Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks

Composite Structures xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

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

Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks Francesca Giulia Carozzi, Pierluigi Colombi, Giulia Fava ⇑, Carlo Poggi Department of Architecture, Built Environment and Construction Engineering (ABC), Politecnico di Milano, P.zza L. da Vinci, 32, 20133 Milan, Italy

a r t i c l e

i n f o

Article history: Received 2 November 2016 Revised 19 December 2016 Accepted 3 February 2017 Available online xxxx Keywords: Carbon and glass fibres Anchor spikes Debonding Mechanical testing Finite Element Analysis (FEA)

a b s t r a c t Fibre Reinforced Polymer (FRP) materials are extensively used to retrofit masonry and reinforced concrete structures. Failure occurs in most cases due to composite debonding from the substrate. The use of FRP anchor spikes was thus proposed to reduce premature debonding failure. In this work, the results of an extensive experimental program on the bond behaviour between the FRP anchor spikes and the substrate are first presented. They include both the mechanical characterization of carbon and glass FRP anchor spikes and pull-out tests from concrete and masonry blocks. In pull-out specimens, FRP anchor spikes are embedded into the block (embedded anchor spikes) or fanned-out on the substrate surface (fanned-out anchor spikes) with an epoxy resin. For different specimen configurations and materials, the bond behaviour of the FRP anchor is analysed. The mostly observed failure modes were the tensile failure of the anchor spikes, the debonding of the anchor spikes from the substrate or a combination of both. Nonlinear finite element simulations were finally performed to understand the bond behaviour between FRP anchors spikes and concrete substrate. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Fibre Reinforced Polymer (FRP) materials are nowadays widely used to strengthen and retrofit existing masonry and concrete structures [1]. FRP external reinforcement is considered to be a more efficient technique than other traditional methods due to a lot of advantages of the FRP materials, such as superior mechanical properties, the durability and the light weight that leads to rapid installation as well as low transportation and labour costs. Nonetheless, one of the major points of concern in the use of FRP strengthening systems is related to the bond strength between the FRP reinforcement and the substrate and to the bond durability under both environmental actions (harsh environments) [2] and mechanical loading (fatigue) [3]. In particular, premature debonding may occur due to the low substrate tensile strength or the weak bond strength at the interface between the FRP reinforcement and the substrate. Debonding may be delayed by using anchorage systems [4–8]. Many types of anchorage systems have been already studied in past researches [9], such as FRP anchor spikes, FRP transversal wrappings, FRP strips, U-Anchors, patch anchors, longitudinal chases, plate anchors, bolted angles, cylindrical hollow sections ⇑ Corresponding author. E-mail address: [email protected] (G. Fava).

and ductile anchorage systems. They present different geometries, installation limits and force transfer characteristics [10,11]. Above all the types of anchorage system already described, the FRP anchor spikes will be studied in depth in this paper. Only few researches on this topic are available, furthermore, there are no specific rules or guidelines for the anchor spike design. 1.1. Problem statement and previous studies The adhesion between the reinforcement and the substrate is a critical issue in FRP reinforcing systems. In order to prevent the debonding failure, the application of FRP anchor spikes, which are commonly assumed to resist axial and shear forces, is considered. FRP anchor spikes are strands of bundle fibres (see Fig. 1a) which are embedded or fixed into the concrete or masonry substrate by using epoxy resin. The free end of the anchor spikes is fanned-out and fixed by epoxy resin in order to join the anchor spike to the FRP reinforcement (see Fig. 1b). In some cases, the anchor spikes are extended on the external surface opposite to the FRP reinforcement, fanned-out and fixed again by using epoxy resin (see Fig. 1b). In the first case, the embedded portion of the anchor spike (embedded anchor spikes) contributes to the axial resistance while in the second one the anchor end is fanned-out on the external surface (fanned-out anchor spikes) and provides a significant

http://dx.doi.org/10.1016/j.compstruct.2017.02.026 0263-8223/Ó 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

2

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Fig. 1. Anchor spikes: (a) carbon (top) and glass (down) and (b) different uses of anchor spikes in practical situations.

contribution. They are usually installed orthogonal to or in plane with the FRP system, although other orientations may exist [9]. The role of the FRP anchor spikes in improving the FRP system strength is an interesting item studied by different research groups. Orton et al. [12] reported the outcomes of an experimental program on concrete beams strengthened with CFRP strips anchored both with U-Anchors and with anchor spikes. The beams were notched at the centreline and subjected to bending tests. It was observed that the use of anchor spikes was more effective than U-Anchors. Moreover, anchor spikes having fan angle less than 90° showed a better mechanical performance. Niemitz et al. [13] examined the efficiency of CFRP anchors by means of single shear tests on reinforced concrete blocks strengthened with CFRP sheets. They found that the use of FRP anchors was a viable technique in combination with FRP bonded sheets and then studied the effectiveness of FRP anchors with respect to geometric characteristics of the anchor. A wide research on carbon and glass FRP anchors on concrete members has been performed at the Hong Kong University. As an example Zhang et al. [14] used FRP anchors to enhance the strength capacity of the FRP plates bonded to RC members. They highlighted the importance of the anchor fan configuration and of the dowel angle and defined a simply linear relationship between the angle of the anchor dowel and the joint strength enhancement. Smith et al. [15] studied the ability of FRP anchors to influence the load and deflection capacity of FRP-strengthened RC slabs. They analysed the influence of anchor location, geometry, inclination of anchor dowel angle and fibre content. Optimal anchor arrangements enabled the load and deflection capacities of the FRP-strengthened slabs, in relation to the unanchored but strengthened ones.

Kalfat and Al-Mahaidi performed a wide experimental campaign for the development of an anchor system able to improve the bond performance at the interface between the FRP reinforcement and the concrete substrate and to moderate premature failure of the joint [16,17]. In detail, anchor spikes were found to provide some anchorage improvement due to the dowel action. Concerning the use of anchor spikes on masonry members, fewer studies were recently presented. Borri et al. [18] applied anchor spikes on masonry arches reinforced with GFRP sheets or CFRP plates. By performing an experimental campaign, it was observed that the use of anchor spikes could avoid the premature debonding of the reinforced system as well as the premature peeling. No evident increment in the load carrying capacity was highlighted due to the damage in the masonry caused by the drilled holes used for the insertion of the anchor spikes. Caggegi et al. [19] performed experimental tests to quantify the efficiency of the CFRP anchors on the strength of a reinforced brick. When CFRP anchors were properly placed, an increase of the reinforcement resistance and of the dissipation capability was observed. Then, Fagone et al. studied the effects of mechanical anchors applied to FRP-to-masonry joints [20] and to masonry pillars [21]. In both cases they implemented near-end supportedsingle shear tests and analysed how the use of anchor spikes improved the performance of the reinforcement system. The experimental campaigns showed that properly designed mechanical anchors increased both the failure load and the ductility of the reinforcement. Finally, Caggegi et al. [22] performed an experimental study on bricks reinforced by using FRP sheets and fibre anchors with different configurations. The displacement field was used to describe the debonding process along the reinforced substrate and it was measured by means of a Digital Image Correlation optical non-destructive method. The analysis of the displacement field

