Cyclic Behavior of Long Concrete Interfaces Crossed by Steel Screws

Abstract

This study focuses on long concrete interfaces tested under cyclic actions, fastened with post-installed industrial steel screws. The overall behavior and the effect of roughness were investigated in three long interfaces, representative of connections between, e.g., a slab and a wall, a beam and a wall, etc. The results were compared with those of short interfaces tested by the authors in previous campaigns. It was observed that rough long interfaces activate their maximum resistance at small values of imposed shear slip. When roughness was reduced, the maximum resistance was also reduced, the corresponding shear slip was increased, and the overall behavior was stable. For large values of the shear slip, imposed at one end of the interface, the shear slips along it tended to be uniform, both in short and long interfaces. The limited embedment length of the screws led to their pronounced pullout. Finally, the asymmetry of resistance between the two loading directions that was observed in short interfaces was alleviated in the long ones, where also the scatter of the results was limited among duplicate specimens.

1. Introduction

Interventions to existing structures constitute a par excellence field of application of post-installed anchors of various types (see Section 2). Indeed, repair and/or strengthening of structures involve the increase of dimensions of existing elements and/or the addition of new members and, thus, the creation of interfaces between concretes cast at different times. To reach a composite function of the resulting structural members, adequate behavior of interfaces is to be ensured. This is a topic of interest, especially for structures in seismic areas, where interfaces are expected to be subjected to cyclic actions, affecting their stiffness and resistance characteristics. However, up to now, the testing effort relevant to the seismic behavior of interfaces is rather limited [1,2] compared to the significance of the topic.

Furthermore, in many interventions to existing structures, interfaces are of significant length, and they are crossed by a series of reinforcing elements. There cannot be an unequivocal definition of long or short interfaces. The meaning of those terms within this work is the following: An interface is considered as short when its length is of the order of the sectional dimension of a linear element (e.g., a column or a beam) and it is crossed by a small number of reinforcing elements (e.g., two to three), to respect the requirements for minimum-edge and inter-anchor distances [3]. An interface is termed as long when it is representative of the connection between two plane elements (e.g., a slab and a wall) or a linear and a plane element (e.g., a beam and a wall) and it is crossed by a large number of reinforcing elements, always fulfilling the requirements related to edge and inter-anchor distances [3]. As shown in Section 2, most of the tests on interfaces are performed on short interfaces. Thus, sensible questions, such as “are all connectors equally activated over the entire length of a long interface” or “is there any chance of redistribution of forces among a large number of connectors”, cannot be answered on the basis of the published experimental results.

Another aspect needing more attention is the investigation of behavior (monotonic and cyclic) and, thus, the proper use of types of connectors other than pieces of steel reinforcing bars. For the time being, the latter constitute the most used type of interface reinforcement, and definitely the most investigated one (see Section 2). Nonetheless, industrial anchors (such as powder-driven nails or steel screws, etc.) may constitute a viable alternative in cases where the intervention is to be urgently applied, either for safety or for economic/social reasons. For example, to avoid fast deterioration of the state of a building after a strong earthquake, the filling of some frame spans with reinforced or unreinforced concrete walls might be sought. In such an application, the installation of industrial anchors (not requiring the drilling of holes to concrete and time for hardening of the bonding material) as reinforcement of interfaces between the existing frame and the to-be-added wall may be a good choice. Moreover, to reinstate the decayed deck of a bridge, the construction of a thin RC overlay may be needed, connected to the existing concrete using a large number of connectors. In such applications, the use of industrial anchors, easy and fast to apply, is worth investigating as an alternative to steel reinforcing bars.

Being focused on the seismic behavior of concrete-to-concrete interfaces, the authors of this paper have also investigated the case of using steel screws as interface reinforcement. The results of tests, published in [4], were obtained on interfaces 500 mm long and 200 mm wide, crossed by three 10 mm diameter screws and subjected to cyclic shear.

