Quasi-static cyclic in-plane testing of masonry walls strengthened with a single-sided fabric-reinforced cementitious matrix overlay and flexible anchorage

An experimental program was undertaken to assess the effectiveness of a new retrofit concept to improve the in-plane behavior of unreinforced clay brick walls by means of full-scale static-cyclic in-plane tests. The proposed seismic retrofit system combines two standalone retrofit measures for in-plane and out-of-plane strengthening of masonry walls. The in-plane reinforcement consists of a single-sided carbon Fabric-Reinforced Cementitious Matrix (FRCM) overlay, and anchors embedded with a flexible adhesive in the masonry. The out-of-plane reinforcement, which consisted of deep mounted Carbon Fibre Reinforced Polymer (CFRP) strips embedded with a flexible adhesive in the masonry, was included in the study to investigate the possible degrading effects of the deep groove on the in-plane behavior. A total of nine full-scale reinforced masonry walls with three different geometries were tested under three different axial loads. None of the specimens showed shear failure at both the reinforced and the as-built side. Cracking predominantly occurred at the interface between the bottommost bed-joint and the foundation beam. The out-of-plane reinforcement did not affect the in-plane strength, as no vertical shear cracks occurred. Moreover, it was found that the anchors increased both the rocking and sliding resistance of the walls. An analytical model was proposed covering the rocking and sliding resistance of the reinforced walls, providing a good approximation of the experimentally obtained in-plane strengths. Additional pull-out experiments showed that the testing scenario where the tensile forces in the anchor were transferred to CFRP strip, provided a good approximation of the analytically determined anchorage strength.


Background
In Groningen, an area in the Northeast of the Netherlands, earthquakes occur because of gas production from the Groningen field. Decades of gas production led to the depletion of the pressure of hydrocarbon gas within the reservoir pore space, causing the reservoir to compact. In turn, this compaction increases the mechanical loads acting on preexisting geological faults within and close to the reservoir. Some small fraction of these faults become unstable and are therefore prone to slip. Abrupt slip on such a fault results in an earthquake that radiates seismic energy [1]. Although the magnitude of these induced earthquakes on Richter's scale is relatively low (< 3.5), they have a big impact on the buildings in the region due the soft surface soils in the area and the shallow depth (3 km beneath earth's surface) at which they occur [2]. As the majority of the buildings in Groningen are composed of cavity walls with single whyte load bearing walls of unstrengthened clay brick masonry, and are designed to only resist wind loads, it is essential to improve the earthquake resistance of the existing buildings in the area to prevent collapse, with likely casualties. This definition of the problem in Groningen is identical to the first part of the introduction of the previous work of the authors [3]. The remaining part of the introduction contains duplicated (modified) passages from the previous work of Türkmen, De Vries, Wijte and Vermeltfoort [3].
A broad range of strengthening techniques for enhancing the capacity of Unreinforced Masonry (URM) walls are available nowadays. Traditional strengthening methods such as reinforced concrete jacketing and steel frames, however, add considerable mass to the structure, are labor intensive, and generally alter the esthetics of a building [4]. A system of cement-based mortar matrix reinforced by continuous dry-fiber fabric is proposed for retrofitting masonry walls [5][6][7][8] and it is becoming more and more in use especially for in-plane strengthening. This strengthening system is known under different appellations: Fabric-Reinforced Cementitious Matrix (FRCM), Textile Reinforced Mortar (TRM), Textile-Reinforced Concrete (TRC) and Fiber Reinforced Cement (FRC) [9]. The FRCM strengthening technique is a sound alternative to FRP epoxy-bonded materials for reinforcing existing masonry elements for several reasons: (a) higher compatibility of the cement-based or lime-based mortar with the masonry supports than the epoxy adhesives, (b) higher adaptability to not perfectly plane surfaces (i.e. masonry textures with mortar joints or blocks with irregular shapes and dimensions, (c) low sensitivity to debonding phenomena [8]. Other advantages are breathability, fire resistance, reversibility, safety for the operators, better quality in terms of reversibility of the intervention in the case of historical and preserved buildings, and advantages in term of cost and time of installation [9][10][11][12][13].
Past experimental programs assessing innovative strengthening systems based on cementitious mortar matrices, highlighted a significant improvement in both in-plane and/or out-of-plane lateral strengths of masonry walls [6,7,10,[14][15][16][17][18][19][20][21][22]. Based on the response of medium-scale clay brick shear walls, beam-column type walls and beam type walls subjected to cyclic in-plane loading, Papanicolaou, Triantafillou, Karlos and Papathanasiou [10] concluded that FRCM overlays provide a substantial gain in strength and deformability. The authors reported that FRCM jacketing is much more effective than FRP. The increased effectiveness is about 15-30% in shear walls, 135% in beam-column type walls and 350% in beam type walls, on the basis of tests conducted [10]. With an experimental campaign medium-scale walls, Babaeidarabad, De Caso and Nanni [14] showed that the increase in ultimate in-plane strength is proportional to the amount of FRCM and ranged between 2.4 and 4.7 times that of the unstrengthened specimens.
The authors reported that toe-crushing failure occurred for wallettes with a calibrated reinforcement ratio higher than 4%, and therefore increments of FRCM beyond this value were claimed ineffective [14]. Ismail [15] investigated the in-plane behaviour of medium-scale clay brick shear walls strengthened with different types of FRCM systems. The shear strength of single-sided retrofitted wallettes ranged from 113 to 148%, whereas the shear strength of test wallettes with a double-sided FRCM retrofit ranged from 446 to 481% compared with the strength of the unstrengthened wallettes. The lower increase in shear strength for the singlesided retrofitted specimens was attributed to the unrestrained boundary conditions of the diagonal compression tests, as these specimens showed out-of-plane bending behaviour. This out-of-plane bending during the diagonal compression testing of single-sided FRCM retrofitted specimens, caused by the eccentric stiffness resulting from the application of the reinforcement on a single side, was also reported in other researches [16,17]. Shabdin, Zargaran and Attari [16] tested medium-scale URM walls strengthened with FRCM under diagonal compression in order to consider the effect of strengthening system on the behavior of brick walls. The authors concluded that FRCM improved the diagonal load carrying capacity and deformation capacity, which caused the strengthened walls fail in a ductile manner. Marcari, Basili and Vestroni [17] investigated the effectiveness of using a Basal TRM system for in-plane shear reinforcing of volcanic tuff stone masonry. The average increase in shear strength was approximately 40% for the single-side reinforced specimens, while the increase was 60% with doubleside reinforced specimens. The authors also reported that the TRM system changed the failure mode of the panels from joint-sliding to diagonal cracking. Another experimental campaign carried out on medium-scale tuff-masonry walls also showed that strengthened walls did not fail for the characteristic diagonal sliding fracture, generally developing in unreinforced masonry walls at the mortar-to-brick interface [18]. The ultimate load in diagonal compression (and the corresponding shear strength) for strengthened walls were reported to be between four and six times greater than the one observed for bare walls [18]. Ismail and Ingham [19] conducted an experimental program with full scale reversed cyclic in-plane testing of FRCM strengthened URM walls. The authors observed the strength increment due to TRM strengthening to range from 128% to 136% when the URM test walls were loaded in-plane, with a notable increment in deformation capacity and ductility.
For the out-of-plane strengthening of masonry walls the FRCM system requires application to both faces of a wall. In case of strengthening of load bearing leafs of cavity walls, this would require the temporary rehousing of the occupants in order to install the FRCM layer from within the building, as well as the removal of the façade of the building for the installation of the FRCM layer on the cavity facing side of the load bearing wall. Cost effective retrofitting can be enhanced by installing deep mounted CFRP out-of-plane reinforcement, where FRP strips or rods are inserted into deep vertical grooves cut in the surface of a wall. The CFRP strips offer additional out-of-plane flexural strength to the wall for both out-of-plane loading directions whilst only installing the reinforcement from one side of the wall, leading to costeffective retrofitting [23,24]. This system uses a viscouselastic epoxy is instead of a conventional stiff epoxy for the installation of the FRP strips, because the implementation of a conventional stiff adhesive for deep mounting the CFRP strips results in premature splitting failure in the masonry [25]. Next to making deep mounting possible, using a viscous-elastic epoxy (Young's modulus < 50 N/ mm 2 ) instead a conventional stiff adhesive (Young's modulus ~ 10,000 N/mm 2 ), a significant increase in terms of ductility and maximum withstandable load is reached and critical crack development is prevented [23,24,26]. Similar findings were previously reported by Kwiecień [27] and Derkowski, Kwiecień and Zajac [28].
A schematic overview of the proposed combined reinforcement concept is shown in Fig. 1. While the influence of FRCM reinforcement on the in-plane behaviour of masonry wallettes has been a popular subject of research for the past years, the influence of the aforementioned combination of retrofit measures on the in-plane shear capacity of masonry has not been investigated before. Moreover, this experimental program aims to investigate the possible degrading effect of the proposed out-of-plane strengthening system on the in-plane shear strength of masonry panels. The aim of the experimental campaign for this study was to determine the effectiveness of a one-sided FRCM reinforcement combined with DM CFRP strips to improve the in-plane shear resistance of clay brick masonry walls. In this experimental campaign, static-cyclic in-plane shear tests were performed (cantilever configuration) on full-scaled masonry walls strengthened with this combined reinforcement system. Additional pull-out experiments were conducted to gain more insight on the behavior of foundation anchors, embedded in the same flexible adhesive as the CFRP strips. Finally, the experimental results were compared to the outcomes of existing design codes, to check the validity of these models for this combined retrofit system.