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

underlines that the anchors change the behaviour of the strengthened system and take part in the bond between the reinforcement and the substrate. The importance of FRP anchor spikes in improving the performance of the reinforcement system has been widely discussed on the basis of experimental campaigns performed in the literature. Nonetheless, at the system level, no specific guidelines or design criteria still exist to define the strengthening enhancement of reinforced structures after the installation of FRP anchors. Besides, at the component level, the behaviour of the anchor spikes should be better investigated. At both levels, it is essential to ensure that the anchor spike has a proper bond behaviour with respect to the concrete or masonry substrate. In fact, in the design of structural elements reinforced with anchor spikes, it is considered that if debonding collapse occurs, the materials are not completely exploited up to their best performance and thus bond failure should be avoided. On the other hand, attaining a good bond strength from FRP strongly orthotropic anchors may be troublesome due to the presence of high strength unidirectional fibres inside a low strength polymeric matrix. The aim of this research is therefore to present and discuss the results of an extensive experimental campaign on the bond behaviour of different types and configurations of anchor spikes inserted in concrete or masonry blocks and therefore to identify the key parameters and the phenomena that affect the bond performance of FRP anchors. 1.2. Scope of the research As stated in the previous section, even though various experimental campaigns clearly show the significant role of FRP anchors in the strength enhancement of reinforced members, the motivation of this study consists in the evaluation of the axial resistance of the anchor spikes and in the identification of the main factors influencing the relevant performance. Reference is made to both concrete and masonry substrates. In this study the pull-out strength of Carbon Fibre Reinforced Polymers (CFRP) and Glass Fibre Reinforced Polymer (GFRP) anchor spikes applied on brick and concrete members was firstly investigated from the experimental point of view. The main objectives of the experimental campaign included: (a) the analysis of the influence of mechanical properties and geometry of the substrate on the bond strength; (b) the analysis of the failure modes that characterized the different experimental configurations; (c) the study of the various phenomena that influenced the variability of the results, particularly in terms of maximum load reached and failure mode. Besides, some of the pull-out tests were simulated by means of a nonlinear finite element analysis to understand the bond behaviour between the FRP anchor and the concrete substrate. Finally, a parametric analysis was performed in order to investigate the effect of some design parameters on the bond strength. The anchor system studied in this work is constituted by two anchor spikes configurations, that is, carbon or glass fibres embedded anchor spikes or two different arrangements of fanned-out anchor spikes bonded on the substrate surface with an epoxy resin. It must be highlighted that the system under investigation is prepared and installed on-site, without a pre-impregnation of the fibres. This aspect included a lot of advantages related to the absence of discontinuities in the anchors between dry and impregnated fibres, and the possibility to adapt the geometry of the system to the properties of the substrate. On the other hand, a drawback in the on-site application of anchor spikes is related to the difficulties of ensuring a homogeneous impregnation of the fibres and to the fact that it is not possible to control the effective volume ratio.

3

2. Experimental program In this section an experimental program performed at the Materials Testing Laboratory of the Politecnico di Milano is presented. The fibre reinforced anchor systems consisted of carbon or glass fibres embedded anchor spikes or fanned-out anchor spikes. Tensile tests were carried out to characterize the mechanical properties of glass and carbon fibres and several pull-out tests were performed. The anchor spikes were applied with various configurations on substrates with different mechanical and geometrical properties. Details on the materials characterization, specimen geometry and preparation and pull-out tests are reported in the following. 2.1. Materials The reinforcement system consisted of dry anchor spikes in carbon or glass fibres (see Fig. 1a) impregnated with an epoxy resin and applied on the substrate with different configurations. Two different epoxy resins were employed, resin ‘‘type R-A” was used to impregnate the dry fibres while ‘‘resin R-B” was inserted in the drilled hole (embedded anchor spike) or applied on the external surface of the block (fanned-out anchor spike) to bond the anchor system. Different substrate materials were involved, that are concrete blocks and masonry blocks. The masonry blocks were realized using two different types of mortar and brick. The results of experimental tests aiming at determining the mechanical properties of the materials listed above are reported in the following. 2.1.1. Tensile tests of the epoxy resins Tensile tests were carried out for evaluating the tensile strength and the elastic modulus of the two types of epoxy resins used in the anchor system. Adhesive panels, with an average thickness of 4 mm, were prepared on a support which was previously covered with greaseproof paper to ease the detachment. After the adhesive panel curing, conforming to the requirements of ASTM Standard D638 (2003) [23], dumb-bell specimens were cut from the panels and tested in tension. Specimen dimensions and test set-up are shown in Fig. 2. At the ends of the specimens, FRP tabs were applied to avoid the local damage of the samples by the wedges of the tensile machine. A tensile machine with a maximum load of 2 kN and an extensometer with a base length of 20 mm were used. For each resin type, three samples were tested and the experimental results are reported in Table 1. Clearly, the mechanical properties of the adopted resins are quite different. The first one (‘‘resin R-A”), in fact, was used just to impregnate the anchor system and distribute the stresses between the dry fibres and then no specific mechanical strength was required. On the other hand, the second one (‘‘resin R-B”) was adopted to fix the anchor spike in the drilled hole or to fix the fanned-out spike to the external surface of the block. A specific mechanical resistance was then required in order to guarantee the required bond strength. 2.1.2. Compression tests on masonry and concrete blocks The masonry blocks were of 18.5  25  25 cm (length, thickness, height) and they were realized with two different types of mortar and brick. The first type (labeled ‘‘M-A”) was made of sandblasted bricks with lower mechanical characteristics and lime based mortar. The second one (labeled ‘‘M-B”) was made of no sandblasted bricks with high mechanical properties and a cementitious mortar. These two different types of materials were

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

4

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Fig. 2. Tensile tests on epoxy resin: (a) dimensions (in mm) of a dumb-bell specimen on resin ‘‘type R-A”, (b) dimensions (in mm) of a dumb-bell specimen on resin ‘‘type R-B” and (c) experimental set-up.

type, were prepared by impregnating the anchor spike dry fibres with the epoxy resin (‘‘type R-A”). The fibres were manually impregnated and much attention was paid to guarantee homogeneous fibres saturation and constant volumetric fibre ratio. The average volumetric fibre ratio was equal to about 38% for CFRP bars and to 49% for GFRP bars. The specimens were cured for one week and finally presented a bond length of 700 mm and an average diameter of 6 mm. The tensile tests were carried out with two testing machines, that is an Amsler horizontal machine with maximum capacity of 300 kN and an MTS machine with maximum capacity of 250 kN. The deformations were recorded with an extensometer with base length equal to 100 mm. Tests were performed under displacement control and the test speed was equal to 0.5 mm/min. In order to guarantee a homogeneous stress distribution and no damage in the samples, carborundum abrasives and aluminium tabs were located at the ends of the specimens. The failure mode was brittle and in some cases the collapse of the impregnated fibres showed an explosive failure, as shown in Fig. 3 for glass fibre. The experimental results are reported in Table 3.

