Modiﬁed hinged beam test on steel fabric reinforced cementitious matrix (SFRCM)

An experimental campaign based on modiﬁed hinged beam test (MhBT) set-up has been reported in the present study. The samples consist of two concrete blocks coupled by a proper hinge device and laminated with steel wire fabrics embedded in a cementitious mortar layer. Two kinds of fabrics, made of gal-vanized steel strands with diﬀerent mesh spacing, have been used to reinforce the concrete joists. With the aid of a DIC monitoring system, slippage proﬁle at the interface between the concrete support and the mortar laminate along the contact region has been assessed, together with the fracture opening. Force vs slippage at the interface has been retrieved for the sampled tested according to the MhBT set-up. With the aim to obtain predictive ultimate load design formulas, a novel classiﬁcation of laminate here proposed will be argued and related to a MhBT design formula. The inﬂuence of peel and shear stresses interaction on the ultimate strength of the system has been discussed in detail.


Introduction
The use of both long and short fibres has known a wide development in 2 various engineering fields.This is mainly due to the high chemical and mechanical fibres properties with respect to the same bulk materials form [1,2].In the 4 framework of aerospace and mechanical engineering a significant enhancement of the base materials can be obtained by exploiting nano-tube technology [3,4,5].
Furthermore in the field of Civil Engineering, a basic and simple technology together with a low manpower requirement have contributed to increase the 8 demand of fibre reinforced (FR) systems.In particular, fibre-based composites are nowadays the most common material used as reinforcement and rehabilitation engagement on existing structures in the framework of civil engineering [6,7,8,9].Indeed, continuous FR composite materials are commonly used as fibre reinforced polymers (FRP) and fabric reinforced cementitious matrix (FRCM).
In the case of reinforcements made of wire fabric embedded in a cementitious matrix, the composite system is generally named "ferrocement".For the FRP systems, almost always, the epoxy resin promotes the loading transfer between the matrix and the fibres.Conversely, in FRCM, the cementitious matrix allows the stress transfer to the embedded fibres.Generally, FRP systems achieve higher mechanical performance with respect to FRCM, but this leads to relevant drawbacks, typical of polymeric materials.
Fire resistance loss as well as lack of permeability, introduced by polymer matrix, make FRP unreliable materials for some European cultural heritage authority.Moreover, previous studies about the viscous effects in FRP [10] shown that epoxy-concrete interface under constant load can exhibit displacement that moving twice within few months.This relevant time-dependent effect, scoped on structures subjected to cyclic or high temperature load [11,12,13,14,15] can leads to a improper or useless repairing action.However FRCM, differently from the FRP, are significantly affected by ageing strength loss under aggressive environments which may reduce significantly the design limits [16].

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
Towards the end of the 20 th century, different experimental tests have been employed by the scientific community to investigate the bond-slip behaviour of composite materials.In particular, the available shear test schedules are subdivided in direct and indirect tests.Direct tests include the single lap setup [17,18,19,20] (Figures 1(a) and 1(b)) and the double lap test layout [21] (Figures 1(c) and 1(d)).Both single and double lap tests can be performed in near end (NE) or far end (FE) version.On the other hand, the indirect methods are the MhBT [7,22], (Figure 1(e)) and the most common four points bending test [23,24], Figure 1(f).The indirect tests are characterized by a more complex stress state with respect to the direct tests, which are more suitable for the calibration of the design parameters.On the other hand, indirect methods reproduce a stress state closer to that encountered in structural elements during their life.
In order to connect the results provided by direct and indirect tests, a numerical FE simulation based on non-linear fracture mechanics theory was performed by [25], investigating the differences induced on the stress state by single, double lap and beam tests.It's shown that, although of single and double lap tests provide similar results, the MhBT provides higher ultimate bond strength values.
The load transfer mechanism between the matrix and fabric plays a crucial role in the design of composite structures.The investigation of the transfer mechanism has led the scientific community to carry out extensive experimental programs to better understand the interfacial stress distribution and its dependence on the geometric and material parameters.
The experimental programs performed in literature pointed out that the crack initiation at the nearest loading point shifts the active bonded zone toward a new region farther from the loading point.The bond activation zone plays a crucial rule, for which it has been introduced the concept of a characteristic length, named effective or transfer length L e , beyond which the load transfer capability become negligible.A definition of the effective bond length as the region in which the strain assumes a linear distribution was proposed in [22].Differently, in [26] the effective bond length was denoted as the length needed to attain the 97% of the ultimate load depending on the assumed bond-slip relationship provided by an assumed theoretical interface constitutive models.
As reported in [27], the monitoring of the shear tests by using conventional strain gauges does not provide sufficient information to grasp local phenomena and high strain levels which typically occur in the neighbourhood of the laminate edges.Conversely, the use of full-field monitoring methods are suitable to properly describe local effects.In particular, the constant growth of non-