Materials and characterization
This materials and characterization section is mostly identical to §2.1 of the previous work of the authors [3]. The clay bricks used in this research had dimensions of 205 (± 4) × 95 (± 2) × 50 (± 2) mm 3 l b × w b × h b . Several mechanical characteristics of the used clay bricks were determined with an experimental program conform the corresponding standards. The bricks had a mean compressive strength f c of 31.7 N/mm 2 (n = 12; COV = 7.4%), where n is the amount of tested specimens respectively. The compressive strength of the clay bricks were determined following the EN 772-1 [29] standard using gypsum capping and half bricks. The mean splitting tensile strength f st (determined conform ASTM C1006-07 [30]) and flexural tensile strength f t (obtained following ASTM C67-03 [31]) of the bricks were found to be 3.34 N/mm 2 (n = 12; COV = 8.7%) and 5.89 N/ mm 2 (n = 9; COV = 7.4%). The mean flexural tensile strength of the mortar specimens was found to be 3.6 N/mm 2 (n = 8, COV = 16.5%), and the mean compressive strength of the mortar was 10.6 N/mm 2 (n = 16; COV = 20.7%). Both the flexural tensile strength and the compressive strength of the mortar specimens were determined according to EN 1015-11 [32]. Compression tests were performed on three masonry specimens, consisting of 6 brick high masonry prisms, under displacement control with a loading speed of 0.20 mm/min. The average compressive strength f m of the specimens was 14.8 N/mm 2 (COV = 6.1%). The Young's modulus was determined as a secant modulus at 35% of the compressive strength in accordance with EN 1052-1 [33]. The average Young's modulus (E) of the masonry prims was found to be 3100 N/mm 2 (COV = 2.5%).
In order to determine the mechanical properties of the masonry under shear in accordance with EN 1052-3 [34], a total of 9 triplet shear tests was performed at three different normal stress levels: 0.2, 0.6 and 1.0 N/mm 2 . For each specimen the relation between the applied normal stress and the shear strength has been established. The ratio between the compressive stress and the shear strength of the masonry was obtained using a linear regression. The parameters for the Coulomb's friction criterion follow from Eq. (1): with f v,0 , ma and n being the initial shear strength, friction coefficient and axial load respectively. The residual shear strength f v,0,res and residual coefficient of friction ma,res were determined by applying the same linear regression analysis when a plateau was reached in the post-peak phase. The mechanical properties of the materials used for building the specimens for this study are summarized in Table 1.
The obtained values regarding the mechanical properties of masonry under shear correspond well with the results of the study carried out by Jafari, Rots, Esposito and Messali [35] on the material properties characterization of Dutch URM, and the values proposed in the Dutch Practical Guideline for the seismic assessment of local buildings in Groningen, NPR9998 [36].
The reinforced mortar used for the mortar matrix of the FRCM overlay was a polymer modified mortar based on organic binders, polymer fibres and selected aggregates, with a maximum grain size of 1.8 mm. The dry mortar and polymer fibres are shown in Fig. 2a. The additional reactive components, which were mixed into the reinforcement mortar, bonded with the amorphous silica on the CFRP mesh. This ensured an improved adhesion between the mesh and (1) f v = f v,0 + ma n the cementitious matrix. For the preparation of the reinforced mortar, a plastic bonding agent was used in order to improve the adhesion of the cementitious matrix to the clay brick substrate. This was done by mixing 110 g of the plastic bonding agent per 10 kg of prepared mortar. The reinforced mortar was prepared following the manufacturer's instructions by adding 2.6 L of water to a bag of 25 kg dry mortar. Both the flexural tensile strength and the compressive strength of the reinforced mortar specimens were determined according to EN 1015-11 [32]. The average flexural tensile strength of the reinforced mortar specimens was found to be 7.58 N/mm 2 (n = 9; COV = 11.7%). The mean compressive strength of the reinforced mortar was 62.6 N/mm 2 (n = 12; COV = 1.6%). The weight density was 2138 kg/m 3 (n = 6; COV = 1.7%).
The bidirectional CFRP mesh, with a fibre weight density of 1.79 g/cm 3 and about 3 mm width per thread, had a square aperture dimension of approximately 50 × 50 mm 2 . The theoretical cross section of the carbon fibre for design was 44 mm 2 . The Young's modulus, tensile strength and elongation at rupture of the mesh as provided by the supplier are > 240 kN/mm 2 , > 4300 N/mm 2 and 1.75% respectively for the carbon FRP mesh. The prefabricated (pultruded) CFRP strips were 20 mm in width and 1.4 mm in thickness and have a fibre volume content of > 68%. The Young's modulus, tensile strength and elongation at rupture of the CFRP strip was found to be 215 kN/mm 2 , 2876 N/mm 2 and 1.59% respectively.
The material properties for the two-component viscouselastic adhesive were obtained following ISO 527-1 [37] using three specimens at a loading rate of 10 mm/min. The Young's modulus was determined as the secant modulus between 0.5% and 5% of the tensile strength, and was found to be 16.0 N/mm 2 (COV = 1.7%). The tensile strength and elongation of rupture were determined as 4.3 N/mm 2 (COV = 0.9%) and 72.1% (COV = 3.5%).