Table 1 Tensile tests on epoxy resin: experimental results.

Average CoV Average CoV

Sample

Tensile strength [MPa]

Elastic modulus [GPa]

Type R-A

19.11 3.04% 30.67 4.36%

1.83 3.67% 6.28 2.27%

Type R-B

involved to study the influence of the substrate characteristics in the failure mode and in the maximum load reached before collapse. To characterize the materials, the following quantities were experimentally determined [24], that is, compression strength and elastic modulus, tensile strength and pull-off strength. In Table 2, the main experimental results and the corresponding standard requirements are listed. The selected bricks have quite different mechanical strength in order to investigate the effect of this parameter on the bond strength of the fanned-out anchor spikes. The concrete blocks were of 20  20  20 cm (length, thickness, height) and only the compressive strength was evaluated after a curing of 28 days. An average compressive cubic strength equal to 48.7 MPa was found.

2.2. Specimen geometry and preparation

2.1.3. Tensile tests on carbon and glass anchor spikes The tensile strength and the elastic modulus of the carbon and glass fibres of the anchor spikes were derived from tensile tests on carbon and glass composite bars. The specimens, for both fibre

In the following the specimen geometry is described, two different procedures were adopted for anchors spikes embedded into the block or fanned-out on the substrate surface. Concrete blocks with dimensions of 20  20  20 cm (length, thickness, height) and

Table 2 Experimental results of tests on mortar and bricks. Sample Standard Average CoV Average CoV

Type M-A

Average CoV Average CoV

Type M-B

Brick Mortar Brick Mortar

Compressive strength [MPa] EN 1015-11 [25] EN 772-1 [26]

Elastic modulus [GPa] EN 14580 [27]

Tensile strength [MPa] EN 12390-6 [28]

Pull-off strength [MPa] EN 1542 [29]

22.32 (12) 12.05% 5 (data sheet) –

– – – –

1.81 (3) 20.08% – –

– – – –

68.87 (12) 5.88% 27.13 (6) 13.07%

12.33 (4) 5.02% 7.60 (data sheet) –

6.24 (3) 14.12% – –

3.07 (12) 7.52% 1.60 (data sheet) –

Note: within brackets it is specified the number of tested samples or if the experimental value was taken from datasheets.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

5

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Fig. 3. Tensile test on a glass anchor spike showing the brittle failure mode.

masonry blocks with dimensions of 18.5  25  25 cm (length, thickness, height) were used for testing the anchor spikes. Details concerning the specimen preparation are reported in the following. The different types of tested anchor spikes configurations are represented in Fig. 4. The first one (configuration c1) refers to embedded anchor spikes while the second one to fanned-out anchor spikes with two different arrangements (configurations c2 and c3). These configurations were selected in order to represent the practical situations reported in Fig. 1b. For embedded anchor spikes (see configuration (c1) in Fig. 4), a central hole with a depth of 100 mm and a diameter of 16 mm was

drilled in the concrete and masonry blocks. The hole was blown up with an air compressor, paying attention to perfectly clean the inner surface from the dust. The anchor spikes were then impregnated with the epoxy resin ‘‘type R-A” (see Fig. 5a) and fixed before curing of the resin (see Fig. 5b) with resin ‘‘type R-B”. Attention was paid in order to avoid the presence of air bubbles that damages the correct adhesion to the substrate. The samples were cured for a week keeping the anchors in a vertical position. For fanned-out anchor spikes on the substrate surface, a 25 mm-diameter central hole was drilled in the concrete and masonry blocks passing through the depth of the specimen. In order to avoid damage of the fibres when applying the anchor system, a chamfer was bevelled on the block and then smoothed. The chamfer radius was equal to approximately 35–40 mm. The dust was blown up from the hole. Then the anchor spikes were impregnated using the epoxy resin ‘‘type R-A”, inserted into the blocks and finally fanned-out and fixed with resin ‘‘type R-B” on the external surface of the blocks (see Fig. 5c). A Teflon layer was inserted in the hole in order to prevent bonding between the anchor spike and the block. In this way, just the contribution of the fanned-out portion of the anchor spike was investigated in the experimental tests. Two anchor spike arrangements were considered. The first one did not include a specific surface preparation and the anchor spike was homogeneously fanned-out on the external surface (see configuration (c2) in Fig. 4) with a fan angle equal to 360°. In the second configuration, eight star-shaped slits with a depth of about 10 mm

Table 3 Experimental results of tensile tests on GFRP and CFRP composite bars. Sample

Tensile load [kN]

Tensile strength [MPa]

Elastic modulus [GPa]

Average CoV

Carbon

52.33 (10) 8.33%

1772 (10) 8.33%

226.2 (3) 4.12%

Average CoV

Glass

45.61 (5) 12.69%

1629 (5) 12.69%

39.6 (3) 5.03%

Note: within brackets it is specified the number of tested samples.

(c1)

(c2)

(c3)

Fig. 4. Configurations of the tested specimens: (c1) anchor spike fixed in the block; (c2) anchor spike homogenously fanned on the substrate; (c3) eight star-shaped anchor fanned on the substrate.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

6

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Fig. 5. Specimen preparation for carbon anchor spike: (a) anchor system impregnation, (b) the anchor is inserted in the embedded anchor spike and (c) the anchor is fixed in the fanned-out anchor spike.

were cut into the external surface. Then the anchor spike was fanned-out in eight parts and bonded into the slit with the epoxy resin (see configuration (c3) in Fig. 4). For both arrangements the radius of the fanned-out part was equal to 100 mm. All the samples were cured for a week with the connector in vertical position. 2.3. Pull-out test on embedded or fanned-out anchor spikes Bond strength between the CFRP or GFRP anchor spikes and the concrete or masonry substrate was experimentally measured by performing pull-out tests. A total of fifty-three pull-out tests on CFRP and GFRP anchor spikes anchored in concrete or masonry blocks were performed. The experimental matrix, concerning the specimen configurations and materials used is reported in Table 4. In Table 4, specimen labels are as follows. The first two digits indicate (see Fig. 4) the specimen configuration: ‘‘c1” for the anchor spike embedded in the block, ‘‘c2” for the anchor spike fanned-out homogenously on the substrate and ‘‘c3” for the eight star-shaped anchor spike fanned-out on the substrate. Then the letters (‘‘c”, ‘‘mA” or ‘‘mB”) refer to substrate material where ‘‘c” stands for concrete while ‘‘mA” and ‘‘mB” stand for the different types of masonry bricks, respectively. Finally, the capital letters ‘‘C” or ‘‘G” define the type of fibres (‘‘C” stands for carbon and ‘‘G” for glass) adopted in the anchor system. As an example, specimen type ‘‘c1-mA-C” refers to carbon anchor spikes embedded in a masonry block of type ‘‘M-A” while specimen type ‘‘c2-c-G” refers to glass anchor spikes homogenously fanned-out on a concrete

block and specimen type ‘‘c3-c-C” refers to eight star-shaped carbon anchor spikes fanned-out on a concrete block. The adopted pull-out test setup satisfied the recommendations of standard [30] for steel rebars in concrete. A mechanical testing machine with maximum load capacity of 1000 kN was used. Tests were driven under displacement control with a rate of 1 mm/min. Both the load and the displacement values were provided by the testing machine. In order to avoid alignment problems, a spherical hinge was located between the sample and the rigid plate. As the tensile tests on the anchor spikes, carborundum abrasives and aluminium tabs were located at the ends of the specimens in order to guarantee a homogeneous stress distribution and no damage in the samples. The pull-out experimental set-up is shown in Fig. 6.

