Superelastic shape memory alloy cables for reinforced concrete applications

Construction and Building Materials 148 (2017) 307–320

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Superelastic shape memory alloy cables for reinforced concrete applications B. Mas a, D. Biggs b, I. Vieito c, A. Cladera a,⇑, J. Shaw b, F. Martínez-Abella c a

University of Balearic Islands, Dept. of Physics, Ctra. Valldemossa km 7.5, 07122 Palma, Spain The University of Michigan, Dept. of Aerospace Engineering, 1320 Beal Ave, Ann Arbor, MI 48109, USA c University of A Coruña, Dept. of Construction Technology, Campus Elviña s/n, 15071 A Coruña, Spain b

h i g h l i g h t s
Relatively large-diameter Ni-Ti cables characterized for their use in RC members.
Uniaxial tensile tests at different temperatures and loading rates performed.
Bonding tests between the cable and conventional concrete carried out.
The cables have been used to reinforce two small scale concrete beams.
Modulus of elasticity should be increased by adjusting the alloy or heat treatment process.

a r t i c l e

i n f o

Article history: Received 24 December 2016 Received in revised form 18 April 2017 Accepted 5 May 2017

Keywords: Reinforced concrete Shape memory alloy Ni-Ti Thermo-mechanical properties Bond Beam tests

a b s t r a c t The research on shape memory alloys (SMAs) has attracted a lot of attention in recent years for different structural engineering applications. In this paper the performance of relatively large-diameter Ni-Ti SMA cables is depicted for their use in reinforced concrete applications. The cable was characterized through a complete experimental program, including electrical resistance tests to determine the phase transformation temperatures, monotonic uniaxial tensile tests at different temperatures and loading rates, and cyclic tests to quantify the amount of strain ratcheting. Moreover, bonding tests between the cable and conventional concrete were performed. Lastly, the cable was placed inside real beam specimens to work as longitudinal reinforcement. The low modulus of elasticity of the SMA cables used here was a limiting factor that needs improvement, but the good strain recovery of superelastic SMA cable presents novel and untapped opportunities for its use as reinforcement in concrete. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Shape memory alloys (SMAs) are singular materials that have the ability to achieve high deformations and to return to a predefined shape after either mechanical unloading or upon heating above an activation temperature. SMAs are considered smart materials because of this distinctive characteristic and are regularly used in different fields and industries such as aviation, surgical medical equipment, and implants [1]. In terms of structural engineering, there are three key properties of SMAs: superelasticity (or pseudo-elasticity), shape memory effect (SME) and damping capacity. Superelasticity is the phenomenon whereby SMAs may be able to undergo large non-linear deformations and, despite ⇑ Corresponding author at: Mateu Orfila Building, Ctra. Valldemossa km 7.5, 07122 Palma, Spain. E-mail address: [email protected] (A. Cladera). http://dx.doi.org/10.1016/j.conbuildmat.2017.05.041 0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.

these, return to their original shape upon unloading. Shape memory effect refers to the phenomenon whereby SMAs are capable of returning to a predefined shape upon heating. The damping capacity is the ability of these alloys to convert mechanical energy into thermal energy and, thereby, possibly reduce movements or vibrations of a structure. All these properties are the result of thermoelastic martensitic transformations. The interesting properties of SMAs have inspired the research of possible applications of these alloys in the field of civil engineering [2–9]. For some alloys, such as Ni-Ti, the manufacturing of SMA cables is a promising way to resolve longstanding impediments to economically realizing large scale SMA elements [10,11], although some researchers have already successfully used 12– 32 mm diameter SMA bars [12–15]. By leveraging the highly optimized manufacturing processes currently available for wire, the cable form results in a large-force SMA element with superior properties for substantially less cost than a monolithic bar of

308

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

comparable size [10]. Moreover, conventional steel cables are a very common construction product. The study of the mechanical properties of SMA cables is still in an incipient stage. Reedlunn et. al. [10,11] studied the superelastic behavior of two Ni-Ti cable constructions, a 7  7 right regular lay and a 1  27 alternating lay. The external diameter of the cables studied were 2.475 mm and 1.582 mm. Ozbulut et al. [16] explored the performance of a 8 mm diameter Ni-Ti cable for its potential use in civil engineering, focusing in cyclic uniaxial tests. Carboni et al. [17] studied the hysteresis of multi-configuration assemblies of Ni-Ti and steel strands. In this paper, an 8 mm Ni-Ti cable specifically manufactured for this study was characterized through a complete experimental program, including electrical resistance tests to determine the phase transformation temperatures, uniaxial tensile tests at different loading rates and temperatures, a cyclic test to quantify the shakedown behavior, and a test to ultimate failure. Moreover, bonding tests between the cable and conventional concrete were performed to measure the bonding properties. Lastly, the cable was placed inside real beam specimens and shown to work as longitudinal reinforcement. 2. Thermo-mechanical characterization of the Ni-Ti cables 2.1. Materials Relatively large superelastic Ni-Ti cables were provided by Fort Wayne Metals Research for thermo-mechanical characterization. The cable specimens were provided in 305 mm (12 inch) lengths, each consisting of 49 wires (each of 0.885 mm nominal diameter) in a conventional 7  7 cable construction with about an 8 mm outer diameter. This is a hierarchical structure consisting of seven wires (six wires wrapped helically around a central wire) within each strand, and seven strands (six strands wound around a central straight strand) in the entire cable. The Ni-Ti wires were superelastic at room temperature, indicating a slightly Ni rich chemical composition (50.95 at.% Ni, certified by the manufacturer). The precise heat treatment was not released by the manufacturer, but after being wound into the 7  7 structure, the wires were shape set at an aging temperature (typically near 500 °C) to maintain their helical forms and avoid unwinding at room temperature.

1.20

Electrical resistance tests were performed to determine the phase transformation temperatures of Ni-Ti material. The experimental technique used was four wire alternating current (a.c.) impedance measurements at a frequency of 686 Hz [18]. The real part of the impedance R was measured by means of a lock-in amplifier, which provided high resolution measurements. The temperature range analyzed was from 190 °C to 80 °C. More information about the test setup and the interpretation of the results may be found in [18]. Fig. 1 provides the results, showing an austenite finish temperature of Af = 43 °C and a martensite finish temperature of Mf = 148 °C. Note that the Af temperature is slightly high according to the Ni content of the alloy declared by the manufacturer. A third martensitic phase, called the R-phase (rhombohedral), was detected in the test during cooling between 43 °C and 0 °C, the presence of which is common in commercial Ni-Ti alloys, arising from residual stress fields and dislocations in cold worked/ heat treated alloys [19]. The cables used here were quite similar to one of the cables used in the study by Reedlunn et al. [10,11] provided by the same manufacturer, although those used smaller wires (0.27 mm wire diameter). Despite using, in this study, a different technique to establish the transformation temperatures, it is expected that the differential scanning calorimetry (DSC) thermogram of Fig. 3a in Reedlunn et al. [10] could be also representative, after about a 20 °C shift (increase), of the transformation temperatures of the cables used in this study. The DSC thermogram from Reedlun et al. [10] also shows the presence of the R-phase. After heat treatment by the manufacturer and allowed to cool, the material began to transform from austenite to R-phase at 43 °C, thereby leaving a mixture of austenite and R-phase at room temperature. The asreceived condition, therefore, was not fully austenite, which although it is often ignored in other applications (and models), it actually does affect the initial mechanical response of the cables shown later. 2.2. Thermo-mechanical experiments The experiments performed for the thermo-mechanical characterization of the SMA cables explored various aspects of their mechanical behavior, including their strain rate sensitivity, temperature dependence, core wire behavior, cyclic shakedown, and