Building the in-plane test specimens
The wall specimens for the static-cyclic in-plane shear tests were built in the testing laboratory of QuakeShield in Grijpskerk, the Netherlands. The specimens were built on a reinforced concrete foundation beam by an experienced mason. The masonry walls were nominally 2450 mm high (h w ) and had a thickness (t w ) of 100 mm. The lengths of the specimens (l w ) were 1100 mm (for the S specimens), 2000 mm (for the M specimens) and 4000 mm (for the L specimens). For each configuration three specimens were built in order to test the effect of axial load q v on the in-plane behavior for each geometry. The axial loads were 0.15, 0.3 or 0.5 N/mm 2 , with the only exception being specimen S1 (0.20 N/mm 2 ), due to control difficulties of low axial forces. An overview of the test specimens is provided in Table 3. The concrete foundation beam had a height (h c ) of 180 mm and a width (t c ) of 180 mm. It should be noted that the reinforced concrete beam used for the M-specimens was 60 mm shorter than the length of these wall specimens. These wall specimens therefore extended 30 mm over the foundation beam at both ends. The mortar for the masonry specimens was prepared in the laboratory and the walls were built in running bond. Both the bed and head joints had a nominal thickness of 12 mm and were fully filled. All walls cured in air in the unheated laboratory (0-10 °C) for at least 28 days before strengthening.  Table 4. After the walls were sufficiently cured (± 28 days) the strengthening process started by milling a number (n g ) of vertical grooves (n g = 2, n g = 3 and n g = 5 for S, M and L specimens respectively) of 65 mm depth (d f ) and 10 mm width (b f ), spaced 850 mm (l strip ) apart. The distance of the outer grooves to the edges (l edge ) was 150 mm for the S and M specimens and 300 mm for the L specimens. The dust in the groove was removed with compressed air. The CFRP strips with a cross-section of 20 × 1.4 mm 2 (b p × t p ) were cleaned with acetone after cutting the strips into the specified length. A layer of primer was then applied to the groove to obtain an improved bond of the adhesives to the masonry. Afterwards, the CFRP strips were inserted into the groove that was partially filled with the flexible adhesive. The vertical grooves at the side ends were widened to 25 mm (b f ') till a depth of 35 mm (d f' ) and a height of 500 mm. In each specimen two 12 mm (D s ) ribbed reinforcement bars (B500B) with a length of 650 mm (l a ) were fixed with a conventional stiff adhesive (HIT-HY 100) inside a borehole of 150 mm in depth (l a,stiff ) and angle of 30° degrees in the foundation beam. The remaining part of the steel rods were embedded in the visco-elastic adhesive within the aforementioned widened grooves over a length of 500 mm (l a,flexible ). Excess adhesive till a depth of 30 mm (d ff ) in the standard grooves and 10 mm (d ff ') in the widened grooves was removed using a scraper.