3. Experimental results In this section, the experimental results for the embedded and fanned-out anchor spikes will be presented and compared. In particular, the observed failure modes will be illustrated together with the relevant bond strength and the load-displacement curves. 3.1. Embedded anchor spikes in masonry or concrete blocks A total of twenty-two specimens were tested for the analysis of the bond behaviour of embedded CFRP and GFRP anchor spikes fixed in concrete and masonry substrates.

Table 4 Pull-out test configuration and materials. Anchor

Substrate

Spec. type c1: embedded anchor spike Carbon ‘‘C” Masonry ‘‘type M-A” Concrete ‘‘c” Glass ‘‘G” Masonry ‘‘type M-A” Concrete ‘‘c” Spec. type c2: fanned-out anchor spike (homogeneous Carbon ‘‘C” Masonry ‘‘type M-A” Masonry ‘‘type M-B” Concrete Glass ‘‘G” Masonry ‘‘type M-B” Concrete ‘‘c”

Specimen

d [mm]

L[mm]

r [mm]

R [mm]

# of tests

c1-mA-C c1-c-C c1-mA-G c1-c-G

16 16 16 16

100 100 100 100

– – – –

– – – –

4 10 5 3

25 25 25 25 25

– – – – –

35–40 35–40 35–40 35–40 35–40

100 100 100 100 100

2 4 5 7 5

25

–

35–40

100

8

distribution) c2-mA-C c2-mB-C c2-c-C c2-mB-G c2-c-G

Spec. type c3: star fanned-out anchor spike (eight star-shaped) Carbon ‘‘C” Concrete ‘‘c” c3-c-C d = central hole diameter. L = bond length. r = chamfer radius. R = radius of the fanned-out part.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

7

Fig. 6. Pull-out test set-up for anchor spike embedded in a masonry (left) and concrete (right) block.

For all the specimens consisting of embedded CFRP and GFRP anchor spikes in masonry blocks, the observed failure mode was the cohesive fracture of the substrate (Fig. 7a). Such failure mode occurred at first in the mortar vertical joint, and then vertical cracks were also observed into the bricks. In the lower part of the pull-out specimen, the cracks showed an angle of about 45° and, after the collapse, the substrate remained bonded to the anchor. The failure load was correlated to the mechanical properties of the substrate and to the installation of the anchor spike. For the samples consisting of embedded CFRP and GFRP anchor spikes in concrete blocks, two failure modes, or a combination of them, were mainly detected. They are the anchor failure in the free length or at the end of the free length (Fig. 7b) and the debonding failure (Fig. 7c). Note that this last failure mode is anticipated by the cohesive failure of the top (free) surface of the concrete block (see Fig. 7c). The experimental results are collected in Table 5 in terms of peak load and failure mode. Based on the comparison between the test results of carbon anchors in masonry (specimen type ‘‘c1-mA-C” and Table 5) and the ones of glass systems (specimen type ‘‘c1-mA-G” and Table 5), no substantial differences were found. On the other hand, a large variability in the experimental results was observed due to the handmade preparation of the samples. An extremely important aspect is the distribution of the epoxy resin into the hole due to the presence of voids or air bubbles that could affect the bond strength between the anchor system and the substrate.

The failure mode for CFRP anchor spikes embedded in concrete blocks was either anchor failure (specimen types ‘‘c1-c-C1”, ‘‘c1-cC2”, ‘‘c1-c-C3”, ‘‘c1-c-C4” and ‘‘c1-c-C5”) or debonding from the substrate (specimen types ‘‘c1-c-C6”, ‘‘c1-c-C7”, ‘‘c1-c-C8” and ‘‘c1-c-C9”). The relevant average failure load (Table 5) was of 37.35 kN and 41.67 kN, respectively. Since these two average failure loads are quite similar, it may be deduced that the main causes determining the occurrence of a certain failure mode were the specimen preparation, the interface properties and the alignment of the anchor spike, the block and the testing machine. A large variability of the results was observed, equal to approximately 30%. Only in one specimen (sample ‘‘c1-c-C-10”), the collapse was due to the tensile failure of the anchor. In that case, the value of the peak load corresponded to 52.94 kN which is very close to the tensile strength of the carbon anchor spike listed in Table 3 (52.33 kN). For GFRP anchor spikes and concrete substrate, the failure was due to the rupture at the end of the free length, with an average failure load of 30.31 kN (Table 5). Again, a variability close to the one recorded for CFRP anchor spikes (30%) was found. This means, again, that the specimen preparation, the interface properties and the specimen installation have a significant impact on the failure load. Load vs. displacement curves are depicted in Fig. 8 for CFRP and GFRP anchors inserted in masonry blocks and in Fig. 9 for anchors in concrete blocks. In detail, Fig. 8a and b respectively show the load-displacement curves for embedded carbon and glass anchor spikes in masonry

Fig. 7. Failure modes for anchor spikes fixed into masonry or concrete blocks: (a) cohesive failure of the masonry block, (b) tensile failure at the end of the free length and (c) debonding failure for anchor spikes embedded in a concrete block.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

8

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Table 5 Failure modes and peak loads of pull-out specimen tests on anchor spikes fixed in masonry or concrete block: CFRP and GFRP anchor spikes. Specimen c1-mA-C-1 c1-mA-C-2 c1-mA-C-3 c1-mA-C-4 Average c1-c-C-1 c1-c-C-2 c1-c-C-3 c1-c-C-4 c1-c-C-5 Average c1-c-C-6 c1-c-C-7 c1-c-C-8 c1-c-C-9 Average

Failure mode

cohesive failure

anchor failure (a) debonding and anchor failure (a) anchor failure (a) debonding and anchor failure (a) anchor failure (a)

debonding

Peak load [kN] 22.73 13.54 13.33 15.17 16.19 32.38 33.06 38.42 38.76 44.12 37.35 41.61 32.94 43.33 48.81 41.67

c1-c-C-10 anchor failure (b) 52.94 Note: In the failure mode, ‘‘a” indicates the anchor failure at the end of the free length while ‘‘b” indicates the fibre tensile failure in the free length. Specimen c1-mA-G-1 c1-mA-G-2 c1-mA-G-3 c1-mA-G-4 c1-mA-G-5 Average c1-c-G-1 c1-c-G-2 c1-c-G-3 Average

Failure mode

cohesive failure

anchor failure (a)

Peak load [kN] 17.24 19.82 18.50 14.09 21.67 18.26 34.43 31.06 25.44 30.31

Note: In the failure mode, ‘‘a” indicates the anchor failure at the end of the free length while ‘‘b” indicates the fibre tensile failure in the free length.

