate was virtually identical to the peak load of Slab S5.6. the number of anchors in Slab S7 was doubled and the amount of fibre used to make each anchor was half when compared with Slab S5. thus accounting for the drop of load at a deflection of 25 mm. e.3. partially ruptured anchor. it is based on visual observation. Debonding crack propagation The presence of the anchors influenced the rate of propagation of debonding cracks along the length of the FRP plate. (iii) debonding . the rate of crack propagation for Slab S7 was much reduced than that of Slab S6. { debonding Type 1 FRP anchor. it is important to stress the limitations of presenting such data. crack. Overall comments to be made include (i) debonding initiates well before complete plate debonding (i. This slab is singled out as it is considered one of the most optimally designed (i.. Regardless. Fig. (b) debonding cracks at specific level of load and deflection. The strength enhancements for Slabs S5 and S7 were. greatest strength enhancement. one edge of the FRP plate was monitored with strain gauges. S8).. Upon complete debonding of the plate. Type 2 FRP anchor. 12. all the Type 2 anchors ruptured (Figs. completely ruptured anchor.e. region (Fig. The increase in deflection of about 16% above that observed in Slabs S3 and S5 proved the effectiveness of the close spaced Type 2 anchors nearer to the ends of the FRP plate. virtually identical (see column 4 of Table 3). For example. CMR = constant moment region. partially intact anchorage post-plate debonding).T. the closer spaced anchors in Slab S7 were found to increase the deflection of the slab over the larger spaced anchors in Slab S5. However.S. 30% above that of Slab S2) and the second greatest mid-span deflection. . Smith et al. Slab S8 achieved the greatest enhancement in strength (i. undamaged anchor}. Propagation of debonding cracks (Slab S8): (a) load–deflection response (Slabs S1. As a result. Slab S7 was able to achieve much greater enhancement in strength and deflection than Slab S6. debonding cracks were found to propagate more slowly for anchors spaced more closely together. 9(f) and 11(c)) however the partially and fully intact Type 1 anchors contributed to the post-peak reserve of strength gain of 10%. 12 provides a detailed account of the propagation of debonding cracks in relation to the applied load (and deflection) for Slab S8. The significance of this second point is that the extent of debonding through the width of the plate may not be uniform. In Fig. / Engineering Structures 33 (2011) 1075–1087 1085 a b Fig. Fig. The enhancement in strength in this case was also due to the use of more FRP anchors. however. S2. second greatest mid-span deflection. adequate warning of failure is provided). 12 provides a qualitative overview of the debonding crack propagation process. 3.e. First. Overall.load position. the extent of debonding in accordance with the level of load and deflection are related to each other and the fifteen data points correspond to observed propagation of the debonding cracks. (ii) debonding initiates near the loaded region and propagates towards the free ends of the FRP plate. Second. 12. When the same total amount of fibre was used to cross the shear plane. 8). 13. 13(b)) due to the clamping effect of the FRP anchor which sustains interaction along the debonding crack by aggregate interlock and friction. Smith et al. the distribution of strain along the FRP for the same Slab S8 is provided in Fig. partial debond. to near final debonding as well as corresponding drawings showing the extent of debonding cracks.T. however. Anchorage of the FRP strengthening plate with FRP anchors can build robustness into the member. 13. Strain gauge results are notoriously difficult to interpret especially once the concrete has cracked and the FRP has debonded. The following observations are. 12. the strains at load levels of 10–40 kN are provided in addition to the strains at the 15 different debonding crack levels identified in Fig. 3. the internal tension steel has also yielded. Conclusions The effectiveness of FRP anchors in increasing the strength and deflection of FRP flexurally strengthened RC slabs has been reported in this paper. The experimental results have revealed the following issues of importance. 