Reinforcing the in-plane test specimens
After the placement of the strips and anchors, the walls were left to cure for one day before applying the single-sided FRCM overlay. The masonry surface was wetted prior to the mortar matrix application to secure proper adhesion. A thin layer of mortar was subsequently applied to the masonry surface by hand. The remaining parts of the grooves were also filled with the same mortar. The CFRP mesh was then applied on the mortar matrix surface and was pressed into the mortar matrix. After placing the CFRP mesh in the mortar a new thin layer of mortar was applied to embed the CFRP mesh, resulting in a nominal FRCM layer thickness of 15 mm (t FRCM ). The specimens were left to cure in the laboratory environment for a minimum of 28 days before starting the experimental program.

In-plane test setup
The static-cyclic in-plane shear tests were performed in the test setup of QuakeShield. The test setup is illustrated in Fig. 6, and shown in picture in Fig. 7. The frame of the test setup is formed by two post-aligned steel frames which were interconnected. The basis of the frame is a rectangular closed portal frame (A in Fig. 4) in which the specimen was placed. The bottom beam of it passes through the portal frame and supported the steel shore (C) that connects to the portal frame at the position of the horizontal actuator. Steel beam B in Fig. 4 takes care of the stability of the test setup in transverse direction. All steel profiles of the test frame are interconnected by weld connections thus minimizing movement in the connections. The test specimens were supported at the bottom side by a foundation formed by a reinforced concrete beam. This foundation beam was connected to a 15 mm steel plate with a fast-curing epoxy. The steel plate allowed for a connection between the specimen and the test setup. Besides the epoxy-connection with the bottom steel plate, the foundation beam was also kept in place by a mechanical connection (steel brackets) that covered the topside of the beam.
At the topside the specimens were provided with a steel plate which enabled a connection between the specimens and the loading beam. In order to ensure a uniform distribution of the vertical load, rubber pads were placed between the loading beam and the steel plate which connected the specimens to the test setup. This steel plate was in turn provided with 40 mm thick steel blocks which were positioned such that they fitted exactly between the rubber pads. These steel blocks were used to transfer the horizontal load from the loading beam to the specimen (Fig. 6). The steel plates were attached to the specimens with a fast-curing adhesive after lowering the loading beam. After a minimum curing time of 24 h, the experiment was started.
The horizontal load was generated by a horizontally positioned hydraulic actuator with a capacity of 500 kN. The horizontal load was applied in both West direction (also referred to as push direction) and East direction (also referred to as pull direction). A loading beam distributed this horizontal Table 2 Mechanical properties of the materials used for reinforcing the specimens a Flexural, b Stresses calculated with respect to the cross-sectional area of the reinforced mortar, c Stresses calculated with respect to the crosssectional area of the CFRP mesh, d n = 3; COV = 1.7%, e n = 3; COV = 0.9%, f n = 5; COV = 1.0%, g n = 5; COV = 3.4%, h n = 5; COV = 14.7%, i n = 6; COV = 1.7%, j n = 9; COV = 11.7%, k n = 12; COV = 1.6%, l n = 8; COV = 6.9%, m n = 9; COV = 14.9%, n n = 8; COV = 13.5%, o n = 6; COV = 15.3%, p n = 9; COV = 10.2%, q n = 6; COV = 19.4%, r ISO 527-1 [37], s EN 1015-11 [32], t AC434.13 [38] (kg/m 3   load over the topside of the specimen. For the sake of symmetry in the push and pull cycles the force exerted by the horizontal actuator engaged in the center of the load beam using two steel arms. The load beam was stiffened at the top with an IPE-profile on which two vertically positioned hydraulic actuators with a capacity of 200 kN each, provided the desired vertical load on the test specimen. At the topside the vertical jacks were fixed in position against the frame of the test setup. Because the load beam would translate horizontally throughout the test the vertical jacks were connected to the beam by means of crane trolleys. The crane trolleys were in turn connected to the hydraulic actuator by a load pin, which also enabled rotation of the crane trolley. This load pin also monitored measured the load exerted by the actuators. Both load pins were calibrated to a maximum force of 120 kN and had an accuracy of 0.3%. The horizontal load which was exerted by the horizontal actuator was monitored by a load cell positioned at each of the two steel arms which connects the actuator to the center of the loading beam. In addition to these load measurement devices the horizontal and vertical loads were checked by measuring the oil pressure within the hoses of the hydraulic power pack. Displacement and deformation measurements were conducted by laser sensors with an accuracy of 7 microns. Laser sensor 12 in Fig. 6 was used for controlling the test and for constructing the force-displacement plots. This sensor was connected to the steel plate on top of the specimens and displacements were measured relative to a detached frame. Vertical deformation measurements were conducted on the small specimens. These deformations were measured by means of two laser sensors which were applied on telescopic tubes (8 and 9). The telescopic tubes covered nearly the complete height of the specimens and monitored the vertical deformation of the specimen. Shear deformations were measured on the medium and large specimens by laser sensors on diagonally positioned telescopic tubes (6 and 7). Slip of the wall specimens over the bottommost bed joint was measured by a laser sensor positioned on a  (5). The same frame was used for the positioning of a laser sensor which measured the displacement of the East top of the specimen (4). Possible vertical uplift at the position of the steel anchors was measured by two laser sensors placed on detached measurement frames (10 and 11). All measuring devices were connected to a PLC-system which processed the data in real time. This data was then forwarded by the PLC-system to a laptop and monitored by the researcher. During the tests the nature and extent of the cracking pattern was continuously observed and noted. The cantilever configuration was chosen for this testing program, were vertical deformation and rotation of the loading beam is not restrained (i.e. free rotation at the top). The vertical force of both actuators was kept constant during the complete course of the tests. The vertical loads exerted by both actuators was therefore not depending on the exerted horizontal load nor the horizontal displacement. After applying the vertical pre-compression load the wall specimens were subjected to cyclic shear loading. The cyclic shear load was applied using computer controlled displacement steps, starting from 0.26 mm with a speed of 0.2 mm/s. Each displacement step was applied two times in both loading direction forming one load cycle, before increasing the target displacement with 40%, based on the FEMA [39] protocol. This continued until a target displacement of ± 40 mm, which was the maximum stroke of the horizontally oriented hydraulic actuator.