Fig. 8. Load-displacement behaviour of anchor spikes embedded into masonry blocks: (a) CFRP and (b) GFRP anchors spikes.

blocks. It could be highlighted that the carbon anchor systems in masonry appear to be stiffer than the glass ones. In fact, the average stiffness of carbon anchor systems is equal to 1.18 kN/mm, while the one of glass anchors in masonry is equal to 0.97 kN/m. The same holds for embedded carbon and glass anchor spikes in concrete blocks (Fig. 9a and b) 3.2. Fanned-out anchor spikes on masonry or concrete blocks A total of thirty-one specimens were tested for the analysis of the bond behaviour and strength of fanned-out CFRP and GFRP anchor spikes fixed in concrete and masonry blocks. The samples realized with masonry blocks showed two different failure modes depending on the mechanical characteristics of the substrate. They were mainly due to: (a) the debonding of the anchor spike from the substrate (b) the anchor failure at the end of the free length. The samples consisting of sandblasted bricks with lower mechanical properties and lime based mortar showed the debonding of the fibres applied on the surface and the consequent failure of the substrate (Fig. 10a). The samples made of bricks and cementitious mortar with higher mechanical properties presented the anchor failure at the end of the free length. In those cases, in fact, the failure is not influenced by the substrate properties but occurs at the end of the free length due to the anchor failure (Fig. 10b). The samples realized with concrete blocks showed mainly the anchor failure at the end of the free length (Fig. 10c). The experimental results for fanned-out CFRP and GFRP anchor spikes on concrete and masonry blocks are collected in Table 6, where the values of the peak loads and the failure modes are reported.

Fig. 9. Load-displacement behaviour of anchor spikes embedded into concrete blocks: (a) CFRP and (b) GFRP anchors spikes.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

9

Fig. 10. Failure modes for anchor spikes fanned-out on the surface of masonry or concrete blocks: (a) debonding from the masonry substrate, (b) anchor failure at the end of the free length for masonry substrate and (c) anchor failure at the end of the free length for concrete substrate.

Table 6 Failure modes and peak loads of pull-out specimen tests on anchor spikes fanned on masonry or concrete substrate: CFRP and GFRP anchor spikes. Specimen c2-mA-C-1 c2-mA-C-2 Average c2-mB-C-1 c2-mB-C-2 c2-mB-C-3 c2-mB-C-4 Average c2-c-C-1 c2-c-C-2 c2-c-C-3 c2-c-C-4 c2-c-C-5 Average c3-c-C-1 c3-c-C-2 c3-c-C-3 c3-c-C-4 c3-c-C-5 Average c3-c-C-6 c3-c-C-7 c3-c-C-8 Average Specimen c2-mB-G-1 c2-mB-G-2 c2-mB-G-3 c2-mB-G-4 c2-mB-G-5 c2-mB-G-6 c2-mB-G-7 Average c2-c-G-1 c2-c-G-2 c2-c-G-3 c2-c-G-4 c2-c-G-5 Average

Failure mode debonding

anchor failure (a)

anchor failure (a)

anchor failure (a)

debonding and anchor failure (b) anchor failure (b) Failure mode

anchor failure (a)

anchor failure (a)

Peak load [kN] 23.82 25.05 24.43 18.12 22.68 21.57 20.03 20.60 24.33 34.34 35.87 21.26 27.16 28.59 31.67 36.00 32.09 43.78 25.63 33.83 50.93 45.91 40.55 45.80 Peak load [kN] 22.49 20.14 20.12 21.98 26.35 19.37 18.04 21.21 22.24 23.62 16.16 20.09 14.55 19.33

Note: In the failure mode, ‘‘a” indicates the anchor failure at the end of the free length while ‘‘b” indicates the fibre tensile failure in the free length.

Comparing the values of the peak loads for homogenously fanned-out carbon and glass anchor spikes on masonry substrate

(specimen types ‘‘c2-mB-C” and ‘‘c2-mB-G”), no substantial differences were found. Besides, a small variability in the experimental results was observed in this case. On the other hand, for homogenously fanned-out CFRP anchor spikes on concrete blocks (specimen type ‘‘c2-c-C”), the average value of the peak load was remarkably larger than the one of GFRP anchor spikes (specimen type ‘‘c2-c-G”). In detail, for specimens type ‘‘c2-c-C”, the average peak load was 28.59 kN with a significant variability of the results (40%) while for specimen type ‘‘c2c-G”, the average peak load was 19.33 kN with a similar variability of the results. In the configuration consisting on eight star-shaped fanned-out anchor spikes on concrete substrate (specimen type ‘‘c3-c-C”), the cut of eight slits into the substrate led to a better control of many geometrical parameters and factors related with samples preparation. With respect to specimen type ‘‘c2-c-C”, specimens tested under such configuration showed higher peak loads and, for failure due to the rupture of the anchor at the end of the free length (specimens ‘‘c3-c-C-1”, ‘‘c3-c-C-2”, ‘‘c3-c-C-3”, ‘‘c3-c-C-4” and ‘‘c3-c-C5”), the average peak load was increased up to 33.83 kN. In three cases (specimens ‘‘c3-c-C-6”, ‘‘c3-c-C-7” and ‘‘c3-c-C-8”) the collapse occurred due to the tensile failure of the anchor in the free length. The relevant average failure load corresponded to 45.80 kN which is close to the tensile strength of the carbon anchor spike listed in Table 3 (52.33 kN). The variability of the results (25%) is lower than the companion system (homogeneous fanned-out anchor spike) since the cut of eight slits into the substrate led to a better control of the geometrical parameters. Load vs. displacement curves are depicted in Fig. 11 for homogenously fanned-out CFRP and GFRP anchor spikes and masonry substrate, in Fig. 12 for homogenously fanned-out CFRP and GFRP anchor spikes on the concrete substrate and, finally, in Fig. 13 for eight star-shaped fanned-out CFRP anchor spikes on the concrete substrate. In all the tested specimens, irrespectively from the considered configuration, the amount of epoxy resin used to bond the anchor spikes to the concrete or masonry substrate was manually controlled and was strictly correlated to large variability of the experimental results. Besides, it was observed that the radius of the chamfer bevelled in the block influenced the fibres curvature at the opening of the anchor and thus played an extremely important role in the bond strength and failure mode of the system. The effect of the chamfer radius on the variability of the experimental results will be investigated by means of the numerical model illustrated in the next section.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

10

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

Fig. 11. Load-displacement behaviour of anchor spikes homogenously fanned-out on the surface of masonry blocks: (a) CFRP and (b) GFRP anchors.