1. Also. The greatest increase in strength and deflection recorded. was 30% and 110%. The strain profile does not completely flatten though when a large proportion of the FRP has debonded (i. Fig. Ample warning of distress is thus provided well before complete debonding of the FRP. over the unanchored but strengthened control counterparts. In addition. in the anchored regions propagates in lengths approximately equal to the anchor spacing. respectively. The FRP anchor fan influences the strain results (the strain in the anchored regions is less due to the use of more FRP material). FRP strain distributions for Slab S8: (a) complete load range. 4.1086 S. the usable strain in the FRP plates was increased from 45% of the capacity of a flat coupon (for the unanchored but strengthened control slab) to almost 80% (for optimally designed anchorage schemes). In Fig. 12. made in light of Fig. During debonding. (b) selected load levels (no bond. the strain profile becomes flatter as the debonding cracks propagate to the ends of the plate. debond crack 14 in Fig. and (iv) complete debonding of the plate signifies a complete loss of load carrying capability of the strengthened slab although post-peak reserves of strength (above the unstrengthened slab) due to intact anchorage can be sustained. / Engineering Structures 33 (2011) 1075–1087 a b Fig. early debonding. near complete debond) {CMR = constant moment region}. Further inspection of Fig.4. 13. 13(a). 12 leads one to the conclusion that the addition of FRP anchors introduces a considerable degree of robustness into the FRP-strengthened slabs by allowing the debonding crack growth process to be controlled. Load–strain response To enable comparison with the results contained in Fig. .e. 13(b) provides selected results from pre-debonding. [34] ACI 440. 2007 [in Danish]. 2005. Tian Y. Seismic strengthening of rectangular reinforced concrete columns using fiber reinforced polymers.29(9): 2158–71. Building code requirements for structural concrete and commentary. 2008. J Compos Constr. UK: British Standards. Basunbul IA. [21] Antoniades KK. Smith ST. 91(2):160–8. ACI. Pecce M. 2010 [in press]. Baluch MH. . [7] Smith ST. Forstærkning af betonkonstruktioner med bolte-limede kulfiberbånd. [5] Smith ST. USA): American Concrete Institute.D. Poulson E. Binici B. Nanni A. [23] Teng JG.D. [12] Kim SJ. A case study on the restoration of flexural capacity of continuous one-way slabs with cutouts. The following individuals from The University of Hong Kong are thanked for their assistance with the experimental work: Messers Huawen Zhang (Ph.6(3):183–99. Anchorage of composite laminates in RC flexural beams.T. ASCE 2008.8(1):79–87. Guide for the design and construction of externally bonded FRP systems for strengthening concrete structures. Jiaqi Yang (M. Deflection calculation of FRP-strengthened RC flexural members. Strengthening of initially loaded reinforced concrete beams using FRP plates. Ozcebe G. 35–44. Galati D. Eng Struct 2010. 2001. Anchorage system for externally bonded FRP laminates using near surface mounted rods. Tang WC. 1087 [11] Kim SJ. Jirsa JO. Method for determination of compressive strength of concrete cubes. Pullout strength models for FRP anchors in uncracked concrete.14(4):406–14. 2009. Smith ST. [20] Ozcan O. [26] Zhang HW. APFIS 2009. [9] Su RKL. Bayrak O. Optimisation of carbon and glass FRP anchor design. and the LG laboratory technicians. J Compos Constr. Seracino R. Smith ST. Ghaleb BN. Rizzo A. Kim SJ. 3. Tensile behaviour of FRP anchors in concrete. Closer spaced anchors were found to reduce the rate of debonding crack propagation and also enabled higher deflections to be achieved. 2008.2R-08. Teng JG. FRP anchors: recent advances in research and understanding. [17] Lam L. centre for infrastructure engineering studies. [8] Ceroni F. Constr Build Mater 2003. 1999. 1983. Effectiveness of CFRP strengthening on curved soffit RC beams. Chen JF. [29] BS 1881-116. Lam L. Seracino R. USA): American Concrete Institute. Taerwe L. Evaluation of hysteretic response and strength of repaired R/C walls strengthened with FRPs. Special issue on bond behaviour of externally bonded FRP reinforcement. Salonikios TN. Adv Struct Eng 2009. J Compos Constr.32(6):1769–78. Ibell TJ. Effects of bolt–plate arrangements on steel plate strengthened reinforced concrete beams. stiffness and ductility. [27] El-Reedy M. [6] Gose SC. Assessment and repair of corrosion.12(6):845–65. Report. [37] ACI 318M-05. ASCE 2004. [24] Eshwar N. [13] Orton SL. Smith ST. UK: British Standards. Eng Struct 2008. Performance of FRP bonding systems under fatigue loading. Shear-bending interaction in debonding failures of FRPplated RC beams. References [1] Oehlers DJ. UK: British Standards. 