Direct pull-out tests
The force in the anchors, especially during the rocking behavior of a reinforced specimen, is a relevant parameter for modelling purposes. With the test setup presented in the previous paragraph and the limitations in the lab, the measurement of the force in the anchors was difficult to realize. In order to gain more understanding on the mechanical behavior of the flexible anchor connection, additional small-scale direct pull-out tests were performed. Two possible scenarios were considered as shown in Fig. 8, regarding the transfer of the tensile forces in the anchor into the reinforced specimen, Only anchors were installed on the DPT-A specimens. The anchors installed in the DPT-B specimen, were bonded to the bottom 490 mm of the specimen. This was done to maintain a consistent anchorage length with the anchors of the DPT-A specimens. The installed CFRP strip on the DPT-B specimen, were bonded over the entire specimen length of 740 mm. The specimens are illustrated in Fig. 9. In experimental conditions, scenario A was realized by restraining the masonry prism using a steel plate. Scenario B was realized by clamping the extended piece of the CFRP strip, rather than using a steel plate. An overview of the direct pull-out test setup is provided in Fig. 10. The direct pull-out tests were conducted on a 250 kN Instron universal testing machine (Fig. 11). For the DPT-A specimens (Fig. 11a), the installation process started by carefully positioning the prism under the loading grips of the testing equipment, with the specimen resting on two support blocks. Hard cardboard was put on the top of the prism to prevent stress concentrations due to a possible non-flat surface of the brick. Afterwards the steel restrain plate at the loaded end was placed on top of the prism. To prevent undesirable wedge type failure modes when using partial end restraint, a full restraint in the form of a 25 mm thick solid steel plate with three openings was selected. The centrally located opening allowed the loaded end of the ribbed steel anchor to pass through. The smaller two openings allowed the LVDT (linear variable differential transformer) sensors to rest on the specimen. The specimen was then lifted up, which made it possible for the prism to find its own balance point and thus minimize the eccentricity caused by imperfect installation of the anchor. Using M12 threaded steel rods, the steel restraint plate was bolted tightly to the base of the setup. Four LVDT's were installed prior to the load application process. The upper two LVDT's (1 and 2) measured the loaded end slip, whereas the bottom two LVDT's (3 and 4) measured the free end slip.
The installation process of the DPT-B specimens (Fig. 11b) was slightly different from the DPT-A specimens. Aluminium plates of 100 mm in length, 20 mm in width and 2 mm in thickness were used to tab the CFRP strips. After the tabs were roughened with sandpaper and thoroughly cleaned with acetone, the tabs were glued to both sides of the extending CFRP strip using high strength and fast curing 2-component epoxy adhesive.
The prism was positioned under the loading grips of the testing equipment, with the specimen resting on two support blocks. The specimen was then lifted up via the aluminium grip plates. The extending anchor was clamped using the grips at the base of the installation. Prior to the load application process, four LVDT sensors were installed. The upper two LVDT's (1 and 2) measured the loaded end slip of the CFRP strip, whereas the bottom two LVDT's (3 and 4) measured the loaded end slip of the steel anchor. The mean of two LVDT's was used to establish the corresponding slip. The pre-tension load was 1.0 kN for all specimens. After resetting the sensors, the experiment was started with a pull-out speed of 0.5 mm/min on the loading grips.

Test results
This chapter will provide an overview of the obtained results during the experimental campaign. The first five paragraphs focus on the in-plane shear experiments, covering overall results, failure modes, strength and drift values, wall response parameters and finally the uplift of the anchors. The final paragraph covers the direct pull-out experiments of both load transfer scenarios.

Overview in-plane shear test results
A summary of the obtained in-plane test results is given in Table 5. The maximum forces in both the pull (H max,east ) and push cycles (H max,west ) is reported. Moreover the maximum displacement δ max , the maximum drift Δ max [calculated in accordance with Eq. (2)] and the failure modes are shown. These parameters are discussed in the following paragraphs.

Failure modes in-plane shear tests
The failure modes of all specimens are illustrated in Fig. 12. None of the specimens showed any shear damage at the treated or untreated surfaces. As failure only occurred at the bottom side of the specimens, the upper part of the walls is not shown in the illustrations in Fig. 12. All tested specimens showed initial cracking at the bottom corners. These initial cracks propagated over the complete length of the specimen during subsequent loading cycles. Cracking mainly occurred at the interface between the bottommost bed-joint and the concrete foundation beam. The expected formation of cracks due to the vertical shear stress concentrations caused by the deep grooves (needed for the out-of-plane reinforcement) did not occur.
All specimens started to show rocking behaviour during the tests. For specimens S3, M1, M2, M3 rocking of the specimen was followed by crushing of the specimens' bottom corners. The uplift of the specimen had the effect of reducing the compressive zone which eventually led to toe-crushing. Due to the relatively low compression forces acting on specimen S1 and S2 these specimens did not show compressive failure.
For specimen S1, first cracking was observed at a horizontal force of about 8 kN in push direction. During the pull cycles not only similar cracking of the bed joint took place but also partial detachment of the concrete foundation beam from the steel plate occurred. Consequently some uplift of the West side of the foundation beam was observed due to (2) Δ max = max ∕h w Fig. 8 Possible stress transfer scenario's from the anchor to the masonry/CFRP strip rocking of the complete specimen-foundation assembly during pull cycles. This partial detachment of the foundation beam was not observed at the East side during any of the push cycles, leading to different force-displacement behavior in push and pull cycles. It was found that the bracket at the West side of the specimen, which should keep the foundation beam in place was not tightened sufficiently. For specimen M1, uplifting of the foundation beam was observed during the pull cycles. After observing this malfunction, the brackets were tightened further. This detachment of the foundation beam was not observed during any of the push cycles. During subsequent cycles of the L-specimens, both rocking and sliding of the specimen was observed. Subsequently detachment of the bottom corners next to the steel anchors occurred.