Fig. 12. Load-displacement behaviour of anchor spikes homogenously fanned-out on the surface of concrete blocks: (a) CFRP and (b) GFRP anchors.

4. Numerical model Finite element models to predict and better understand the behaviour of the bond between FRP rebars and concrete were already presented in the literature [31,32]. In the present investigation, finite elements models were developed and calibrated to understand the bond behaviour between FRP anchors and concrete substrate. In detail, the software ABAQUS [33] was adopted for the numerical simulation considering two different models corresponding to specimen types ‘‘c1-c-C” (embedded anchor spike in a concrete substrate) and ‘‘c2-c-C” (homogeneous fanned-out anchor spike on a concrete substrate). The masonry substrate was not taken into account since the dominant failure mode did not involve the anchor spike. Finally, the calibrated numerical model was used to investigate the effect of two different design parameters on the bond strength of CFRP anchor spikes. More precisely, the effect of bond length was considered for embedded anchor spikes while the effect of the chamfer radius was investigated for homogeneous fannedout anchor spikes. 4.1. Geometry, mesh and boundary conditions The first configuration under analysis represented the specimen type ‘‘c1-c-C”, that is, an embedded CFRP anchor spike in concrete (see Fig. 4(c1)). In modelling such configuration, five different bond lengths of 50 mm, 75 mm, 100 mm, 125 mm and 150 mm were taken into account in order to investigate the effect of such design parameter. The second experimental configuration was representative of the specimen type ‘‘c2-c-C”, that is, homogenously fanned-out CFRP anchor spikes on the concrete substrate (see Fig. 4(c2)). In

Fig. 13. Load-displacement behaviour of eight star-shaped CFRP anchors fannedout on the concrete surface.

modelling such configuration, the influence of the chamfer radius was analysed by considering three different types of chamfer radii; that is 25 mm, 40 mm and 55 mm. The geometric solid models contain two parts, that is the FRP anchor and the concrete block. An axisymmetric model was adopted for the analysis of the pull-out specimens. As already observed in previous numerical simulations [34,32], for FRP anchor systems embedded in concrete or in a resin head, the results of a 2D axisymmetric finite element model are very close to 3D ones. As a result, the 2D model was used in order (a) to verify the debonding onset and propagation at the interface between the anchor and the substrate, (b) to detect the stress distribution at the interface, (c) to eventually identify critical zones in the speci-

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

men and (d) to investigate the effect of some design parameters on the bond strength. The two axisymmetric specimen configurations have the same geometrical properties, i.e., the anchor diameter is 16 mm and the concrete block has a height of 185 mm and a width of 250 mm. Note that the anchor diameter is equal to the diameter of the hole drilled in the concrete block and filled by the resin type ‘‘resin R-B”. In the second configuration, since the anchor fibres are homogenously distributed over an anchor fan radius of 100 mm and since the amount of fibres and the fibre volumetric ratio are considered the same as the ones of the embedded part of the anchor spike, the thickness of the CFRP fanned-out fibres is assumed to be equal to 0.25 mm. Both the specimens are regarded as axisymmetric. Thus, fournode bilinear axisymmetric quadrilateral elements (CAX4R) and three-node linear axisymmetric triangle elements (CAX3) were used. Besides, boundary conditions were enforced to reproduce the symmetries. A refined mesh was generated around the bond surface, while the dimensions of the finite elements gradually increased toward the support. Note that a refined mesh is required due to the significant mismatch of the adherent stiffness (carbon fibre and concrete). The displacement of the bottom surface of the concrete block was constrained as in the experimental setup. The displacement control experimental set-up was simulated in the finite element analysis applying an increasing displacement at the loaded section of the CFRP anchor.

11

mechanics criterion adopting the Virtual Crack Closure Technique (VCCT) along a selected crack surface. Indeed, VCCT was firstly developed by Rybicki and Kanninen [35] and is a proper crack propagation criterion for brittle cracks occurring along predefined surfaces. In the VCCT failure criterion, the Linear Elastic Fracture Mechanics (LEFM) principles are assumed and it is stated that the strain energy released when a crack is extended by a certain amount is the same as the energy required to close the crack by the same amount. In the numerical simulation, the interfacial crack onset and growth were activated on the basis of the VCCT through the eval-

4.2. Material models and interface behaviour The constitutive laws of the materials were assumed to be isotropic and linear elastic. Based on the results of experimental tests, the concrete Young’s modulus was equal to 35313 MPa (see Table 2), while the CFRP Young’s modulus was of 226,200 MPa (see Table 3). For the Poisson ratio, typical values of 0.3 and 0.2 were assumed for the concrete and the CFRP anchor spike, respectively. For both ‘‘c1-c-C” specimen type and ‘‘c2-c-C” configuration, the progressive debonding at the interface between the anchor and the concrete substrate was numerically simulated. The interfacial crack onset and growth were detected on the basis of a fracture

Fig. 15. Results of the finite element simulations for pull-out test on specimen type ‘‘c1-c-C” with different bond length: (a) load-displacement behaviour and comparison with the experimental results and (b) load-slip curves.

Fig. 14. Results of the finite element simulations for pull-out test on specimen type ‘‘c1-c-C”: (a) stress distribution at failure in the anchor spike and (b) interface status.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

12

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

uation of the critical energy release rates (ERR) at the crack tip along the initially partially bonded interface. Strain energy release rates at the crack tip were computed for the mode-I due to opening tension (GI), the mode-II due to in-plane shear (GII) and the modeIII due to out-of-plane shear (GIII) [36]. In order to simulate the experimental outcomes, the critical ERR for modes I, II and III were defined after [37], considering GII/GI = 10 and GII/GIII = 1. This Table 7 FEM results as function of the bond length (specimens type ‘‘c1-c-C”) or the chamfer radius (specimens type ‘‘c2-c-C”). Spec. type

Bond length [mm]

Pu [kN]

c1-c-C

50 75 100 125 150

26.35 30.54 31.55 33.45 33.75

Spec. type

Chamfer radius [mm]

Pu [kN]

c2-c-C

25 40 55

25.79 29.39 33.18

Fig. 16. Anchor spike strength as function of the bond length for pull-out test on specimen type ‘‘c1-c-C”.