To investigate possible differences between short and long interfaces crossed by screws and subjected to cyclic actions, and to thus contribute to the design of relevant applications, a testing campaign was planned and executed. In the current study, the results of testing long interfaces reinforced with steel screws of short length are reported. For the sake of comparison, selected experimental results obtained from testing short interfaces, presented elsewhere [4], are repeated here to allow for identification of similarities and differences with the long ones.

2. Literature Survey

The literature is rich in tests of reinforced interfaces between concretes cast at different times. As reported in [1], more than 60% of the 868 test results on cold joints are tests of interfaces reinforced with cast-in reinforcing bars. Another 20% consist of interfaces reinforced with post-installed reinforcing bars, bonded by means of adhesives. Less than 16% of all interfaces were reinforced using various types of alternative connectors. More specifically, Choi et al. [5,6] tested interfaces crossed by powder-driven nails, whereas Saari et al. [7] and Wallenfelsz [8] investigated the case of headed studs as interface reinforcement, and Alkatan [9] tested interfaces using cast-in GFRP bars of various shapes (closed hoops, headed bars, etc.). Finally, Menkulasi and Roberts- Wollmann [10] simulated connections between precast concrete elements, using various types of reinforcement.

Among the 148 interfaces reinforced with alternative connectors, more than 100 were of length smaller than 600 mm, while the interface length exceeded 1000 mm in 20 specimens, simulating concrete girder to deck connections [11]. The interface reinforcement was in the form of cast-in threaded rods provided with a nut at their ends. The total embedment length was equal to 7 d or 9.33 d (d denotes the diameter of the rods). All interfaces were subjected to monotonic loading, whereas they were crossed by a limited number of connectors (1 to 4). Thus, the reinforcing ratio ranged between 0.016% and 0.27%. In that work, no information was provided about the observed failure mode or the effect of the embedment length. Furthermore, the shear slip and the crack width were measured at the two ends and at the mid-length of the interfaces, respectively, and thus, there are no data about the development of those displacements in various locations along the interface.

It is noted that, in all the aforementioned studies [5,6,7,8,9,10,11], independently of the dimensions of interfaces crossed by various types of connectors (other than reinforcing bars), the number of reinforcing elements was small, whereas the shear slip and the crack opening were measured as in the case of tests by Kamel [11]. Furthermore, all tests reported in the literature were performed under monotonic loading, with the exception of those reported in [4,12]. Cyclic loading was also applied in the tests by Saari et al. [7], where connections between a concrete infill and a steel frame were simulated.

The same type of reinforcement used in the work reported herein, namely, steel screws, was used by Cattaneo et al. [12] and Palieraki et al. [4]. The interfaces in both testing campaigns were 500 mm long and 200 mm wide, crossed by three screws in line. In the publication by Cattaneo et al. [12], the experimental results were presented to the extent that serves the purpose of the work, i.e., to investigate the effect of specimen configuration on the interface resistance and overall behavior and to provide a comparison between numerically calculated and experimentally obtained results. Thus, herein, a comparison is presented between the experimental results obtained on long interfaces and those reported by the authors in [4].

3. The Experimental Campaign

3.1. Geometry of Specimens

Three specimens with long interfaces were constructed. Their shape was similar to those presented in [4]. They consisted of two reinforced concrete blocks (Figure 1) cast separately into metal molds (Figure 2), approximately 28 days one after the other. The interfaces were 1200 mm long and 300 mm wide. The shape of the specimens allowed for cyclic shear slips to be imposed to the interfaces with zero eccentricity. The specimens were provided with dense reinforcement (Figure 2) to avoid parasitic cracking, e.g., at the regions of fixing the specimens to the setup. Furthermore, this reinforcement was placed out of the region where the interface reinforcement was located (100 mm away from the interface) to avoid its contribution to the behavior of the interface. Finally, to make sure that the crack would occur at the interface, the width of the specimens was gradually reduced from 500 mm to 300 mm, as shown in Figure 1. The steel screws were positioned mid-width of the interface. Their side cover (~15 d, d being the anchor diameter = 10 mm) and the clear distance between consecutive anchors (9 d) were chosen (Figure 1), such that premature concrete splitting and interaction between anchors, respectively, was avoided. The aforementioned distances, checked against the provisions of [3], are valid for reinforcing elements crossing interfaces subjected to shear.