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
destructive technologies based on optical physics for surface field monitoring allows a higher resolution description of the delamination process.A far-field airbone Radar (FAR) technique has been used in [28] to analyse the influence of humidity and temperature on debonding phenomenon in FRP specimens subjected to four points bending test.Digital image correlation (DIC) instrumentation allows monitoring the fracture opening in a NE single lap test to capture the evolution of the interfacial stress distribution as the external load increases [29].Conversely, the use of full-field monitoring methods are suitable to properly describe local effects.In particular, the constant growth of nondestructive technologies based on optical physics for surface field monitoring allows an higher resolution description of the delamination process.
The present paper deals with the bond behaviour of FRCM laminated on a concrete support through a MhBT.The test program aims to the discussion of the reliability, in terms of unwanted and encountered failure mode, of the MhBT set-up.In addition, the possibility of this set-up to be performed on specimens with reduced span length, with respect to the original structural element, should be make this test particularly suitable for the bond test on existing structural element.The test is performed by recording the load vs deflection, under the upper knives, up to the sample failure.In order to investigate the monitored fracture opening, for which an analytical treatment can be found in [12,30] and [31,32], for infinitesimal and finite elasticity theory, respectively, DIC monitoring has been used to investigate the slip at the interface as well as fracture growth within the contact region.The paper is organized as follows: bonded to an elastic half-plane [33,34].

Research significance
The experimental program carried out in the present study has been organized to investigate the following issues: • analyse the load transfer capability of two different types of unidirectional steel fabrics (Figure 2(a)) laminated on a groove grid concrete support (Figure 2(b)); • evaluate the capability of the MhBT to run as reference test for characterization of reinforcing and rehabilitation engagements on existing structures owing to its wide adaptability both in terms of set-up arrangement and in particular cases the sample number could be increased.Indeed starting from two existing beam elements of 1.5m length, the MhBT allows to carry out three tests instead of two, by the beams cutting into three portions of 50cm length, in order to obtain three specimens realized from the union of two 50cm beams block (as described in Section 2.2 and performed in [7]).In addition, considering as an acceptable specimen block length 50cm, the hinge presence, which length is about 8cm, allows to earn an important length compared to the entire specimen.In the field of the existing building structural element the sample number increase could represent a relevant aspect.Particular attention will be paid to analyse how vacuum thrust, an unwanted failure not encountered in the original state of the manufacture, can influence the beam-test results; • identify the serviceability load and the optimal lamination length over which the load can be bear without exhibit crack patterns [30,35];