means that the ERR for mode-I is assumed to be negligible with respect to the ERR for mode-II and mode-III and that the critical ERR is the same for mode-II and mode-III. The ERR values were thus calibrated in the following way: values for GI ranging from 0.1 N/mm to 1.1 N/mm were considered and the relevant numerical results were compared to the experimental outcomes. Finally, both for specimen type ‘‘c1-c-C” and ‘‘c2-c-C”, ERR values of 0.3 N/mm, 3 N/mm and 3 N/mm were found for modes I, II and III, to fit the experimental results. 4.3. Numerical results In this section, the results from the proposed numerical model are discussed and compared to the experimental findings. For ‘‘c1-c-C” specimens, stress distribution in the anchor spike at failure is reported in Fig. 14a while the interface status is reported in Fig. 14b. Fig. 14a shows that the stress in the anchor spike is well below the tensile strength listed in Table 3. Fig. 14b shows that at failure a large portion of the interface is debonded revealing that debonding is the dominant failure mode. The numerical load-displacement plot and the load-slip curve are shown in Fig. 15. The load was calculated by integration of the stresses in the anchor in correspondence of the loaded section while the slip was the relative displacement between the CFRP anchor and the concrete substrate measured at the beginning of the bonded region. A reasonable agreement among the experimental curves and the numerical simulations was observed (see Fig. 15a). In Fig. 15a, the experimental curves should be compared with the numerical results for a bond length equal to 100 mm. Note that the bond length has also a significant influence on the value of the bond slip at failure, which increase with the bond length (see Fig. 15b). Maximum load levels as function of the different bond length (ranging from 50 mm to 125 mm) for embedded anchor spike are reported in Table 7. From the numerical results it is possible to define an effective bond length equal to 100 mm. For bond lengths larger than the effective one, in fact, no increment of the bond strength is observed. Finally, Fig. 16 illustrates the relationship between the anchor spike strength and the bond length. For ‘‘c2-c-C” specimens, the stress distribution in the anchor spike at failure is reported in Fig. 17a while the interface status is reported in Fig. 17b.

Fig. 17. Results of the finite element simulations for pull-out test on specimen type ‘‘c2-c-C”: (a) stress distribution at failure in the anchor spike and (b) interface status.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

13

face with an epoxy resin. In this last case, two different arrangements were adopted, that is, the anchor spike was fanned homogeneously or bonded into eight star-shaped slits cut into the substrate. The main conclusions of this work are:

Fig. 18. Results of the finite element simulations for pull-out test on specimen type ‘‘c2-c-C” with different chamfer radii: (a) load-displacement behaviour and comparison with the experimental results and (b) load-slip curves.

From Fig. 17a it is observed that at failure the maximum tensile stress in the CFRP (equal to about 1780 MPa) is greater than the allowable tensile stress (see Table 3). The failure was then also due in this case to the tensile failure of the anchor spike, as experimentally observed. In fact, Fig. 17b shows that at failure a part of the interface is debonded, revealing that debonding is not the dominant failure mode but interacts with CFRP tensile failure as experimentally observed. Finally, the load-displacement curves are plotted with respect to the chamfer radii and compared to some experimental results in Fig. 18a while the load-slip curves are shown in Fig. 18b. Again, the load was calculated by integration of the stresses in the anchor in correspondence of the loaded section while the slip was the relative displacement between the CFRP anchor and the concrete substrate measured at the beginning of the bonded region. A reasonable agreement among the experimental curves and the numerical simulations was observed (see Fig. 18a). In Fig. 18a, the experimental curves should be compared with the numerical results for a chamfer radius equal to 40 mm. Maximum load values (see Table 7) of 25.92 kN, 29.45 kN, 33.18 kN with respect to the chamfer radii of 25 mm, 40 mm and 55 mm were respectively found. Averagely, an increase of 35.7% in the radius led to a growth in the maximum load corresponding to about 30%. The chamfer radius has then a dominant influence on the randomness of the experimental results. Finally, it is observed from Fig. 18b that the chamfer radius has no influence on the bond slip at failure. 5. Conclusions The main aim of this paper was to investigate the bond behaviour between FRP anchor spikes and the concrete or masonry substrate, based on both an experimental campaign and numerical modelling. Pull-out tests were performed on CFRP and GFRP anchor spikes applied on brick and concrete blocks. The anchors were embedded into the block or fanned-out on the substrate sur-

 In most of the experiments on embedded anchor spikes, the observed failure modes for CFRP anchor spikes in a concrete substrate consisted of debonding and of anchor failure at the end of the free length. This was not the case for GFRP anchor spikes where debonding was not observed. For masonry substrate, the failure mode was the cohesive fracture of the substrate, both for CFRP or GFRP anchor spikes. The observed failure modes for concrete substrate corresponded to different ultimate loads. As an example, the anchor failure at the end of the free length for CFRP anchor spikes resulted in an ultimate strength equal to 60% of the CFRP tensile strength while debonding failure mode presented a maximum load equal to 80% of the CFRP tensile strength. A similar trend was observed for GFRP anchor spikes and concrete substrate. The observed different failure mode and bond strength are related to the influence of the handmade preparation. This resulted also in a significant variability in the experimental results.  For fanned-out anchor spikes and concrete or masonry substrates, collapse was generally due to the anchor failure at the end of the free length for both CFRP and GFRP anchor spikes. Homogeneous fanned-out CFRP anchor spikes led to a bond strength equal to 50% of the tensile strength. This value is lower than the corresponding one for embedded anchor spikes (60%). Finally, GFRP homogeneous fanned-out anchor spikes presented even a lower bond strength, equal to about 40% of the tensile strength. This was due to the difficulties related to the system preparation and installation. On the other hand, the cut of eight slits into the substrate led to a better control of the geometrical parameters and to higher peak loads, up to 90% of the tensile strength of the anchor spikes. This did not result in a lower variability of the experimental results due to the effect of the specimen preparation and in particular to the effective value of the chamfer radius.  Numerical modelling was also implemented aiming to assess the bond behaviour between CFRP anchors and concrete substrate. Numerical results agree with the experimental ones. Besides, the numerical model was also used to investigate the effect of some design parameters on the bond strength. In particular, the influence of the bond length for embedded anchor spikes was explored. Results revealed an effective bond length equal to 100 mm. Additionally, the influence of the chamfer radius on the bond strength of homogeneously fanned-out anchors spike was investigated in order to explain the randomness of the experimental outcomes. Results indicate a strong influence of this parameter on the bond strength. Since it is quite difficult to control this parameter from the experimental point of view, it is believed that the chamfer radius has a significant influence on the randomness in the experimental results. Definitely, the eight-star fanned-out configuration presented the best performance with an experimental bond strength close to the tensile strength. In any case, all the investigated system configurations are very sensitive to handmade preparation which led to a significant randomness of the experimental results.