3.2. Post-Installed Interface Reinforcement, Roughness, and Concrete Strength

The interfaces were reinforced with 10 mm diameter concrete screws (mechanical dowels; Figure 3), installed according to the specifications of the manufacturer. The characteristics of the concrete screws are presented in

Table 1. Characteristics of concrete screws.

Diameter, d (mm)	10
Characteristic value of yield strength, fsy (MPa)	690
Characteristic value of tensile strength, fsu (MPa)	805
Yield strain, εy (‰)	3.45
Embedment into old concrete block (mm)	67
Embedment into new concrete block (mm)	65

. To allow for a comparison between long and short interfaces,

Table 2. Characteristics of specimens, and maximum experimental and calculated capacity.

Specimen	Number of Connectors	ρ (%)	ρfsy	Roughness (mm)	fc,old (MPa)	fc,new (MPa)	τu (MPa)	τu,calc (MPa)	τu/τu,calc
L-R-22/G/65/0.1	10	0.218	1.50	2.84	22	26	1.36	1.18	1.15
L-R-23/G/65/0.1	10	0.218	1.50	2.66	23	27	1.63	1.22	1.34
L-S-27/G/65/0.1	10	0.218	1.50	Smooth	30	27	0.88	0.81	1.09
R-32/G/65/0.1	3	0.236	1.63	6.82	32	42	1.64	1.63	1.00
R2-32/G/65/0.1	3	0.236	1.63	5.46	32	42	1.66	1.63	1.02
R-24/G/65/0.1	3	0.236	1.63	1.60	24	40	1.53	1.30	1.18
S-25/G/65/0.1	3	0.236	1.63	Smooth	25	43	1.28	0.81	1.58
S-32/G/65/0.1	3	0.236	1.63	Smooth	32	43.5	0.74	0.97	0.76

Notation. Specimen name: L, long interface; R, mechanical anchorage at rough interface; S, mechanical anchorage at smooth interface. The first number indicates the compressive strength of the weaker concrete block (in MPa). G, interface reinforced with concrete screws. The second number indicates the embedment length in the new part (in mm). The third number indicates the magnitude of the cyclic shear slip imposed during the first set of cycles (in mm). τ_u (MPa) is the maximum experimental shear resistance of the interface, given as the average value between the two loading directions for the first cycle. τ_u,calc (MPa) is the maximum predicted shear resistance of the interface, based on the NTUA model [2].

presents both groups of specimens, which were comparable in terms of the parameter ρf_sy, with ρ being the geometrical reinforcement ratio and f_sy being the yield strength of the steel. It is noted that the reinforcing ratio was dictated by the decision to use three connectors in short interfaces, to avoid any parasitic rotation of the interface during testing. Thus, in the long interfaces, an adequate number of screws was used to reach approximately the same reinforcing parameter as for their short counterparts.

After casting the portion of the specimen simulating the existing concrete, the interface was either artificially roughened (chipped, using a pickaxe) or remained untreated (Figure 3). Medium roughness (2.50 mm–3.00 mm) of the surface was sought (Table 2), measured by the standardized “Sand Patch” Test [13]. The smooth interface was achieved with as-cast concrete: after the concrete was laid in the formwork and leveled, a trowel was used to smoothen and fine-level the surface, resulting in a rather smooth surface with minimal asperities.

Ready-mixed concrete with limestone coarse aggregates of 32 mm maximum diameter was used for the construction of the specimens. For each batch of concrete, conventional cylindrical specimens (diameter 150 mm and height 300 mm) were cast. The cylinders were tested (Table 2) to failure the day of testing the respective specimens.