Ms (R→M)= -68oC

1.15

Mf (A→R)= 0oC

R/R80oC

1.10

1.05

As = -25oC

Mf (R→M)= -148oC

Ms (A→R)= 43oC

1.00 Af = 43oC

0.95 -200

-150

-100

-50

0

50

100

Temperature, oC Fig. 1. Temperature dependence of resistance for a sample of the cable. Data are normalized to the value of resistance at 80 °C.

309

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320 Table 1 Summary of uniaxial experiments for thermo-mechanical characterization. Subset

Experiment ID

Strain Rate (s1)

Free Length L (mm)

Gage Length Le (mm)

Ambient Ta (°C)

A1 A2 A3 A4

2  105 1  104 1  103 1  102

247.2 244.5 243.0 244.0

99.68 99.74 95.83 96.01

21.7 21.8 21.4 21.6

A1 B2 B3

2  105 2  105 2  105

247.2 89.5 93.0

99.68 49.38 49.94

21.7 30.4 41.0

C1

2  105

98.8

56.27

21.6

D1

1  103

246.0

98.73

21.7

E1

1  103

247.2

99.54

21.7

Strain Rate Sensitivity

Ambient Temperature

Core Wire 50 Cycle Shakedown Failure

ultimate failure strength. A summary of the experimental parameters is provided in Table 1. The first group comprises four experiments (A1, A2, A3, A4), whose purpose was to quantify the sensitivity of SMA cables to loading rate. The second group includes three experiments at different ambient temperatures at or above room temperature (A1, B2, B3), which was used primarily to measure the thermo-mechanical coupling (Clausius-Clapeyron slope), but also helps one appreciate the three-dimensional temperature-stress-strain space that SMA cables occupy. The core wire experiment (C1) provides the underlying uniaxial tension behavior of monofilament wire (0.885 mm diameter) used in the 7  7 cable. The shakedown experiment (D1) shows the cyclic strain ratcheting behavior of the cable. Finally, the failure experiment (E1) quantifies the approximate load the cables can withstand before breakage. 2.2.1. Experimental set-up for uniaxial mechanical tests The primary experimental setup used to characterize the cables is shown in Fig. 2A. The experiments were performed in an Instron load frame, model 5585, with a 200 kN capacity. Attached to the load frame were a 200 kN Instron load cell and wedge grips. A laser extensometer (EIR, model LE-05) was used to measure the relative displacement of two laser tags attached by flexible epoxy to the

middle portion of the specimen. This configuration was used in order to measure the average strain in a gage section away from stress concentrations at the grips and to avoid measurement artifacts from a small, but noticeable, amount of grip slippage. An oscilloscope verified the output signal of the laser extensometer, to check for possible glare or alignment issues. Two thermocouples were used to record the ambient and specimen temperatures. The thermocouples were attached to Fluke thermocouple amplifiers that provided a 1 mV/°C output. Certain modifications to the setup were necessary for parts of the thermo-mechanical characterization. For the elevated strain rate sensitivity experiments (faster than the near isothermal rate of 2  105 s1), a FLIR IR camera recorded the specimen’s fullfield temperature history. To improve the accuracy of the IR measurements, specimens were first airbrush painted with a thin coating of matte white to raise the emissivity to about 0.94. The elevated temperature experiments were performed in an Instron thermal chamber as seen in Fig. 2B, without any IR measurement, relying solely on the thermocouple measurements. Gripping the specimens was a challenge. In our initial attempts, the outer fibers of the cable specimen would break during loading, typically at the onset of phase transformation and again at saturation when local stress concentrations were the largest. The issue

Fig. 2. Experimental setup. A) Rate sensitivity setup. B) Elevated temperature setup.

310

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

was identified to be caused by the sharp teeth in the v-groove of the grip platens. The solution was an extensive specimen preparation procedure that involved epoxying on tin-plated copper ferrules. To prepare a specimen, about 37 mm of the end of the cable’s outer strands were first unwound from around the core strand, and the core strand was coated in epoxy. The outer strands were then rewound around the core strand, and more epoxy was spread over the outside of the cable. A ferrule (32 mm length, 8.3 mm inner diameter, 0.2 mm thick) was then slid over the cable and epoxy, while ensuring enough epoxy filled the gaps between the ferrule and cable. The specimen with epoxied ends was then cured in an oven at 100 °C for 45 min. One end of a cable specimen with the epoxied ferrule is shown in Fig. 3A before an experiment, and the same end is shown in Fig. 3B after the experiment. One can see that during the experiment the grip plastically deformed the ferrule and left teeth marks, but this extra layer effectively distributed the wedge grip’s stress concentrations on the cable wires while maintaining a strong hold on the specimen. With this arrangement, specimens could be taken well beyond the loads needed for stress-induced transformations, and the ultimate breaking strength of the cable could be estimated. 2.2.2. Strain rate sensitivity The mechanical responses of superelastic SMAs are well known to exhibit a strong dependence on the loading rate, even at low strain rates that would normally be considered quasi-static for conventional materials. The cause of the SMA’s sensitivity is not due to inertia effects or viscoelasticity, but is due to temperature excursions caused by latent heat of transformation released or absorbed during stress-induced forward and reverse transformations, respectively. If the ambient medium is unable to extract or

supply heat to the specimen fast enough, the temperature of the specimen must change. Since the transformation stresses in SMAs are strongly dependent on temperature (as shown in the next section), even a small temperature change causes a significant change in stress within the specimen during a constant displacement-rate superelastic test. During the forward transformation (A ? M) while loading, latent heat is released, the specimen temperature rises above ambient temperature, thereby causing the stress to evolve upward. During the reverse transformation (M ? A) while unloading, latent heat is absorbed, the specimen temperature drops, and the stress evolves downward. More details can be found in [20]. While the strain rate sensitivity of Ni-Ti wire and strips has been well studied [21–23], the SMA cables here are significantly more massive and their sensitivity to strain rate was as yet unquantified. All experiments shown below were performed on a new cable specimen to avoid any accumulated damage that otherwise might have occurred. One end of the specimen was held fixed, while the other end of the specimen (vertically oriented) was displaced (d) upward by the testing machine’s crosshead. All experiments were performed in displacement control at constant rate during loading and unloading, during which the axial load P (by the load cell) and gage strain ee (by the laser extensometer) were recorded. The results are reported below in terms of the stress measure r = P/A0, where A0 is the cross-sectional area of 49 parallel wires (as defined in Reedlunn, et. al. [10]). It should be noted that due to the hierarchical and helical construction of the cable, the local stress within a wire involves combined extension-bending-twisting and varies with its location within the cable, so r it should be just considered the normalized axial load.