Acknowledgements The financial support from the Politecnico di Milano – Italy is gratefully acknowledged. The experiments were performed at Laboratorio Prove Materiali Strutture e Costruzioni of the Politecnico

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026

14

F.G. Carozzi et al. / Composite Structures xxx (2017) xxx–xxx

di Milano. Thanks are also expressed to Sika Italia S.p.A. for providing the materials used in the experimental campaign. References [1] Bakis CE, Bank LC, Brown VL, Cosenza E, Davalos JF, Lesko JJ, et al. Fiber reinforced polymer composites for construction - State-of-the-art review. ASCE J Compos Constr 2002;6(2):73–87. [2] Silva MAG, Biscaia H. Degradation of bond between FRP and RC beams. Compos Struct 2008;85(2):164–74. [3] Karbhari VM, Zhao L. Issues related to composite plating and environmental exposure effects on composite-concrete interface in external strengthening. Compos Struct 1998;40(3):293–304. [4] Ceroni F, Pecce M, Matthys S, Taerwe L. Debonding strength and anchorage devices for reinforced concrete elements strengthened with FRP sheets. Compos Part B 2008;39(3):429–41. [5] Bocciarelli M, di Feo C, Nisticò N, Pisani MA, Poggi C. Failure of RC beams strengthened in bending with unconventionally arranged CFRP laminates. Compos Part B 2013;54:246–54. [6] Colombi P, Fava G, Poggi C. End debonding of CFRP wraps and strips for the strengthening of concrete structures. Compos Struct 2014;111(1):510–21. [7] Valluzzi MR, Oliveira DV, Caratelli A, Castori G, Corradi M, De Felice G, Zaja˛ B, Zuccarino G, et al. Round Robin Test for composite-to-brick shear bond characterization. Mater Struct 2012;45(12):1761–91. [8] Carozzi FG, Colombi P, Poggi C. Calibration of end-debonding strength model for FRP-reinforced masonry. Compos Struct 2014;120:366–77. [9] Kalfat R, Al-Mahaidi R, Smith S. Anchorage devices used to improve the performance of reinforced concrete beams retrofitted with FRP composites: state-of-the-art review. J Compos Constr 2013;17(1):14–33. [10] Grelle SV, Sneed LH. Review of anchorage systems for externally-bonded FRP laminates. Int J Concr Struct Mater 2013;7(1):17–33. [11] Kalfat R, Al-Mahaidi R, Williams G. Investigation of efficient anchorage systems for shear and torsional retrofitting of box girder bridges. Proc. of the 10th Int. Symp. on Fiber Reinforced Polymer for Reinforced Concrete Structures, FRPRCS-10 (CD-ROM). London: Thomas Telford Publishing; 2011. [12] Orton SL, Jirsa OJ, Bayrak O. Design considerations of carbon fiber anchors. J Compos Constr 2008;12:608–16. [13] Niemitz CW, James R, Breña S. Experimental behavior of carbon fiberreinforced polymer (CFRP) sheets attached to concrete surfaces using CFRP anchors. J Compos Constr 2010;14(2):185–94. [14] Zhang HW, Smith ST. (). Influence of FRP anchor fan configuration and dowle angle on anchoring DRP plates. Compos Part B 2012;43:3516–27. [15] Smith ST, Zhang H, Wang Z. Influence of FRP anchors on the strength and ductility of FRP–strengthened RC slabs. Constr Build Mater 2013;49:998–1012. [16] Kalfat R, Al-Mahaidi R. Development of a hybrid anchor to improve the bond performance of multiple plies of FRP laminates bonded to concrete. Constr Build Mater 2015;94:280–9.

[17] Kalfat R, Al-Mahaidi R. Mitigation of premature failure of FRP bonded to concrete using mechanical substrate strengthening and FRP spike anchors. Compos Part B 2016;94:209–17. [18] Borri A, Castori G, Corradi M. Intrados strengthening of brick masonry arches with composite materials. Compos Part B 2011;42:1164–72. [19] Caggegi C, Pensée V, Fagone M, Cuomo M, Chevalier L. Experimental global analysis about the efficiency of carbon fiber anchors applied over CFRP straightened bricks. Constr Build Mater 2014;53:203–12. [20] Fagone M, Ranocchiai G, Caggegi C, Briccoli Bati S, Cuomo M. The efficiency of mechanical anchors in CFRP strengthening of masonry: an experimental analysis. Compos Part B 2014;64:1–15. [21] Fagone M, Ranocchiai G, Briccoli Bati S. An experimental analysis about the effects of mortar joints on the efficiency of anchored CFRP-to-masonry reinforcements. Compos Part B 2015;76:133–48. [22] Caggegi C, Chevalier L, Pensée V, Cuomo M. Strain and shear stress fields analysis by means of Digital Image Correlation on CFRP to brick bonded joints fastened by fiber anchors. Constr Build Mater 2016;106:78–88. [23] ASTM D638-03, Standard Test Method for Tensile Properties of Plastics, ASTM International, West Conshohocken, PA, 2003. [24] Carozzi FG, Milani G, Poggi C. Mechanical properties and numerical modelling of Fabric Reinforced Cementitious Matrix (FRCM) system for strengthening of masonry structures. Compos Struct 2014;107:711–25. [25] UNI EN 1015-11:2007. Methods of Test for Mortar for Masonry – Part 11: Determination of Flexural and Compressive Strength of Hardened Mortar; 2007. [26] UNI EN 772-1:2015. Methods of test for masonry units – Part 1: Determination of compressive strength; 2011. [27] UNI EN 14580:2005. Natural stone test methods - Determination of static elastic modulus; 2005. [28] UNI EN 12390-6:2010. Testing hardened concrete. Part 6: Tensile splitting strength of test specimens; 2010. [29] UNI EN 1542:1999. Pull Off Test Method; 1999. [30] UNI EN 10080:2005. Steel for the reinforcement of concrete – weldable reinforcing steel – general; 2005. [31] Pepe M, Mazaheripour H, Barros J, Sena-Cruz J, Martinelli E. Numerical calibration of bond law for GFRP bars embedded in steel fibre-reinforced selfcompacting concrete. Compos Part B 2013;50:403–12. [32] Fava G, Carvelli V, Pisani MA. Remarks on bond of GFRP rebars and concrete. Compos Part B 2016;93:210–20. [33] ABAQUS standard user’s manual, rel. 6.13. Dassault Systémes Simulia; 2013. [34] Fava G, Carvelli V, Pisani MA. Mechanical behaviour modelling of a new anchor system for large diameter GFRP bars. Compos Part B 2012;43(3):1397–404. [35] Rybicki EF, Kanninen MF. A finite element calculation of stress intensity factors by a modified crack closure integral. Eng Fract Mech 1977;9(4):931–8. [36] Krueger R. The virtual crack closure technique: history, approach and applications. NASA/CR-2000- ICASE Report No. NASA/CR-2002e211628; 2002. [37] Krueger R. Development and application of benchmark examples for mixed mode I/II quasi-static delamination propagation predictions. NASA/CR2012e217562. NIA Report No. 2012-01.

Please cite this article in press as: Carozzi FG et al. Mechanical and bond properties of FRP anchor spikes in concrete and masonry blocks. Compos Struct (2017), http://dx.doi.org/10.1016/j.compstruct.2017.02.026