Experimental schedule and specimens design
A total of 24 specimens have been tested through MhBT schedule up to failure.Each specimen consists of two prismatic concrete blocks coupled by a mechanical hinge device placed at the upper part of the cross sections as shown in Figure 3(a).This schedule inspired to [36], born for longitudinal steel reinforcement bond test, is particular useful since both the compression and tension resultant magnitude and position are known during the test.Two test variables have been considered.The first one is the lamination length (L A ), which denotes the length of the composite contact region for each block.The following effective bond length formulas, given by [23], In order to promote adhesion between the laminate and the substrate, after 14 days curing, a groove grating with a square mesh of 1.5 × 1.5 cm and 4 mm depth, see Figure 2(b), has been realized on the lower surface of each block.
The specimens have been composed by coupling two blocks with a hinge device, placed at the upper side of the cross section and fixed by means of proper mechanical inserts, as shown in Figures 3.
Once the block pairs have been realized, the specimen intrados has been wetted and a first thin cementitious mortar layer 4 mm height (t) was laid on the block surface for a chosen lamination length.The steel fabric has been applied on the mortar bed (still wetted) and the laminate has been completed with a second mortar layer similar to the previous one.The attitude of the composite layer to prevent delamination process is provided by the fabric wire mesh, which confers higher internal grip to the laminate.After 28 days, once the laminate has been cured, the specimens have been tested through a bending test, according to Figure 3

Materials characterization
In addition to the block pairs, six cubic samples of 15 cm size have been cast in order to measure the concrete compressive strength (R cm ) from which the concrete tensile strength (f ctm ) has been inferred (Table 2) according to [37].In addition six 4 × 4 × 16 cm prismatic samples of laminate cementitious Type  mortar have been tested through three points bending test and compression test in order to assess the Young modulus (E), mean compressive strength and tensile strength (Table 2) according to [38].The rough wire fabric, according to [39], has been characterized under tensile test, f u , ε u and E, i.e. the ultimate tensile stress, the ultimate strain and the Young modulus, respectively.The mean values provided by the experimental test have been reported in Table 1.

MhBT load and monitoring procedures
The laminated specimen pairs have been tested in a four-points bending test using a MetroCom 7170S02 machine equipped with a 200 kN load cell with 0.6 kN resolution.The test has been performed according to 1 mm/min displacement rate.The machine has been equipped with a LVDT (linear variable displacement transducer) to monitor the deflection at the loading point.For some specimens, the displacement field in the neighbourhood of the sample midspan has been acquired through a DIC instrumentation (Figure 3(b)) in order to grasp slip at the support-laminate or laminate layer interface.The specimens that have not been acquired with DIC have a slash symbol in Table 3 under DIC amounts.DIC has been calibrated in order to acquire a 30×30 cm square window, centred at the middle span of the specimen.The acquisition system configuration ensures 10 µm resolution.
In order to measure the substrate-laminate or laminate layers interface slip, DIC displacement field has been referred to horizontal reference lines L 1 and L 2 oriented with the x axis, as sketched in Figure 4, placed in correspondence of the first and second thin mortar laminate layer, respectively.During the loading phase, the gap between the horizontal displacements of the reference lines provides the slip distribution along the sample.Note that a third reference line placed on the support over and parallel to L 1 should be required to evaluate laminate-substrate slip.Moreover, as discussed in Section 3, the groove grid realized on the support intrados before lamination provides a strong anchor system, hence no slip occurs at the support-laminate interface.

Force displacement results and failure modes
The MhBT results have been reported in Figure 5 for different types of wire fabrics.The detailed results for each type of wire fabric have been reported in Tables 3 and 4 in terms of the following quantities: maximum vertical force , where P M AX is the maximum tensile force on the fabric.It should be noted that the symbol F will be used in the following to denote the vertical resultant force applied by the upper knives, differently from the symbol P that denotes the force acting on the laminate evaluated as, where L k and d, represent distance between the upper and lower knives and the internal lever arm (Figure 3(a)).Moreover the overline symbol will be used to denote mean amounts of the entire specimens group.In addition, the DIC post processed data have been reported in Tables 3 and 4 in terms of maximum slip between the laminate layers (δ u ), external work (L e,τ ) and the relative error P M AX (defined from eq 3) for the estimation of the maximum tensile fibre force with respect to a theoretical predictive design formula eq 2. As reported in Figure 5 for specimens C, as the lamination length increases, the maximum load increases tending to a plateau.For the specimens C, the lamination length affects both the maximum strength and displacement.A different result has been provided by specimen A, where over the 20 cm lamination length, the maximum displacement are nearly the same while the maximum force decreases.Because of its larger mesh size, the wire fabric C provides high mechanical interaction between the laminate layers.As a consequence of the increase in shear transmission capability, both the maximum force and displacement increase with the lamination length, Figure 6.In the region close to the laminate edges the stress level grows with the external load.The stress redistribution, related to the longitudinal displacement field contour map in Figure 7, paints the activation of a more extended concrete reacting zone for specimen C with respect to the specimens A. A longer lamination length, for specimens C, implies a longer reacting region along the interface and, in turn, more compres-  sive trusses inside the concrete samples, according to the Morsch scheme.
A summary of mean maximum load (F M AX ), coefficient of variation (CV) and the mean fibre employment index, defined as the ratio between the mean fibre maximum stress over tensile fibre strength (σ f,M AX /f u ), have been reported in Table 5 for each specimen group.As highlighted by the employment index, fabric C achieved a stress level nearly close to its ultimate stress, conversely to