Strength and drift
The hysteresis loops and backbone curves of all the specimens are provided in Fig. 13, with grey and black colored lines respectively. For all the specimens, a sudden drop in force was observed after the displacement target was reached. This sudden drop was likely a limitation in the hydraulic equipment preventing a smooth transition when the cylinder had to be moved to the opposite direction. The test on specimen S1 was aborted due to the detachment of the foundation beam at a target displacement of 28 mm.
The tests on specimens L2 and L3 were stopped due to significant damage development in the region surrounding the anchors. The test on the specimens S2, S3, M1, M2, M3 and L1 were stopped at a target displacement of 40 mm as the maximum stroke of the actuator was reached.
The difference in strength between the West and East side of the specimens, was less than 16% (West side with respect to East side) for all tested walls, except for S2, where the difference was over 50%. From the hysteresis loop of S2 it can be concluded that the East anchor was activated during the push cycles. This was accompanied by a significant increase in the walls capacity in the push direction. Minimal to none activation of the anchorage occurred during the pull cycles. This is expressed in the lower capacity of the wall in the pull direction. After maximum activation of the anchor in push direction, the force in the load-uplift plot drops back to almost the same value as for the pull cycle. Faulty installation of the anchor could be a possible explanation why the West anchor was minimal to none activated for S2, and thus led to a difference of more than 50% in strength between both sides. The east anchors of both specimen M1 and M2 were pulled out from the foundation, leading to crack formation around the initial location of the anchors in the foundation beam.
Specimens S2 and M1 showed significant different loaddisplacement behavior between the East and West sides. Looking at the hysteresis loop of specimen S2 on the pull direction, no drop in force was monitored during the postpeak phase. This shows that the debonding process of the installed anchors was not initiated. As mentioned before, the asymmetry in the load-displacement diagram of specimen M2 was caused by the detachment of the foundation beam during the pull cycles. The West anchor did not reach full strength until a target displacement of 28 mm (versus 6.1 mm for the East anchor).
The axial load has a significant influence on the in-plane resistance of the specimens. For the M and L specimens, a linear correlation was found between the applied axial load and strength of the specimen, with R 2 values of 0.93

Wall response parameters
In-plane wall response parameters were calculated from the force-displacement graphs presented in Fig. 13. A summary of the parameters calculated is shown in Fig. 14.
The bilinearised ultimate wall force H u is determined in accordance with Eq. (3): The bilinearised initial stiffness K e is determined in accordance with Eq. (4): The wall's structural ductility factor is the ratio of the ultimate displacement of the wall over the bilinearised yield displacement following Eq. (5): where e is the bilinearised yield displacement and u is the wall's ultimate displacement corresponding to a 20% force drop in the post-peak phase. The wall's response parameters for the tested specimens are listed in Table 6.
The limited stroke of the horizontally oriented hydraulic actuator was insufficient to reach a 20% force drop in the post-peak phase for the specimens S1, S3, L2 and L3. For these specimens, the presented ductility factors in Table 6 are a lower boundary. For the push side of specimen S2, the wall's structural ductility factor was found to be 4.9. For the M specimens, the wall structural ductility factors were in the range of 3.7-14.7. It can be observed that higher axial loads have a positive effect on the mean ductility factor for the M specimens. For specimen L1, the mean ductility factor was found to be 4.2. Similar to the M specimens, increasing the axial load seems to increase the wall structural ductility factor. An increasing trend in bilinearised initial stiffness was observed as the axial load was increased, while keeping the geometry of the specimen constant.

Uplift anchors
The uplift of the walls at the moment of the maximum load is provided in Table 7. As the anchorage method was consistent for all specimens, the slip of the anchor (opening between foundation beam and reinforced wall) at maximum applied load was also expected to be consistent. This was not the case as a significant variation was found in the measured uplift values at the failure loads. The observed detachment of the foundation beam was a disturbing factor for the measured uplift of the West anchors of both specimen S1 and M1. The same applies for the West anchor of specimen S2 that did not enter the post-peak phase. The remaining variations in the uplift values, can be explained by the imperfect clamping of the foundation beam, the insufficient tightening of the steel transport plate to the bottom frame and/or the deformation of the used bolt and threaded rods, leading to an uplift of the foundation beam.