4. Test Setup, Instrumentation, and Loading History

4.1. Test Setup

Figure 4a shows the experimental setup. The specimen “S” was placed on top of the heavily reinforced concrete slab “SL” fixed to the strong floor of the laboratory. A system of four steel columns, HEB300 (“c”), connected with two steel beams, HEB400 (“b”), with transverse stiffeners, was used to keep the old concrete block fixed. An actuator, “A” (maximum capacity = ±1000 kN), was placed horizontally, attached to the stiff system of steel reaction frames (“St”). The portion of the specimen simulating the new concrete was attached to the actuator by means of eight steel rods, “R” (passing through steel plates, “P”), so that the axis of the piston was aligned with the level of the interface and no eccentricity occurred.

4.2. Displacement Transducers

Figure 4b shows the locations of the displacement transducers used to measure displacements, with the objectives described in the following.

To measure the shear slip, transducers were placed at several locations along the interface. Thus, in addition to four transducers (two per specimen end) measuring the shear slip at the front and the back end of the long interface and used for the control of the applied displacements during testing, ten more transducers were used to measure the shear slip between adjacent anchors. Due to the limited distance between consecutive anchors compared to the space needed for the arrangement of the measuring devices, half of them were placed at one side and the other half of them at the other side of the specimen. All the transducers used for the measurement of shear slip had a 25 mm range, with an accuracy of 0.10% (i.e., 25 μm).

To measure the width of the crack at various locations along the interface, ten transducers (five per specimen face) were placed perpendicularly to the interface. All the transducers used for the measurement of crack opening had a 50 mm range, with an accuracy of 0.10% (i.e., 50 μm).

Finally, to check the efficient fixing of the specimen and the setup, several LVDTs were placed at adequate locations, to monitor any parasitic movement during testing. Those measurements were practically equal to zero throughout testing and, thus, they are not presented in this paper.

4.3. Strain Gauges

Strain gauges (with gauge length equal to 5 mm) were used to record the strains developed in the interface reinforcement. Two strain gauges were provided to each dowel, both positioned in the new part of concrete, as the part of the screw that was embedded into the old concrete block was threaded. To capture the maximum strains developing in the screws, the strain gauges should be located at the interface, encountering the risk of their damage during casting of the concrete. Thus, the strain gauges were located with their mid-length at approximately 15 mm (1.5 d) from the interface. It is noted that this distance from the interface resulted in measuring strain values somehow smaller than those occurring at the interface. This is valid both for strains due to the clamping effect and those due to dowel action that reached their maximum value at a distance from the interface of the order of 0.5 d–1.0 d [14,15,16].

In Figure 5, each strain gauge code includes a number and a letter: the number indicates the serial number of each dowel, whereas the letter indicates the strain gauge location (left or right side of the screw). The strain gauges, oriented along the direction of loading, were positioned on opposite sides of the screw, parallel to its vertical axis. Their position allowed, simultaneously, (a) to record the strains developing jointly by the clamping effect and dowel action, and (b) to calculate the contribution of each of the two mechanisms to the total strain, following the procedure described in [17] and explained in Section 5.4.

4.4. Loading History

Cyclic-displacement-controlled tests were performed, with the imposed shear slip increasing stepwise until the measured shear resistance was significantly reduced. Three full cycles were performed at each shear slip step, with the first set having an amplitude of ±0.10 mm. The shear slip was imposed to the interface by the actuator at a low speed (0.005 mm/min to 0.10 mm/min, approximately). The slip rate was based on the transducer’s measurements, while the higher speed was adopted for large values of imposed shear slips.

5. Test Results

5.1. Failure Mode

In all three specimens, a crack along the interface occurred at different shear slip values, as shown in

Table 3. Main results at the step of maximum capacity and of final failure.