Slip profiles, bond slip-relation and interface shear energy
Making reference to the system shown in Figure 4, the displacement component x along the L 1 and L 2 reference lines, respectively u L1 and u L2 , have been acquired by DIC optical system.As no cracks occur at the interface between support and laminate, only adhesive slip has been observed.Accordingly, the slip between the laminate layers has been calculated as follows The slip distribution has been assessed based on the difference of spline fitting on the displacement field provided by DIC along the aforementioned reference lines.The characteristic slip distributions have been reported in Figure 9 at different load levels for specimens A 13 , A 22 and A 31 , representative of their own group behaviour.In addition, the slip profile of specimen A 23 has been reported to better understand local phenomena observed and discussed in Section 3.3, as well as for two C specimens, to proof that no slip occurred for fabric C, for fabric A the slip increases with the load, reaching its maximum value in the neighbourhood of the block corners from which it propagates.For each specimen A, the maximum slip has been reported in Table 3.
In this study, slip has been computed as the relative displacement between ing second order differential equation is found from the laminate and support equilibrium and compatibility condition [26,41], where Ω represents the stiffness parameter of the system (Ω = 1 [26] and [41] where E f , A f and t f represent the Young modulus, cross section area and thickness of the fabric, respectively).
The determination of stress, strain and displacement fields are based on the statement of a global τ (δ) constitutive law.Different laws have been widely investigate by [26] for FRP composite, starting from the assumption of linear, multi-linear or multi-linear combined with exponential decay τ − δ relations.
Differently from the direct tests, in the MhBT set-up the peeling stress component becomes relevant and the approach proposed by [26] should be badly grounded.So the external work has been calculated as a measure of the ductility of the system, L e,τ = δu 0 P dδ.
The experimental force-slip curves provided by DIC post-processing have been obtained fixing the x coordinate where the slip profile reaches its maximum (Figure 9) with the associate monitored load absorbed by the laminate.The force-slip curves have been reported in Figure 10.The obtained results show that

Crack opening
For the case of specimen A, in order to establish a critical load threshold over which crack occurs at the interface between the laminate layers, the crack opening plots have been reported in Figure 11.These plots have been referred  to the spatial coordinate system of Figure 4 and the ratio between the actual load and the ultimate load.The fracture opening has been obtained as the difference between the reference lines vertical displacements amongst which the fracture propagates,

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
With the exception of specimen A 23 , an almost homogeneous and congruent behaviour is retrieved for specimens belonging at the same group.As the lamination length increases, the load threshold characterizing the crack formation increases.For specimens A 13 and A 22 the effect provided by the lamination length is more pronounced on the magnitude of the crack opening than the load level corresponding to crack formation.The increase of dissipation capability of the system is enforced by the external work grow, Table 3. Whereas for A 13 shows that, for this specimen, slippage has occurred at higher magnitude and earlier load level compared to the sample A 22 .Slippage recovered for specimen A 23 is related to a counter-clockwise rotation induced by this phenomenon localized near the corners of the block.However, based on data listed in Table 3 and shown in Figure 10 it is remarked that, even if such a local phenomena occurs, external work does not significantly change.