Overview pull-out test results
The results of the direct pull-out tests conducted on masonry prisms with imbedded anchors are shown in Fig. 15, where the pull-out load (pre-tension force not taken into account) is plotted against the loaded end slip of the anchors. The predominant failure mechanism was adhesive failure at the anchor-adhesive interface for both scenarios.
The anchorage strength F max , mean bond strength σ max and loaded end slip s max at maximum pull-out load were For the DPT-B specimens, these values were found to be 17.90 kN, 0.95 N/mm 2 and 3.1 mm respectively. The significant lower value in strength for the DPT-B specimens was likely caused by the relatively higher shear stresses in the adhesive mass between the CFRP strip and the anchor, compared to scenario A. These increased shear stresses could have expedited the crack initiation process. The mean initial stiffness k 50% , determined at 50% of the strength and the corresponding slip, was found to be 12.30 kN/mm (DPT-A) and 12.78 kN/   [40] mm (DPT-B). For the DPT-A-1 and DPT-A-3 specimens, a reduction in stiffness was observed at a loaded end slip of approximately 4 mm (s damage ). For specimen DPT-A-2, the damage initiation stage was entered earlier due to incorrect surface preparation (not made dust free and not made clean with acetone), and thus leading to a significant lower anchorage strength. For the DPT-B specimens, a reduction in stiffness was observed at a significant lower slip (1 mm). An overview of the obtained direct pull-out test results are provided in Table 8.
Looking at values for the corresponding slip at full strength, a major difference was found with the values presented in Table 6. This indicates that the aforementioned disturbing factors in the measurement of the uplift of the anchors, could have an even more significant contribution then previously assumed.

Analytical model based on Eurocode 6
When out-of-plane buckling is not considered, then there are two failure modes that can occur for walls loaded in-plane: moment failure/rocking or shear failure. The governing failure mode is influenced by the aspect ratio (height divided by length) and the ratio between the vertical and horizontal loading. The moment and shear resistance of reinforced walls were modelled using modified rules from Eurocode 6 [42]. The behavior and resistance of a reinforced wall is very much alike that of an unreinforced wall. This with the exception that a tension element (anchor) in vertical direction is present. Therefore some modifications on the existing equations for unreinforced masonry walls were made. The used mechanical model is presented in Fig. 16.

Moment failure
The ultimate eccentricity e u is determined using Eq. (6). The ultimate depth of the compression zone x u , following Eq. (7), is based on a bi-linear stress-strain relation for the masonry (Fig. 15) according to the Dutch NA [41] to Eurocode 6 [42]. where f m is the compressive strength of the masonry, F v,r is the tensile force in the anchors and F v is the axial force following Eq. (8). The own weight of the reinforced wall = 2.12 kN∕m 2 is also taken into account for the determination of the axial force.
The moment resistance M R follows from Eq. (9). The resistance is based on the equilibrium of the wall, which is influenced by the ultimate eccentricity e u of the vertical reaction force F v and the tensile force in the anchors F v,r The effective depth d of the anchors is determined in accordance with Eq. (10) Combing Eqs. (6-10) results in Eq. (11): The moment resistance M R that follows from the experiments is determined by Eq. (12): Combing and rewriting Eqs. (11) and (12), results in Eq. (13), which can be used to determine the tensile forces in the anchors according to the proposed model. The and factors in Eq. (13) are determined using Eqs. (14) and (15) respectively. (11) For the friction coefficient of clay brick masonry the value 0.75 was used, as determined with the triplet shear tests and proposed in NPR9998 [36]. This coefficient was also used to describe the shear resistance of the base joint between the masonry wall and the concrete base beam. It should be noted that this joint has different properties than a regular bed joint in masonry. Firstly, due to temperature effects and drying shrinkage of the mortar, the bond between concrete and masonry can be affected. Secondly, the bond will be completely gone ( f vo = 0) when the masonry is subjected to rocking. Also taking the additional compressive force into account, caused by the tensile force F v,r generated by the anchor in tension, the resistance against shear sliding at the base joint is determined in accordance with Eq. (19): When the contribution of the anchors is not taken into account (F v,r = 0) for the resistance (V R,s0 ) against shear sliding, Eq. (19) changes to Eq. (20):

Comparison between model and experiments
By using the input parameters as provided in Table 9, the horizontal resistance F Rh in case of no anchoring strength can be determined. The calculated resistance F Rh was compared with the maximum horizontal resistance F h,test found with the experiments for the specimens (S and M) that failed due to rocking. An overview of the results of these specimens are provided in Table 10. It should be noted that the strongest side (West or East) is considered.
The maximum horizontal load resistance F Rh (with the absence of anchors) for the S and M specimens are presented with dashed lines in Fig. 17. The added value of the anchors is set as the difference between the experimental backbone curve and the analytically obtained maximum horizontal load resistance F Rh . These areas are marked in light grey. It With the absence of anchors (F v,r = 0), the maximum horizontal load resistance F Rh follows from Eq. (16). This can be used to determine the added value of the anchors for the moment resistance of the wall.