Specimen	τu (MPa)	su (mm)	wu (mm)	Damage (at τu)	smax,cyclic (mm)	Length of the Falling Branch (mm)	Damage at Test Termination
L-R-22/G/65/0.1	1.36	0.10	0.07	Crack along the interface	1.05	0.95	Crack along the interface, crack parallel to the interface (concrete cone) in the old part
L-R-23/G/65/0.1	1.63	0.10	0.01	Crack along the interface	3.00	2.90	Diagonal cracks (concrete struts) in the old part, crack parallel to the interface (concrete cone) in the old part
L-S-27/G/65/0.1	0.88	3.00	0.54	Crack along the interface	11.00	8.00	Crack along the interface and connectors’ fracture
R-32/G/65/0.1	1.64	1.10	3.12	Crack along the interface, crack parallel to the interface in the old part, diagonal crack in the new part	1.40	0.30	Crack parallel to the interface in the old part, diagonal crack in the new part
R2-32/G/65/0.1	1.66	1.50	2.87	Crack along the interface, crack parallel to the interface (concrete cone) in the old part	1.50	0	Crack parallel to the interface (concrete cone) in the old part
R-24/G/65/0.1	1.53	1.20	0.91	Crack along the interface	3.20	2.00	Crack parallel to the interface (concrete cone) in the old part
S-25/G/65/0.1	1.28	1.70	0.71	Crack along the interface	3.00	1.30	Diagonal crack in the new part
S-32/G/65/0.1	0.74	1.20	0.58	Crack along the interface, diagonal crack in the new part	1.70	0.95	Diagonal crack in the new part

Notation. s_u (mm) is the shear slip value corresponding to the mobilization of the maximum capacity of the interface. τ_u (MPa) is the maximum experimental shear resistance of the interface, given as the average value between the two loading directions for the first cycle. w_u (mm) is the interface crack opening corresponding to the mobilization of the maximum capacity of the interface. The average value between the two loading directions for the first cycle is given. s_max,cyclic (mm) is the shear slip value corresponding to the termination of the cyclic test. The decision to terminate the cyclic loading was based on the observed resistance degradation, which should exceed 30–35% of the maximum. Wherever possible, the cyclic test was followed by monotonically increasing shear slips until failure.

. However, the failure mode depended on the interface roughness.

The two specimens with rough interfaces, namely, L-R-22/G/65/0.1 and L-R-23/G/65/0.1, exhibited similar behavior. In both specimens, a crack along the interface was developed at 0.10 mm. For a shear slip approximately equal to 0.75 mm, i.e., after the maximum shear resistance was attained, a crack almost parallel to the interface was formed at the old block of both specimens, at a distance almost equal to 10 cm from the interface, denoting the development of a concrete cone. It is noted that concrete cone failure is expected, given that the length of the interface reinforcement is very limited [4]. Additionally, the distance between the connectors was smaller than that required for the development of independent concrete cones for each distinct connector [18]; therefore, it led to the formation of an entire surface almost parallel to the interface.

In specimen L-R-22/G/65/0.1, this crack was of small width, and no further cracks occurred up to a value of shear slip equal to that corresponding to the termination of the cyclic test, s_max,cyclic = 1.05 mm, (Figure 6a; Table 3). Subsequently, as the resistance of the interface was significantly reduced, the interface was monotonically loaded, reaching an imposed shear slip equal to 23 mm (Figure 7a). The L-R-23/G/65/0.1 test was completed at a shear slip approximately equal to 5.00 mm (Figure 7b), when additional inclined cracks—concrete struts—developed in the old part, along the entire interface length (Figure 6b). This was the main difference between the two tests, since only the latter exhibited extended, dense cracks at the old part of the specimen, which led to the termination of the experiment. After the separation of the two concrete blocks, when the two tests were completed, all the screws remained in the new part, while their deformed shape was that typical for dowels subjected to shear (Figure 6c).