Design consideration
As shown in the previous Sections, a high strength and rigid interface, such as the one provided by the fabric C, leads to a shearing failure in the concrete support.The design of FRCM system bonded to a concrete support mainly depends on the attitude of the laminate to transfer shear stress within two interfaces: laminate-support and between laminate layers.
In the spirit of the Eurocodes classification for steel joints, an analogous distinction should be proposed for the laminate systems, divided into rigid systems, force, based on the shear energy computing, where E f , A f and G f denote the Young modulus, section area and interface fracture energy of the wire fabric, respectively, whereas p f = b.

Peeling influence
The laminates ultimate forces P u have been reported in Figure 6 with respect to the lamination length.For the case of specimen A, the behaviour is really close to some behaviours reported in [19,42,43,44] for both FRPs and FRCMs bonded to concrete support, but not argued.Our results, as well as some other Literature results seem to suggest that over a certain lamination length the ultimate force decreases1 .This curious behaviour might be due to the complex contact interface mechanical interaction between shear (τ ) and peeling stresses (q) investigated in the following.
The mechanical interaction between a thin layer bonded to a substrate allows understanding why the ultimate force of a composite bonded to a concrete support tends to decrease over a certain lamination length.The analytical studies performed in [33,34] describe the mechanical interaction between a cover and an elastic substrate taking into account both the peeling and shear stresses q − τ .
Let us consider the beam-substrate interaction representative of a thin laminate bonded to a concrete support.The key point is to analyse how the stiffness of the covering (laminate) with respect to the support stiffness can influence the interface mechanical interaction.In particular, the superposition of two loading cases discussed in [34], namely a symmetric and skew-symmetric pair of axial forces acting at the beam edges, is significant to clarify this aspect.
The problem is governed by a dimensionless parameters E s χ l /G l and the laminate slenderness ratio t/L, as observed by [34] yet, where χ and G are the shear factor and shear elastic modulus.The subscript l and s will be used to denote laminate and substrate quantities, respectively.The case of a single force acting at the right beam edge is reported in Figure 12(a).Hence, for this loading condition, the ratio between the interfacial peel and shear stresses has been reported in Figure 12(b) vs the dimensionless horizontal coordinate ξ = x/L centred at the middle span of the laminate.In Figure 12(b), the darker grey areas represents the region in which the strength of the interface is promoted.Indeed, a positive peeling-tau ratio means that peeling component  As the laminate slenderness ratio increases, the inversion point (IP), i.e. the point closer to the loading point where the mechanical interaction ratio (q/τ ) changes in sign, moves to the left and, in turn, both the regions characterized by the maximum compressive peeling stresses increase, Figure 12(c).This means that, for high laminate slenderness ratios, the composite system increases significantly its debonding strength.Figure 12(c) highlights that despite of the commonly assumption of negligible peeling stress, the peeling stress component is able to achieve values higher than the 10% of the shear component.The ultimate load decreases as a critical lamination length is exceeded, thus suggesting that an optimal compromise between the materials strength and the q/τ interaction exists.As a consequence, once the geometry and the materials parameters (in particular, the laminate thickness, interface peeling and shear strength) are known as well the pure shear and peeling strength, the optimal bond length can be assessed.The analytical model allows evaluating how the laminate affects the failure mode of the system.By assuming three different models for the laminate, i.e.Timoshenko beam, Euler-Bernoulli beam and truss element, the stress intensity factors (SIFs) at the loading edge have been reported in Figure 12(d) varying t/L.As expected, the problem is governed by mode II, even if the mode I is not negligible with respect mode II.