Fig. 3. Photographs of a 7  7 cable specimen end with epoxied tin-plated copper ferrules. A) Before experiment. B) After experiment.

(A)

(B)

Fig. 4. Rate sensitivity data of 7  7 cable. A) Mechanical responses. B) Temperature change histories.

311

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

Fig. 4 shows the responses of our SMA cables at four strain rates from 2  105 s1 to 102 s1 (based on the normalized grip displacement, ee = d/L) in room temperature, stagnant air. Fig. 4A provides the mechanical ‘‘stress-strain” responses, according to our stress measure r and laser extensometer strain ee, showing higher stresses during loading at progressively higher strain rates. The cause of the strain rate sensitivity is shown in Fig. 4B which plots the temperature changes in the specimen as measured by the specimen thermocouple. At the slowest rate the temperature change in the specimen is small, with a slight increase during loading (between t/tmax = 0 and 0.5) and a slight decrease during unloading (between t/tmax = 0.5 and 1). Nevertheless, this experimental response can be considered to be the nearly isothermal, quasistatic response. By contrast, the response at 104 s1 shows an elevated stress during loading and a reduced stress during unloading (Fig. 4A), and the specimen temperature changes are larger (about ±8 °C relative to room temperature in Fig. 4B). At 103 s1, the effect is larger yet, and the specimen temperature rises by about 20 °C during loading. Since there was no pause between loading and unloading, the temperature of the specimen remained above room temperature for much of the unloading segment, resulting in an elevated stress even during reverse transformation (although latent cooling contributed to drive the temperature downward). The response at the fastest rate 102 s1 was similar to that at 103 s1, and the thermocouple recorded a similar rise in temperature by about 20 °C during loading. The specimen temperature again remained above ambient during unloading and never reached it before the load reached zero. Full-field IR measurements were recorded (for all but the slowest experiment), and unlike typical single wire experiments, the temperature field showed no evidence of localized transformation but rather a uniform heating or cooling along the length of the cable. Since the thermocouple data adequately summarizes the temperature effects, the IR data are not shown. Overall, we conclude that the 103 s1 response is close to the adiabatic response of the cable, and note that this is a slower strain rate than is necessary to reach the adiabatic limit of single SMA wire (typically near 102 s1). Thus, the rate sensitivity is dependent on the size of the specimen, in accordance with heat transfer scaling. The progression of energy dissipated during a superelastic cycle (area within the stress-strain loop) is interestingly non-monotonic with increasing strain rate, starting at a nominal value in the lowest rate (isothermal) case, becoming larger at intermediate rates,

(A)

but then decreasing to below the starting value as the rate increases further and approaches the adiabatic response. However, one should not conclude this is always the case, since it depends on how the experiment is run. If the experiment had been paused between loading and unloading to allow the specimen to reach ambient temperature, the area within the energy dissipated would have increased monotonically with strain rate (see, for example [22]). Thus, the rate dependence in SMAs is largely governing by transient heat transfer between the SMA and the ambient medium. It should also be emphasized that the experiments shown here quantify the rate sensitivity in stagnant air, which is not a very conducive medium for heat transfer, relying largely on natural convection. The strain rate sensitivity within a solid body, like concrete, will probably not be as severe, since thermal conduction will likely improve heat transfer to/from the SMA cable. 2.2.3. Elevated temperature experiments Experiments were performed at elevated temperatures to explore the temperature sensitivity of our SMA cable and to measure the thermo-mechanical coupling in terms of the ClausiusClapeyron slope. These experiments were run at the slow (nearly isothermal) strain rate of 2  105 s1 at ambient temperatures near 20 °C, 30 °C, and 40 °C. The mechanical responses are shown in Fig. 5A with the corresponding specimen and ambient temperature histories shown in Fig. 5B. The initial linear elastic segments and A ? M stress plateaus were fitted with tangent lines, and the intersection was selected as the onset stress. Using these onset stresses, the Clausius-Clapeyron slope for this cable was found to be 4.6 MPa/°C (see Fig. 5C), which is lower than typical values for straight wire near 7 MPa/°C. Also, the effective elastic modulus for extension of the cable, based on the linear fit used above, was 23.9 GPa, 26.6 GPa, and 32.5 GPa for the respective 20 °C, 30 °C, and 40 °C experiments, which is consistent with a progressively decreasing amount of initial R-phase. 2.2.4. Core wire The only straight wire within the 7  7 cable is the central wire in the core strand, which we term the core wire, and it is the only wire that experiences direct uniaxial tension when the cable is stretched. Although the other 48 wires in the cable are loaded in more complex deformation modes, the core wire’s mechanical response provides a useful baseline for comparison to the full cable response. Fig. 6A provides the uniaxial response of a core wire that

(B)

(C)

600

547 MP a

550

496 MP a

500

460 MP a 41.0 °C 30.3 °C 21.2 °C

450

Specimen

400 10

20

30

40

50

Ambient

Fig. 5. Elevated temperature data of 7  7 cable. A) Mechanical responses and fits of A ? M onset stresses. B) Measured temperature histories. C) Fitted Clausius-Clapeyron slope.