Shear failure
With respect to shear failure again two situations have to be considered: • Shear failure in the masonry wall, where the resistance is based on the initial shear strength, the contribution from the normal stress and the effective depth of the compression zone. • Sliding of the masonry wall over the base, where the resistance is based on dry friction only as no cohesive strength will be present due to the rocking behavior.
According to Eurocode 6 [42] the shear resistance of an unreinforced masonry wall V R is derived with Eq. (17). Since no shear tension failure was observed in the compressed area of the tested specimens, the upper limit for the shear resistance is not taken into account: where l c is the length of the compressed area at the end section of the wall, f vo is the initial shear strength, ma is the average coefficient of friction of the masonry and v is the average compressive stress in the compressed part of the cross-section determined using Eq. (18)  Fig. 17, it is observed that the rocking capacity of specimen S2 was not influenced by the West anchor. Combined with the previously mentioned absence of the post-peak phase of the anchor bond, the conducted analyses indicate that the West anchor of specimen S2 was not activated during the experiment. Faulty surface conditions of the anchor (not cleaned and made dust-free) could not be the only explanation, as the direct pull-out results indicated that there still was some significant bonding capacity left in this case. Possible explanations for the West anchor not being activated could be the insufficient confinement of the anchor in the flexible adhesive and/or the faulty preparation of the flexible adhesive.
For the specimens where sliding over the base joint occurred (L specimens), the ratio between the maximum horizontal force on the specimen and the calculated shear sliding resistance V Rs ∕F h,test is < 1.0, as shown in Table 11. For the L-specimens, the dark grey areas in Fig. 17 indicate the difference between the maximum load during the in-plane experiments and the sliding resistance V Rs of the specimen including the extra compression force caused by the anchor (indicated by a dash-dot line). Comparing the determined values for the sliding resistance with and without the contribution of the anchors from Table 11, it can be observed that the additional compression force due to the anchors results in an increase of 14.8 kN (+24.5%) and 15.2 (+14.42%) in sliding shear resistance (following the analytical model) for specimens L1 and L2 respectively. For specimen L3 the sliding shear resistance for the two aforementioned cases remained the same. It should be noted that dowel action of the anchor in the compressed area was not considered for the values regarding the resistance of the base joint. The contribution of the anchors in creating an additional compression force in the joint will result in an extra resistance against sliding due to friction, hence probably explaining the specimen exceeding the calculated sliding resistance V Rs . Looking at the values for the rocking resistance F Rh in case of no anchors and the resistance V Rs against shear sliding in the base joint for specimens L1 and L2 (as provided in Table 11), it can be observed that the anchors first provide additional sliding resistance followed by an increased moment resistance, preventing the rocking failure of the specimen. The added value of the anchors is limited for both the rocking and sliding behavior of specimen L3, likely caused by the high axial load.
For the S and M specimens the determined tensile forces in the anchors using Eq. (13) ranged between 11.7 and 37.3 kN, with a mean value of 20.1 kN (COV = 38.7%).
Comparing the values for the analytically determined anchorage strength from the in-plane experiments with the conducted direct pull-out experiments, the calculated values predominantly fall between the lower bound (scenario B) and the upper bound (scenario A) as shown in Fig. 18. It can be concluded that scenario B, where the tensile forces are pre-dominantly transferred from the anchor to CFRP strip, provided a better approximation of the analytically determined anchorage strength. Testing conform scenario A is not advised in the current study, as the anchorage strength is significantly overestimated.
Taking 17.9 kN as the mean strength of the anchors (following the pull-out tests conform scenario B), the (moment) resistance of the wall F R can be determined for the S and M specimens. The results are shown in Table 12 and Fig. 19. The model/experimental ratio was in the range 0.68-1.08 and 0.90-1.07 for the S and M specimens respectively. The model/experimental ratio of 1.4 for specimen S2 (east) was not included in Fig. 19 as the West anchor minimally to not activated, as was shown in Fig. 17.
With the mean strength of the anchors (following the pullout tests conform scenario B), the (sliding) resistance of the wall F R can also be determined for the L specimens. The results are provided in Table 13 and Fig. 19. The model/ experimental ratio was in the range 0.78-1.03 for the L specimens. The low ratios can partly be explained by the dowel effect of the anchor in the compression zone not being taken into account.

Conclusions
An experimental study was presented that consisted of nine masonry walls retrofitted with a single-sided Fabric-Reinforced Cementitious Matrix (FRCM) layer, DM CFRP strips and flexible anchor connection. Three different wall geometries, with three different axial loads per geometry were tested to investigate the cyclic in-plane behavior of the reinforced masonry walls. Additional pull-out experiments considering two stress distribution scenarios (tensile force in anchor transferred to (A) masonry or (B) CFRP strip) were performed on prisms with steel anchors to determine the anchoring strength in the flexible adhesive. The following can be concluded from the study: • None of the specimens showed any shear damage at both the reinforced and as-built surfaces during the cyclic in-plane experiments. Cracking and eventually sliding always occurred at the interface between the bottommost bed-joint and the concrete foundation beam. • The S specimen loaded by a high axial load and all M specimens had rocking and toe-crushing as pre-dominant failure mechanism. The S specimens with low and moderate axial loads only showed rocking behaviour. The large specimens showed a combined flexure and sliding failure mode. • For the M specimens the wall's structural ductility factors were in the range of 3.7-14.7. For specimen L1, the mean ductility factor was found to be 4.2. For the remaining specimens a lower bound (in the range 3.5-25.0) was determined as the limited stroke of the horizontally oriented hydraulic actuator was insufficient to reach a 20% force drop in the post-peak phase. The exact value could not be determined do to the limitations of the in-plane shear test setup. • The mean anchor strength as determined with the direct pull-out experiments for scenario A, where the tensile forces in the anchor were transferred to the masonry, was found to be 49 kN. For scenario B, where the tensile forces in the anchor were transferred to CFRP strip (due to the chosen boundary conditions), the mean anchor strength was considerably lower at 18 kN. This was likely due to the relatively higher shear deformations of the adhesive between the CFRP strip and the anchor (scenario B) when compared to the shear deformations of the adhesive between the CFRP strip and the masonry (scenario A). • For the mean initial stiffness k 50% of the anchorage (from origin until 50% of the strength) no significant differences were found between the results obtained by direct pull-out testing with scenario A (12.3 kN/mm) and scenario B (12.8 kN/mm).  • A mechanical model based on EC6 was proposed to model the capacity of the reinforced walls during inplane loading, covering both moment failure and shear sliding failure. The mechanical model provided a good approximation of the experimentally obtained ultimate loads, with model/experimental ratios in the range of 0.68-1.08 and 0.78-1.03 for the specimens predominantly failing due to rocking (S and M specimens) and sliding (L specimens) respectively. The sliding resistance following the model was conservative because the dowel effect of the anchor in the compression zone was not taken into account. • Due to the application of the anchors and the single-sided FRCM overlay, the analytical model provides moment resistance amplification factors (moment resistance reinforced masonry/moment resistance URM) in the ranges 1.5-2.1, 1.