The L-S-27/G/65/0.1 specimen of the smooth interface exhibited quite different behavior. A crack along the interface was visible at shear slip equal to 0.10 mm. The new concrete block was sliding relatively to the fixed old block, and no other cracks were formed on the specimen, even for a shear slip up to 11 mm. After that, with the resistance being reduced, the test continued monotonically, reaching an imposed shear slip equal to 23 mm (Figure 7c). Still, only the crack along the interface developed (Figure 6d). At this step, fracture of the screws occurred, and the test was finished. After the completion of the test, the two pieces of concrete were very easily separated, and the fracture of all screws, as well as local crushing of the concrete around the screws, was visible (Figure 6e). The overall behavior of this smooth interface is commented upon in detail in the following sections.

5.2. Crack Openings and Tensile Strains of the Bars Crossing the Interface

Figure 7 presents the hysteresis loops of the three specimens. It is noted that the shear slip values used in the graphs were measured as the average value of the four transducers (two per interface end), recording the shear slip at the front and the back end of the long interface. The response in the two loading directions was rather symmetrical for all specimens.

The rough interfaces reached their maximum resistance at a shear slip value equal to 0.10 mm, and a falling branch followed. However, the post-peak behavior was quite smooth and stable, characterized by a small inclination of the falling branch, allowing for large shear slip values to be imposed. Of course, a gradual decrease in the stiffness of the interface was observed, paired with an increase in the residual slip (at zero resistance). These features were attributed to the gradual deterioration of the concrete around the screws, also observed by inspection after the respective tests were completed.

The smooth interface reached a resistance significantly lower compared to the rough interface. The maximum response was mobilized for a high value of imposed shear slip (3.00 mm), and the only crack that occurred throughout the test was that along the interface.

As shown in Figure 8a, all three specimens exhibited a moderate reduction (by ~20%) in their resistance after three cycles at an imposed slip equal to that corresponding to the maximum interface resistance.

The diagrams presented in Figure 9 are also typical, where the opening of the crack along the interface is plotted against the shear slip values. The slip values in Figure 9 were calculated as the average value of the four transducers located at the two ends of the specimen (in correspondence with the slip values used in the hysteretic loops of Figure 7). The crack width values were calculated as the average value of the ten transducers (five per specimen face) recording the crack opening. The main features observed on the graphs of Figure 9 are (a) the gradual increase of the crack opening, as the imposed shear slip increased, and (b) the rather high value of the residual crack opening (at zero shear slip), which reached 60% of the maximum and almost 100%, for rough and smooth interfaces, respectively. As explained in [4], the large values of residual crack widths are attributed to the limited embedment length of the interface reinforcement, which allows for significant pullout of the connectors, a mechanism that is not reversible when the interface is unloaded. The assumption was further confirmed by the behavior of the smooth interface, where the crack openings as large as 0.50 mm could not be attributed to the roughness of the interface and, furthermore, those large crack openings were permanent, showing that their cause is not reversible.

5.3. Interface Roughness

The data of Table 2 seem to confirm the positive effect of interface roughness on the interface resistance. In the tests reported here, the two rough interfaces were of medium roughness (2.50 mm–3.00 mm), and they exhibited higher shear resistance than their smooth counterpart. It is noted though that, according to the previous study on short interfaces reinforced with screws [4], very rough interfaces (roughness between 5.50 mm and 7.00 mm, approximately) exhibited smaller resistance than those of medium roughness, when subjected to cyclic actions.

In Table 3, it can also be noticed that the two rough interfaces reached their maximum resistance at a small shear slip value (s_u = 0.10 mm), in contrast to the smooth interface, for which s_u was equal to 3.00 mm. The latter may affect the design of interfaces within a displacement-based design of the entire structure. Indeed, in such a case, values of “acceptable shear slip” along the interfaces were defined, depending on the adopted performance level.