Conclusions
An experimental investigation about beam-test on two different types of wire fabric with DIC monitoring has been performed.Through the post-processing of DIC data the specimen behaviour in terms of force-displacement curves, slippage distribution within the contact region and related energy as well as fracture opening have been discussed.The following conclusion can be drawn: • the load transfer performance of wire fabrics A and C differ in terms of fail- • The results provided by fracture opening plot, Figure 11, suggest to use a safety factor aimed at preventing the crack formation ranging between 0.5-0.7 (the serviceability load level).In addition for specimens C, the modified Italian design formula [37], based on the common Morsch truss strength mechanism provides a good design formula for beam-test results; • the analytic model proposed by [34] highlights that fixed the elastic material constants, the laminates with low values of slenderness promote the debonding strength.This fact, as discussed in Section 4, confirms the observed results, from which it can be state that over a certain lamination length the ultimate force decreases; Single lap in far end (FE) version.
Single lap in near end (NE) version.

F 2 F 2 F
(c) Double lap in far end (FE) version.

F 2 F 2 F
h (d) Double lap in near end (NE) version.(e) Modified hinged beam test (MhBT).

F 2 (
f) Four points bending test.

Sec 2
describes the specimen and test schedule design based on the materials characterization, as well as the test variables.The results are discussed in Sec.3 in terms of force-displacement response and DIC post-processed slip fields.The force-displacement result, supported by the DIC monitoring, allow to understand the effect of lamination length in terms of global results, forcedisplacement ultimate values and failure mode, as well as the local results, slip profiles and fracture opening.A design formula has been used to predict the ultimate load as a function of the failure mode strictly connected to the fab-100 M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT ric used.The influence of the peeling stress, generally neglected in theoretical model, will be treated in Sec 4, by considering the analytical model of a beam

•
discuss the external work L e,τ , related to the failure of the interface and defined as the area under the force-slip plot (P − δ).

Figure 2 :
Figure 2: Steel fabric and surface treatment.
has been initially adopted to set the lamination length limits, where E f , t f and f ctm are the fibre Young modulus, ultimate tensile mean strength (Table1) and the concrete ultimate tensile mean strength (Table2), respectively.It has been decided to investigate the behaviour of three different specimen groups for each type of wire fabric by choosing three different lamination lengths, equalM A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPTto 10, 20 and 30 cm.The second variable is represented by two unidirectional wire fabrics characterized by different mesh spacings.These fabrics are made of galvanized steel strands, named type A and C fabric, Figure2(a), which differ in the wire mesh spacing over the fabric section (A f ) fixed the support width (b), as reported in Table(1).Each specimen has been labelled with a code representative of fabric type, lamination length and identification number in the specimen group (e.gA 24 , represents the fourth specimen with 20 cm lamination length made with wire fabric A).The test set-up geometries aim to simulate elements representative of structural beams or slab joists sawed from an existing structure, as carried out in[7].All blocks have a rectangular cross section 15 cm width, 28 cm heigh (h) and 50 cm long (L).During the concrete caste, each block has been reinforced with longitudinal steel bars (f u = 450 MPa) of 6 mm diameter and 2 cm cover, placed at both upper and lower sides of the cross sections.Transverse steel shear reinforcements with double bracket, 15 cm longitudinally spaced, have been inserted in each concrete block, Figure3(c). (a).
M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT (a) MhBT schedule geometries (b) DIC acquisition set-up (c) Cross section for longitudinal and transverse steel bars reinforcement.

Figure 4 :
Figure 4: DIC reference system and reference lines.