312

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

(A)

(B) Core wire

7x7 cable

Fig. 6. Core wire data at room temperature. A) Mechanical response of core wire compared to the full cable. B) History of measured temperature change in core wire.

had been excised from a 7  7 cable. The experiment was performed at room temperature and at a slow strain rate (2  105 s1), and Fig. 6B confirmed that the specimen remained nearly isothermal (specimen changes within about ±0.4 °C). Compared to the full cable, the core wire’s response is stiffer, transformations occur at higher stress levels and transformation plateaus are more distinct and flat. The initial elastic modulus of the core wire was 33.6 GPa (compared to 23.9 GPa for the full cable), and the A ? M plateau stress was 520 MPa (compared to 460 MPa for the full cable). While the modulus of the core wire was larger than the cable’s effective modulus, it was unexpectedly lower than other superelastic Ni-Ti wires we have tested elsewhere. To verify the low elastic modulus, a new core wire was tested at 60 °C ambient temperature, and the elastic modulus was measured to be 66 GPa, which is indeed a typical value for Ni-Ti austenite. This was further evidence that the material was not fully austenite (a mixture of R-phase and austenite) at room temperature, where Rphase detwinning contributed additional compliance in the wire’s apparent elastic response. As discussed later, the low elastic modulus created difficulties for the use of SMA cable as reinforcement in concrete, but it should be possible to minimize the presence of the R-phase by optimizing the cable’s heat treatment in the future. The reason why the cable’s response has more gradual and positive sloped plateaus is due to the heterogeneous stress distributions between and within individual wires in the cable. As previously mentioned, only the core wire experiences pure tension. The other 48 wires have helical configurations that create a combination of tension, torsion, and bending loading modes within a given wire. This leads to the following complications: (1) the relative amounts of tension-torsion-bending varies from wire to wire, depending on their helix angles and mean helix diameter; (2) twisting and bending creates strain gradients across a wire’s cross-section; and (3) the local stress state within a wire’s crosssection is multi-axial, since it involves both normal strains (extension and bending) and shear strains (twisting). Consequently, the load sharing is unequal between wires, and local transformation stresses are reached in different wires at different times during loading/unloading of the cable. Some wires transform early, while others transform late, resulting in a gradual evolution of the apparent tangent modulus in the macroscopic response of the cable. More details of the load sharing between cable subcomponents can be found in Reedlunn, et al. 2013 [11]. 2.2.5. Cyclic behavior and failure The cyclic behavior and ultimate failure strength of a SMA cable are important functional and durability considerations, so the final two experiments in this thermo-mechanical characterization

address how the cable responds to repeated load-unload cycles and to extreme loading to its breaking strength. Fig. 7A shows the mechanical response of a cable specimen subjected to 50 loading cycles to 8% maximum axial strain at a strain rate of 103 s1 in room temperature air. This relatively fast strain rate was chosen in order to complete the experiment in a reasonable amount of time and to explore the cyclic behavior under severe conditions. We consider this experiment to be a rather brutal test, since the maximum axial stresses reached over 700 MPa at maximum strain. Nevertheless, the cable remained intact. Fig. 7A does show, however, a dramatic evolution in the response to progressively lower transformation stresses during cycling, accompanied by a reduction in stress hysteresis and a small amount (less than 1%) residual strain ratcheting. The most severe changes occurred in the first few cycles, and each successively cycle exhibited a decreasing incremental change from its previous cycle, asymptotically approaching a limit cycle. This behavior is known as ‘‘shakedown” or ‘‘functional fatigue”, typically observed in superelastic wires when subjected to large cyclic stresses. Fig. 7B plots the corresponding specimen temperature history, showing significant temperature excursions (as expected for this strain rate) that after an initial transient settled down to about (+13 °C, 8.5 °C) in late cycles. The evolutions of A ? M (during loading) and M ? A (during unloading) onset and saturation transformation stresses with cycling are plotted in Fig. 7C, showing exponential-like decays in all cases approaching stable asymptotic values. Lastly, a cable was loaded monotonically to failure at a strain rate of 103 s1 as shown in Fig. 8. The cable failed at a peak stress of 1150 MPa, which corresponds to an impressive 31.1 kN load. The gage strain at failure was about 20%, which demonstrated reasonably good ductility. The cable broke near the upper grip, indicating perhaps that we had not completely eliminated the stress concentration there. Thus, the measured failure stress should likely be considered a lower bound value. Overall, the thermo-mechanical characterization experiments performed here show that the cable construction offers an effective way to scale up the novel properties of superelastic SMA wire, and SMA cables are shown to be remarkably robust and durable tension elements. 3. Bond tests The bond between concrete and reinforcing steel can be defined as the physical and chemical interactions and other phenomena that make it possible for these two components to work together, resulting in a new material known as reinforced concrete. Bonding is therefore one of the fundamental factors controlling the behav-

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

(A)

313

(B) Cycle 1

Cycle 50

(C) σ* (MPa)

800

800

Cycle 1

Cycle 50

400 0

600

0

2

4

6

8

10 0

2

4

6

8

10

400 200 0 0

10

20

30

40 n

50

Fig. 7. Cyclic test of 7  7 cable. A) Mechanical response. B) Specimen temperature change history. C) Onset and saturation stresses, during loading and unloading, versus cycle number.

There are different experimental tests to characterize the bonding. The pull-out test, which has gone through several variations over time, has become the test supported by the RILEM (Réunion Internationale des Laboratoires et Experts des Matériaux, systèmes de construction et ouvrages). Its use became the norm at the beginning of the 1970s, and the test underwent geometric revisions in 1983 [24]. During a pull-out test, regardless of variations, a curve is obtained that continuously relates the tensile force and the slip measured on an embedded reinforcing bar. This curve can be used to calculate the average bond stress. Compared to conventional steel reinforcement, the Ni-Ti cable in this research exhibits several unique characteristics that may affect its concrete bonding properties:

Fig. 8. Normalized force-displacement graph for load to failure test.

ior of reinforced concrete as a structural material. If there were no bonding, reinforcing steel would slip without meeting any resistance along its length and without matching the deformation of the concrete. If the unbonded reinforcement were sufficiently anchored as a result, for example, of having a welded plate at its end that would prevent slippage, then the component could develop some resistance even while undergoing significant cracking. Otherwise, once the concrete started cracking, component failure would be immediate.

a) Reinforcement cross-section: Reinforcing steel is characterized by surface protrusions, commonly known as ribs. The purpose of these ribs is to create tension in the surrounding concrete, thereby developing compressive stresses that greatly increase the bonding capacity of the reinforcing steel. Without ribbing, a smooth bar relies only on chemical adhesion and friction and thus provides much less resistance to slip than a corrugated bar [25]. However, the production of the Ni-Ti cables being analyzed is unusual in that they are first individually drawn out and then grouped in seven helicoidally arranged sets of seven-strand wires totaling 49 wires; as a result, their cross-sectional shape is completely different (Fig. 9). Fig. 9 shows the two extreme theoretical situations in the arrangement of the wires in a cross-section of the tested cable. Fig. 9B represents the ideal theoretical cross-section; this is the arrangement expected from the drawing and formation of the

314

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

Fig. 9. Cross-section configuration of the Ni-Ti cable.