1983. Cambridge (UK): Woodhead Publishing Limited. In: Proceedings. [33] BS EN 10002-1:2001. Anchors positioned in the shear span were found to be most effective (anchors in the constant moment region were largely ineffective). Farmington Hills (MI. En elementœr. [22] Tan KH. ACI Struct J 2010. fifth international conference on FRP composites in civil engineering. 2004. Intermediate crack induced debonding in RC beams and slabs. 2010. Patoary MKH. Jacob Schmidt (Ph. UK: British Standards. Kin Lung Jason Li (2009–10 undergraduate final year project student). Florida (USA): CRC Press.29(3):429–41. 2004. [35] BS EN ISO 527:1996. Design considerations of carbon fibre anchors. Jirsa JO. J Compos Constr. Siu WH. Kim SJ. J Compos Constr. Nanni A. USA): American Concrete Institute. ASCE 2001. Teng JG. Struct Concr. Kappos AJ. Acknowledgements Funding provided by the Hong Kong Research Grants Council in the form of General Research Fund Grant HKU 716308E is gratefully acknowledged. 13–9.S. [16] Micelli F. [31] BS 1881-117. Adv Struct Eng 2003. ASCE 2010. Teng JG. 1983. ACI. Eng Struct 2010. ACI. ASCE 2010 [in press]. Saatcioglu M. J Compos Constr. Behaviour of handmade FRP anchors under tensile load in uncracked concrete. 228 pages. Smith ST. p. UK: British Standards. Interface debonding failure in beams strengthened with externally bonded GFRP. Metallic materials—tensile testing. candidate). p. Matthys S. Yao J. Strengthening of masonry walls against out-of-plane load using fiber-reinforced polymer reinforcement. teknisk introduktion. Mr. Plastics—determination of tensile properties. ASCE 2001. candidate). Steel-reinforced concrete structures.13(2):82–92.Phil. Knudsen PJ. In addition. [2] Hollaway LC. [30] BS 1881-121. Composites B 2008. ASCE 2009. J Compos Constr. [38] Teng JG. Method for determination of tensile splitting strength. FRPRCS9 special edition. Aust J Struct Eng 2010. Special edition on fibre reinforcement in concrete structures. Smith ST. 2000. [14] Ozbakkaloglu T. Method for determination of flexural strength. Smith et al. [28] ACI 440.12(6):608–16. Sim J. In: Proceedings. [18] Seliem HM. Al-Sulaimani GJ. Wu YF. 5. [25] Brunckhorst L. Esbjerg (Denmark): Rosendahls Bogtrykkeri A/S. candidate) of the Technical University of Denmark is thanked for his assistance with the experimental work and also for producing the threedimensional drawings.17(6–7):447–62. Farmington Hills (MI. 2008. Debonding strength and anchorage devices for reinforced concrete elements strengthened with FRP sheets. metallic and timber construction materials with FRP composites. / Engineering Structures 33 (2011) 1075–1087 2.3R-04. ACI Struct J 1994.11(2):75–86. [32] BS 1881-118. [10] Smith ST. Anchors of greater fibre content positioned closer to the peak bending moment region in addition to anchors of lesser fibre content but spaced close together near the free ends of the FRP plate produced the greatest enhancement in strength with significant deflection capability. UK: Elsevier. Chan W. Retrofitting of deficient RC cantilever slabs using GFRP strips. Design of FRP and steel plated RC structures: retrofitting beams and slabs for strength.107(2):237–47. Wang J. UK: British Standards. second Asia-Pacific conference on FRP in structures. Method for determination of static modulus of elasticity in compression. Farmington Hills (MI. Guide test methods for fiber-reinforced polymers (FRPs) for reinforcing or strengthening concrete structures.11(3):117–26.8(1):55–68. USA: University of Missouri-Rolla. Eng Struct 2007. [3] Sharif A. [36] Smith ST. Strength of RC cantilever slabs bonded with GFRP strips. Strengthening of concrete. Adv Struct Eng 2005.11(1):25–42. Sumner EA. Strengthening and rehabilitation of civil infrastructures using fibre-reinforced polymer (FRP) composites. [4] Yun YC.30(11):3129–40. FIB 2010. 1983.5(4):221–7.4(2):75–84. Seismic rehabilitation of slab–column connections. [19] Widianto Bayrak O.32(4): 964–73. Thorsen T. CICE 2010. Constr Build Mater. Anchors spaced far apart led to gains in deflection capacity but limited gains in strength. Compos Interfaces 2004. 4. [15] Oh H-S.