Figure 6 :
Figure 6: Effect of lamination length, maximum load with anchored length.
the fabric A, in which the achieved stress is significantly lower.Three different failure modes have been observed.An adhesive failure (AF) ([40]) has been found for the wire fabric A, Figures8(a) and 8(b), owing to a progressive fracture propagation along the lower plane of the fabric.This type of failure is due to low mechanical grip induced by the small mesh size.Conversely, for specimens C failure occurred because of high stress level transferred from the laminate into the concrete support.In particular, in the case of C 1 specimens, failure was due to vacuum thrust (VT) generate by shearing stress in the neighbourhood of the block edge, with a consequent detachment of the concrete, as shown in Figure 8(d).Indeed, the concrete corner was subjected to stress concentration, and the increase of lamination length, for C 2 and C 3 group, has led to concrete shearing failure driven by crack formation and propagation starting from the laminate ends, Figures 8(e) and 8(f).For C 2 specimens, crack propagation started from the laminate up to the intersection of the lower longitudinal steel reinforcement.Instead, for C 3 group, failure has been caused by crack propagation starting from the laminate which joins the compressive region.The higher load transfer capability of fabric C is evidenced by the load-displacement curve.Each specimen C has exhibited a sudden loss and recovery in carry capacity before failure, as displayed in Figure 5.This response is related to progressive crack formations at the midspan of the laminate layer, highlighted as the lami-M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT (a) AF in specimen A 11 (b) AF in specimen A 21 (c) AF+VT in specimen A 31 (d) VT in specimen C 12 (e) CS in specimen C 21 (f) CS in specimen C 31

Figures 9 (
Figures 9(e) and 9(f).For the sake of clarity, two vertical lines have been introduced in Figures 9 corresponding to the hinge ends.A stronger skewsymmetric slip profile shown by Figure 9(a) has been found for specimen A 13 with respect to specimen A 22 and A 31 which shown skew trend in the initial load phases only, 9(b) and 9(d).For specimen A 23 a different failure mechanism has been observed.The specimens C have not exhibited slip.As shown by Figures 9(e) and 9(f), the slip was caused by fracture phenomena localized in the neighbourhood of the blocks edges or near the hinge only.Conversely,

L 1 and L 2
at the laminate lateral edge.Typically for the case of single lap test, the slip has been calculated by the use of LVDT placed at the upper surface of the laminate layer, where the acquired displacement became slip through the assumption of a fixed support.The lateral monitoring grasps exactly the M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT

and A 22
the crack propagation started at 0.5-0.6P u , for sample A 3 the critical load shifted over 0.7 P u .Moreover, for specimens A 3 the advantage provided by increasing the lamination length in terms of crack formation is paid in terms of dissipated energy which comes back closer to the values exhibited by sample A 1 .Differently from the previous cases, specimen A 23 has shown a congruent global behaviour with respect to its group, but showing an anomalous local phenomenon.Indeed, sample A 23 has exhibited a delamination process coupled with a leveraging effect induced by the right block corner.Pushing of the right block corner on the laminate induces an early crack formation.Figure 9(c) a) Mechanical interaction reference system, with Es = E l , νs = ν l .
mode.Adhesive fracture has been observed for specimens with fabric type A, which exhibit a pure friction strength mechanism based on slippage between the laminate layers, thus giving an almost linear ascending or plastic force-slip curve.Conversely, specimens C could take advantage of an extra mechanical strength provided by a high mesh spacing affecting the employment index, as reported in Table5;• the vacuum thrust failure mode plays a crucial role only for sample C 1 .This kind of failure indicates a good laminate-substrate shear transfer attitude, which allow to achieve high performance by the increase of lamination length.We suppose that, when vacuum thrust failure occurs, an increase of the lamination length produces failures into the support, as exhibited by samples C 2 and C 3 .Hence the MhBT set-up might be more suitable to investigate the mechanical behaviour of a composite system in which the lamination length exceeds the beam height minus the height of the hinge device to avoid vacuum trust failure.In order to better understand the results provided by MhBT compared with the other tests schedules shown in Figure1, a further experimental program based on double lap and other beam-test, reducing the displacement rate, are in progress;

•
the proposed FR classification, i.e. rigid systems, partial dissipative systems and dissipative systems allows identifying the sacrificial element in order to obtain a given behaviour of the system.In addition, this classification could guide on design formula as a function of the governing failure mode.

Table 1 :
Galvanized steel fabric geometries and single strands mechanical properties.

Table 2 :
Mortar laminate and concrete support properties.

Table 3 :
Results for wire fabric type A

Table 4 :
Results for wire fabric C (a) A 22 (b) C 23