Fig. 10. A) Schematic of the tests. B) Actual set-up of the tests.

cable, according to the manufacturer’s instructions. The crosssection on Fig. 9A shows, in contrast, another possible configuration that diminishes the outer perimeter. These two configurations are very different in terms of the apparent diameter and perimeter of the respective cables. As a tension load is gradually applied, it is expected that the cross-section will adopt a shape between these two extremes. b) Stress-strain behavior: The apparent modulus of elasticity measured in the Ni-Ti cables is strongly dependent on the temperature and is approximately 23.9 GPa for the 20 °C test. This is an order of magnitude lower than that of steel, and it is very similar to the modulus of deformation of conventional concrete. The martensitic transformation from austenite to martensite in the cable occurs at a stress of approximately 460 MPa for the 20 °C test. This value is reasonably similar to the yield stress of conventional reinforcing steel. 3.1. Experimental set-up Fig. 10 shows a schematic of the test setup. In the pull-out test, the reaction pressure on the base of the cubic concrete specimen is

equilibrated by the bond stress developed in the contact zone between the embedded reinforcement and the concrete. The deformation of the materials and degradation of this contact zone causes displacement between the concrete and reinforcement (slip). This displacement is continuously measured at the upper free end of the reinforcement, and the force exerted on the lower end of the rebar is measured simultaneously. The standard RILEM specifications were followed [24]. This standard specifies that the width of the cubic specimen is equal to 10 times the bar diameter (10Ø) and that there are two differentiated regions of the embedded bar, one where bonding is allowed and other where bonding is prevented, both with lengths equal to 5 times the bar diameter (5Ø). The bond-free region can be achieved using a sleeve, grease or a similar component. In the tests conducted, the bond-free region was created with molybdenum disulfide grease that was wrapped with adhesive tape. In this specific case of Ni-Ti reinforcement, the outer diameter of the samples was slightly greater than 8 mm, which would lead one to conclude that an 80  80  80 mm specimen was required. However, on the one hand, detailed analysis of the cable crosssection revealed that its circumference was between 36.79 and

315

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

45.32 mm, which would be equivalent to a cylindrical conventional steel bar with a nominal diameter of 12 mm. On the other hand, Ni-Ti reinforcement components have a nominal crosssectional area of 31.28 mm2, approximately equivalent to that of a 6-mm-diameter conventional reinforcing bar. In the absence of definitive criteria and based on convenience and the availability of formwork, it was decided to use concrete specimens measuring 100  100  100 mm. The RILEM standard specifies a constant loading rate based on the formula v ¼ 0:5  D2 [N/s]. However, it is much more common to establish a test loading rate expressed in units of displacement because its control is safer and more accurate. The corresponding conversion, which is only exact in the elastic range of the rebar and in the absence of slip, requires that the lower free length of the bar and its mechanical characteristics be taken into account. Within the range of loading rates proposed, significant differences in behavior are not expected [26]. Defining the test loading rate poses a series of alternatives, as explained above, because the nominal diameter (physically measured with a micrometer between the two most distant points in the real section) provides little information about the area and circumference of the cable (a variable that will be required later to calculate the average bond stress). This does not occur in conventional reinforcing bars, which are easily approximated as a cylinder without significant error. For this reason, it was decided to use two loading rates. One rate corresponded to the nominal measured diameter of the reinforcing component (8.11 mm). The second rate

corresponded to the diameter of a cross-section with the circumference of the actual section, assuming arrangement 1 (see Fig. 11 and Table 2). Note that both loading rates selected (3.8105 and 7.5105 s1) are inside the range of the uniaxial tests carried out in Section 2 (see Table 1). It is widely accepted that the mechanical properties of the concrete, particularly its compressive strength, are highly relevant to the bonding properties being measured [25,27]. In this case, the selected concrete was of medium to high strength (40–50 MPa), of viable fabrication with generally available materials, and compatible with the selected high-performance reinforcement. To that end, based on several tests, the mix design shown in Table 3 was selected. The cubic samples were made using a water/cement ratio of 0.58 and reached a 28-day mean compressive strength of 47.0 MPa in the cubic samples. The bonding tests were also performed at the age of 28 days. The cubic samples used to determine the strength and the samples for the pull-out tests were cured in a chamber under controlled temperature and humidity conditions, in accordance with standard EN 12390-2:2009, i.e., at a temperature of 20 ± 2 °C and relative humidity above 95% [28]. 3.2. Results and discussion To obtain the average bond stress, the force recorded by the load cell was divided by the bond surface. As previously explained, it is very difficult to determine the surface area of the cable

Fig. 11. Schematic cross-section of a sample and its essential geometry. In both cases, the calculated area refers to the zones shown in blue, which represents the area occupied by the cable inside the concrete specimen. The calculated circumference is the real value in each arrangement. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2 Loading rates applied in the tests. Diameter (mm) 8.11 11.71 (36:79=p)

Criterion Nominal measured diameter Diameter of a cross-section with the same circumference

Water Cement Sand, 0–4 mm Gravel, 4–12 mm Coarse aggregate, 12–25 mm Nafta-based superplasticizer (MasterRheobuild 1000)

5

32.89 68.61

3.810 7.5105

Rate (mm/min) 1.14 2.28

Table 4 Bond stresses in the tested cubes.

Table 3 Concrete composition. Component

Rate (s1)

Rate (N/s)

Amount (kg) per 100 l of concrete 15.85 27.50 99.03 48.75 56.30 0.042

s0:1 (MPa)

smax (MPa) and slip (mm)

Loading rate: 3.8105 s1 B25-1 5.0 B99-2 8.0 B99-3 7.2

6.3 8.1 8.2

7.9 (0.49 mm) 8.2 (0.21 mm) 9.2 (0.48 mm)

Loading rate: 7.5105 s1 B22-2 7.9 B22-3 4.6 B25-3 3.1

8.3 6.5 5.9

8.4 (0.26 mm) 8.3 (0.54 mm) 8.4 (0.52 mm)

Sample

s0:01 (MPa)

316

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

because it does not resemble conventional cylindrical reinforcement. It was decided to calculate the mean surface area by multiplying the circumference of the sample by the bonded length. The bonded circumference was considered to be the average between both possible extreme arrangements of the cable (Fig. 11), and the bonded length was 50 mm. The mean bond surface was consequently considered equal to 2053 mm2.

3.2.1. Loading rate of 3.8105 s1 The bond stress-slip graphs obtained from the samples tested at a loading rate of 3.8105 s1 indicate that the maximum force was developed before reaching even 1 mm of slip. The three tested samples replicated this behavior, although they exhibited strong variations both in the stresses at a slip of 0.01 mm and in the slip corresponding to the maximum stress (see Table 4 and Fig. 12). The maximum stresses, however, are similar among the three samples.

3.2.2. Loading rate of 7.5105 s1 The tests performed at the higher loading rate reveal very similar behavior. The apparent homogeneity in the maximum stresses remains, as shown in Table 4 and Fig. 13, but there are important differences in the slip at which the maximum bond stresses develop. 3.2.3. Discussion The bond stress-slip relationships obtained in the bond tests show a behavior different from what would be expected for conventional reinforcing steel. The samples developed their maximum bond stress with hardly any slip, exhibiting a very brittle behavior. In addition, the curves are not smooth and continuous but instead exhibit numerous peaks and irregularities, in contrast to what occurs in a pull-out test of conventional reinforcement. These peaks were accompanied by crackling sounds from the specimen during the testing.

10

Average bond stress (MPa)

9

B99-3

8

B99-2 B25-1

7 6

5 4 3

2

Sample B25-1 Sample B99-2 Sample B99-3

1 0 -0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

Slip (mm) Fig. 12. Bond stress-slip graph of the samples tested at a loading rate of 3.8105 s1.

10 9

Average bond stress (MPa)

B22-2

8

B22-3

7

B25-3

6

5 4 3

2

Sample B22-2 Sample B22-3 Sample B25-3

1 0 -0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Slip (mm) Fig. 13. Bond stress-slip graphs of samples tested at a loading rate of 7.5105 s1.

1.1

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

When studying the bond behavior between reinforcement and concrete, it is usually assumed that the three fundamental mechanisms (chemical adhesion, friction between surfaces in contact and mechanical interaction produced by the bar shape) occur at increasing loads. It is plausible that the complex surface geometry of the cable was the cause of the erratic behavior. This shape is characterized by an abundance of cavities, holes and concave surfaces, which considerably increase the circumference in contact with the concrete and thereby increase the apparent bond surface area. However, their small sizes make it difficult for the holes to fill

317

completely, and if they were filled, the filling would consist of heterogeneous concrete very similar to a mortar or grout. The mechanical tension, therefore, would exhibit significant local maximums and variable strengths, depending a great deal on mechanical blocking of the slippage. Perhaps more importantly, the low modulus of elasticity of the Ni-Ti reinforcement (23.9 GPa at 20 °C), a value very close to that of concrete, may play a significant role. High longitudinal strains would be accompanied by a significant reduction in the cables cross-section, which would reduce the mechanical blocking effect, friction and adhesion. During the tests,

Fig. 14. A) Free end of the reinforcement before the test. B) Photograph taken immediately after failure of the sample, demonstrating the slip of the free end.

(A)

(B)

(C)

(D)

Fig. 15. RC beams tested. A) 05-Steel/3Ø6/–/–/1. B) 06-Steel/3Ø6/Steel/130/1. C) 07-NiTi/2x3Cable-/-/1. D) 08-NiTi/3ØCable/Steel/130/1.

318

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

maximum forces of 19 kN were recorded, which corresponded to longitudinal cable strains of approximately 1.74% and transverse deformations of approximately 0.57% (m = 0.33). These values are much higher than those observed in conventional steel reinforcement, which exhibits maximum longitudinal deformations of approximately 0.25% (B500S reinforcement). All samples exhibited unconventional behavior in that the slip at the free end of the bars occurred abruptly and the reinforcement almost withdrew into the concrete specimen, see Fig. 14. This behavior seems compatible with the loss of all bonding mechanisms, even residual friction, with the Ni-Ti reinforcement recovering the majority of its deformation instantly. Once the deformation disappears and the cross-section’s original size is restored, a certain amount of residual friction between the concrete and the reinforcement would be reactivated. The same behavior was observed in all samples. Practically throughout the test, deformation at the loaded end of the cable increased without producing slip at the free end. Slip at this end always remained near zero, at times even becoming negative, which suggests that rearrangement of the wires occurred within the cross-section. The test continued in this way for a long period of time (approximately 45 min at the slower loading rate) until a small amount of slip suddenly occurred and the free end withdrew into the concrete specimen. This occurred so suddenly that the displacement recorder fell because there was no longer a cable to support it (Fig. 14).

4. Tests on reinforced concrete beams In a preliminary assessment of the behavior of the Ni-Ti cables inside real concrete beam specimens, two small-scale beams longitudinally reinforced with the Ni-Ti cables were tested. One beam had no stirrups while the other had conventional 4 mm diameter stirrups at 130 mm. In addition, two other beams, longitudinally reinforced with conventional steel bars, were tested as reference beams. To make the two different types of reinforcement comparable, the reinforcement layouts, including the transverse reinforcement, were very similar in both Ni-Ti cable reinforced beams and conventional steel reinforced beams (Fig. 15). The number of rebars or cables were the same in all beams, and the total area of the longitudinal reinforcement was very similar (As = 169.6 mm2 for the reference beams and As = 187.7 mm2 for the Ni-Ti cable reinforced beams). The experimental tests reported in this paper were an extension to previous work investigating the use of Ni-Ti as shear reinforcement [9]. The nomenclature used for the beam specimens is shown in Table 5. The reference beams were named 05-Steel/3Ø6/--/--/1 (beam without shear reinforcement) and 06-Steel/3Ø6/Steel/130/1 (beam with steel stirrups). The beams reinforced with the Ni-Ti cables were named 07-NiTi/2x3Cable/-/-/1 (non-shear reinforced beam) and 08-NiTi/2x3Cable/Steel/130/1 (beam with steel stirrups). The concrete compressive strength at the age of testing, using 150 mm standard cubes, is presented in Table 5. The dimensions of the RC beams are shown in Fig. 15. The shear span in the

critical zone, a, was equal to 520 mm (see Fig. 15) with a/d, the shear span-to- effective depth ratio, approximately equal to 2.97 (nominal value). The shear span at the other side was 740 mm, with 1260 mm of beam length between support axes. The tests were carried out under displacement control using a hydraulic actuator with a maximum load capacity of 100 kN. The load was applied at a distance of 520 mm from a sliding roller support and 740 mm from a fixed roller bearing. The length of the supporting plates in contact with the beam was 60 mm (in the longitudinal axis of the beam), and the length of the loading plate was 100 mm. To monitor the behavior of the tested beam specimens, the applied loads and the displacements were measured using different instruments such as load cells, strain gauges and magnetostrictive transducers (Linear Position Transducers, see Fig. 16). Photography was also used. All of the parameters were monitored continuously by the data acquisition system. The two beams reinforced with the Ni-Ti cables (07NiTi/2x3Cable-/-/1 and 08-NiTi/3ØCable/Steel/130/1) and the reference beam without shear reinforcement (05-Steel/3Ø6/--/--/1) failed in shear, meanwhile beam 06-Steel/3Ø6/Steel/130/1 failed in bending. Table 5 and Fig. 17 summarize the most important results. When comparing to the steel reinforced beams, the specimens reinforced with Ni-Ti cables show a remarkable lower stiffness after the initial flexural cracking, mainly attributable to the lower modulus of elasticity of the cables (23.9 GPa at 20 °C). Beams without stirrups, 05-Steel/3Ø6/--/--/1 and 07-NiTi/2x3Cable-/-/1 failed for a similar shear force. However, in the case of the beam specimens with shear reinforcement, the beam reinforced with conventional steel, 06-Steel/3Ø6/Steel/130/1, was able to sustain a higher shear force, changing its failure mode from shear to bending, as may clearly be seen in Figs. 17 and 18. Fig. 18 shows the crack patterns at failure. The shear cracks at failure in beam 06-Steel/3Ø6/Steel/130/1 (Fig. 18B) indicates that although this beam failed in bending, the shear failure was not far away. Fig. 18C shows the final crack pattern for beam 07NiTi/2x3Cable/--/--/1 for the load of 25.23 kN and a deflection under the application load point equal to 17.18 mm. In this case there was practically a single crack near the application load point,

Fig. 16. Test set-up and transducer position.

Table 5 Summary of test results for the reference beams (05-Steel/3Ø6/–/–/1 and 06-Steel/3Ø6/Steel/130/1) and beams with the Ni-Ti cable (07-NiTi/3ØCable/-/-/1 and 08-NiTi/3ØCable/ Steel/130/1). Beam No.

05-Steel/3Ø6/–/–/1 06-Steel/3Ø6/Steel/130/1 07-NiTi/3ØCable/–/–/1 08-NiTi/3ØCable/Steel/130/1

fcm (MPa)

61.0 61.5 61.0 62.5

Failure mode

Shear Bending Shear Shear

Max. shear force (kN)

24.39 45.73 25.23 35.18

Deflection at max. force (mm)

d/l

2.93 38.41 17.18 20.05

1/430 1/33 1/73 1/63

319

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

50 45

Shear force [kN]

40 35 30 25 20

15

05-Steel/3Ø6/-/-/1

10

06-Steel/3Ø6/Steel/130/1 07-NiTi/3ØCable/-/-/1

5

08-NiTi/3ØCable/Steel/130/1

0 0

5

10

15

20

25

30

35

40

45

50

55

Deflection [mm] Fig. 17. Shear force-deflection diagrams for the tested beams.

Fig. 18. Crack patterns. A) 05-Steel/3Ø6/–/–/1. B) 06-Steel/3Ø6/Steel/130/1 (bending failure). C) 07-NiTi/2x3Cable-/-/1. D) 08-NiTi/3ØCable/Steel/130/1.

not evolving towards the support of the shear critical side of the beam. For comparison, note that the shear crack in the conventionally reinforced beam without stirrups (05-Steel/3Ø6/--/--/1) is located closer to the support (Fig. 18A). Fig. 18D shows the crack pattern at failure for beam 08NiTi/2x3Cable/Steel/130/1. The critical crack appears to be closer to the supports than in the beam with Ni-Ti cables without stirrups, denoting a slightly better bonding. For beams having bondfree reinforcement, or with a poor quality of bonding, cracks form

in the central part [29], compared to the situation of beams reinforced with bars with good bonding properties, in which more flexural cracks develop in the shear span and the shear critical crack is closer to the support [30]. In relation to the shear strength, the formation of a single crack in the central region does not necessarily decrease the shear strength, but it may even increase it [29]. However, it is widely known that the poor quality of bonding leads to increases in crack widths and, for service loads, it could cause serviceability and durability problems.

320

B. Mas et al. / Construction and Building Materials 148 (2017) 307–320

It is not possible to arrive to definitive conclusions regarding the design of RC members with Ni-Ti cables with only the tested beams presented in this paper. However, the results presented along this paper show that, with an increase in the modulus of elasticity of the cables, the behavior of the beams reinforced with Ni-Ti cables could offer new possibilities in structural engineering. At this research stage, to successfully design RC beams with these cables, it would be necessary to focus in the research carried out for RC beams reinforced with Glass Fiber Reinforced Polymers bars [31,32], a material with also a low elasticity modulus (approximately between 30 GPa and 70 GPa) and different bonding properties than conventional corrugated steel. Finally, performing cyclic tests on RC beams using this superelastic cable could be useful to assess the ability of the beams to recover deflection, especially if the elastic modulus of the cable is increased. 5. Conclusions In this paper, the results of a collaborative research work carried out at three universities have been presented and discussed. The paper presents the performance of an 8 mm diameter Ni-Ti SMA cable for their use in reinforced concrete application. It is concluded that the cable has huge possibilities due to its good superelastic properties, with only a marginal permanent strain ratcheting (less than 1% over 50 large strain cycles). With respect to the bonding tests, the cable has presented a different behavior compared to the conventional rebars or cables for which these tests have been designed. The low modulus of elasticity of the cable is a handicap for its application in reinforced concrete members. The modulus of elasticity should be increased by adjusting the alloy or heat treatment process to lower the transformation temperatures, in order to avoid excessive longitudinal and transverse deformations and to improve the bonding properties. The test on a RC beam with stirrups showed a slightly improved bonding behavior due to the existence of transversal reinforcement. Although this research was intended to study non-prestressed bonded applications of Ni-Ti cables, it must be highlighted that the utilization of these cables in non-bonded applications and/or in posttensioning solutions could be very interesting, bringing new possibilities to the structural concrete industry. Acknowledgments This research was developed in the framework of projects BIA2012-31432 (MINECO/FEDER – UE) and BIA2015-64672-C4-3R (AEI/FEDER – UE) and the FPU13/03028 grant. Work at the University of Michigan was funded by the General Motors/UM Collaborative Research Laboratory in Smart Materials and Structures. The authors acknowledge helpful discussions and testing specimens from Ray Bouthot of Fort Wayne Metals Research. B. Mas and A. Cladera also want to thank Dr. Kustov and his coworkers at the University of the Balearic Islands for his help in determining the phase transformation temperatures. References [1] K. Otsuka, C.M. Wayman, Shape Memory Materials, Cambridge University Press, United Kingdom, 1998. [2] L. Janke, C. Czaderski, M. Motavalli, J. Ruth, Applications of shape memory alloys in civil engineering structures – overview, limits and new ideas, Mater. Struct. Constr. 38 (2005) 578–592, http://dx.doi.org/10.1617/14323. [3] M.S. Alam, M.A. Youssef, M. Nehdi, Utilizing shape memory alloys to enhance the performance and safety of civil infrastructure: a review, Can. J. Civ. Eng. 34 (2007) 1075–1086, http://dx.doi.org/10.1139/L07-038. [4] A. Cladera, E. Oller, C. Ribas, Pilot experiences in application of shape memory alloys in structural concrete, J. Mater. Civ. Eng. 26 (2013), http://dx.doi.org/ 10.1061/(ASCE)MT.1943-5533.0000974.

[5] A. Cladera, B. Weber, C. Leinenbach, C. Czaderski, M. Shahverdi, M. Motavalli, Iron-based shape memory alloys for civil engineering structures: an overview, Constr. Build. Mater. 63 (2014) 281–293, http://dx.doi.org/10.1016/ j.conbuildmat.2014.04.032. [6] C. Czaderski, M. Shahverdi, R. Brönnimann, C. Leinenbach, M. Motavalli, Feasibility of iron-based shape memory alloy strips for prestressed strengthening of concrete structures, Constr. Build. Mater. 56 (2014) 94–105, http://dx.doi.org/10.1016/j.conbuildmat.2014.01.069. [7] M. Shahverdi, C. Czaderski, P. Annen, M. Motavalli, Strengthening of RC beams by iron-based shape memory alloy bars embedded in a shotcrete layer, Eng. Struct. 117 (2016) 263–273, http://dx.doi.org/10.1016/j.engstruct.2016.03.023. [8] A.H.M.M. Billah, Alam M. Shahria, Plastic hinge length of shape memory alloy (SMA) reinforced concrete bridge pier, Eng. Struct. 117 (2016) 321–331, http:// dx.doi.org/10.1016/j.engstruct.2016.02.050. [9] B. Mas, A. Cladera, C. Ribas, Experimental study on concrete beams reinforced with pseudoelastic Ni-Ti continuous rectangular spiral reinforcement failing in shear, Eng. Struct. 127 (2016) 759–768, http://dx.doi.org/10.1016/j. engstruct.2016.09.022. [10] B. Reedlunn, S. Daly, J. Shaw, Superelastic shape memory alloy cables: Part Iisothermal tension experiments, Int. J. Solids Struct. (2013), http://dx.doi.org/ 10.1016/j.ijsolstr.2013.03.013. [11] B. Reedlunn, S. Daly, J. Shaw, Superelastic shape memory alloy cables: Part IIsubcomponent isothermal responses, Int. J. Solids Struct. (2013), http://dx.doi. org/10.1016/j.ijsolstr.2013.03.015. [12] M.S. Saiidi, H. Wang, Exploratory study of seismic response of concrete columns with shape memory alloys reinforcement, ACI Struct. J. 103 (2006) 436–443. [13] M.A. Youssef, M.S. Alam, M. Nehdi, Experimental investigation on the seismic behavior of beam-column joints reinforced with superelastic shape memory alloys, J. Earthquake Eng. 12 (2008) 1205–1222, http://dx.doi.org/10.1080/ 13632460802003082. [14] M. Nehdi, M.S. Alam, M.A. Youssef, Development of corrosion-free concrete beam–column joint with adequate seismic energy dissipation, Eng. Struct. 32 (2010) 2518–2528, http://dx.doi.org/10.1016/j.engstruct.2010.04.020. [15] A.H.M. Muntasir Billah, M.S. Alam, Bond behavior of smooth and sand-coated shape memory alloy (SMA) rebar in concrete, Structures 5 (2016) 186–195, http://dx.doi.org/10.1016/j.istruc.2015.11.005. [16] O.E. Ozbulut, S. Daghash, M.M. Sherif, Shape memory alloy cables for structural applications, J. Mater. Civ. Eng. 28 (2016) 4015176, http://dx.doi.org/10.1061/ (ASCE)MT.1943-5533.0001457. [17] B. Carboni, W. Lacarbonara, F. Auricchio, Hysteresis of multiconfiguration assemblies of nitinol and steel strands: experiments and phenomenological identification, J. Eng. Mech. 141 (2015) 4014135, http://dx.doi.org/10.1061/ (ASCE)EM.1943-7889.0000852. [18] S. Kustov, D. Salas, E. Cesari, R. Santamarta, J. Van Humbeeck, Isothermal and athermal martensitic transformations in Ni–Ti shape memory alloys, Acta Mater. 60 (2012) 2578–2592, http://dx.doi.org/10.1016/j.actamat.2012.01.025. [19] P.K. Kumar, D.C. Lagoudas, Introduction to shape memory alloys, in: D.C. Lagoudas (Ed.), Shape Mem. Alloy. Model. Eng. Appl., vol. 1, Springer, 2008, pp. 1–51, http://dx.doi.org/10.1007/978-0-387-47685-8_1. [20] J.A. Shaw, S. Kyriakides, Thermomechanical aspects of NiTi, J. Mech. Phys. Solids 43 (1995) 1243–1281, http://dx.doi.org/10.1016/0022-5096(95)00024-D. [21] M.A. Iadicola, J.A. Shaw, Rate and thermal sensitivities of unstable transformation behavior in a shape memory alloy, Int. J. Plast 20 (2004) 577–605, http://dx.doi.org/10.1016/S0749-6419(03)00040-8. [22] B.-C. Chang, J.A. Shaw, M.A. Iadicola, Thermodynamics of shape memory alloy wire: modeling, experiments, and application, Continuum Mech. Thermodyn. 18 (2006) 83–118, http://dx.doi.org/10.1007/s00161-006-0022-9. [23] X. Zhang, P. Feng, Y. He, T. Yu, Q. Sun, Experimental study on rate dependence of macroscopic domain and stress hysteresis in NiTi shape memory alloy strips, Int. J. Mech. Sci. 52 (2010) 1660–1670, http://dx.doi.org/10.1016/j. ijmecsci.2010.08.007. [24] T.C. Rilem, RC 6 Bond test for reinforcement steel. 2. Pull-out test, 1983. In: RILEM, editor. RILEM Recomm. Test. Use Constr. Mater., E & FN SPON, 1983, pp. 218–220. [25] CEB Task Group 2.5. fib Bulletin 10 – Bond of Reinforcement in Concrete: stateof-art report. Lausanne, Switzerland: International Federation for Structural Concrete (fib), 2000. [26] E. Vos, Influence of Loading Rate and Radial Pressure on Bond in Reinforced Concrete, Delft University of Technology, 1983. [27] H. Kimura, J.O. Jirsa, Effects of bar deformation and concrete strength on bond on reinforcing steel to concrete, 1992. [28] AENOR, Norma UNE-EN 12390-2. Ensayos de hormigón endurecido. Parte 2: Fabricación y curado de probetas para ensayos de resistencia, 2009:10. [29] A. Muttoni, J. Schwartz, Behaviour of beams and punching in slabs without shear reinforcement, IABSE Colloq. 62 (1991) 703–708. [30] A. Marí, J. Bairán, A. Cladera, E. Oller, C. Ribas, Shear-flexural strength mechanical model for the design and assessment of reinforced concrete beams, Struct. Infrastruct. Eng. 11 (2015) 1399–1419, http://dx.doi.org/ 10.1080/15732479.2014.964735. [31] A. Marí, A. Cladera, E. Oller, J. Bairán, Shear design of FRP reinforced concrete beams without transverse reinforcement, Compos. Part B Eng. 57 (2014) 228– 241, http://dx.doi.org/10.1016/j.compositesb.2013.10.005. [32] C. Barris, L. Torres, I. Vilanova, C. Miàs, M. Llorens, Experimental study on crack width and crack spacing for Glass-FRP reinforced concrete beams, Eng. Struct. 131 (2017) 231–242, http://dx.doi.org/10.1016/j.engstruct.2016.11.007.