5.4. Steel Strains

Figure 10 presents typical diagrams of the recorded steel strains (positive sign for tension) in function of the crack width. For the sake of clarity of the graphs, the measurements of the strain gauges on the left side of the five screws closer to the actuator are presented. The measurements for the remaining five screws were similar.

It is well known (see also [2,19,20]) that when an interface is subjected to shear, strains develop in the interface reinforcement due to the joint action of two distinct mechanisms, namely, friction and dowel action. Indeed, shear slip at the interface causes opening of the crack due to the asperities of the interface. As the crack at the interface occurs and is gradually increasing (for increasing shear slip), the interface reinforcement develops tensile strains (and stresses) due to its pullout. The tensile stresses of the reinforcement are equilibrated by compressive stresses on the interface. These stresses, known to be the result of the clamping effect, multiplied by a friction coefficient depending on the characteristics of the interface, constitute the contribution of the friction to the shear resistance of the interface. In parallel, the reinforcement crossing the sheared interface is deforming, and strains develop (of opposite sign in the compressed and tensioned zones of the connector under bending), while compression is exerted by the interface reinforcement to the concrete, close to the deformed portion of the interface reinforcement. The reaction to this compression constitutes the contribution of the deforming connectors to the shear resistance of the interface, a mechanism termed dowel action. The strains plotted in Figure 10 are due to both shear-resisting mechanisms. Strain values higher than 3.45‰ denote the yield of the screws (Table 1). It is interesting to observe that in all three specimens, the strains developed in the screws due to the joint action of the two shear transfer mechanisms exceeded the yield strain of the steel. This result seems to be unexpected, due to the quite small embedment length of the anchors to the concrete and, by way of consequence, to the expected limited contribution of the clamping effect, known to generate friction and be the main mechanism contributing to the transfer of shear across interfaces. The arrangement of the strain gauges (Figure 5) in the part of the screw that protrudes to the new concrete allowed to calculate the contribution of each mechanism to the strains measured by the strain gauges. In order to evaluate the contribution of each mechanism to the development of tensile strains in the screws, the following procedure (described in detail in [17]) was applied, also illustrated in the example of Figure 11. The distribution of the steel strains at a screw’s cross-section was the result of the combination of the uniformly distributed tensile strains due to pullout and the strains due to dowel action (triangular, with opposite values at the two ends). The strain measured at the most tensioned screw side represents the sum of the pullout strain and the positive dowel strain. The strain measured at the least tensioned screw side represents the algebraic sum of the pullout strain and the negative dowel strain.

By applying this procedure, the plots of Figure 12 were derived. Those plots show that the contribution of dowel action prevailed, even for rough interfaces. This is in accordance with the measurements of crack opening in function of the imposed shear slip, as presented in Figure 9. As commented in Section 5.2, the large values of residual shear slips were attributed to the pullout of the shallow screws, occurring for small tensile strains and respective stresses.

6. Comparison between Long and Short Interfaces

During previous experimental campaigns [4,17,21], the observation was made that tests on short interfaces are representative of the behavior of single connectors. With the measured shear slip being uniform along the interface, all reinforcing elements crossing it are subjected to the same value of shear slip. Thus, the question arose whether real applications, where long interfaces are crossed by a large number of connectors, can be handled based on the formulae derived mainly from tests on interfaces of rather small dimensions [2,20]. Is there any redistribution of actions among the connectors that may allow for long interfaces to be subjected to larger shear slips without significant degradation? Does the cyclic behavior of long interfaces exhibit less asymmetry than that of short interfaces? Such a behavior would be favorable, especially under seismic conditions. Thus, in the following sections, a comparison is presented between short and long interfaces of similar characteristics.

6.1. Failure Mode and Overall Behavior

The data of Table 3 show that the cracking and the damage of long interfaces was similar to their short counterparts. In short rough interfaces, the typical failure mode was that of the formation of a crack in the old part, parallel to the interface, close to the end of the embedment length of the reinforcement, denoting the creation of a concrete cone. In long rough interfaces, the failure was due to the crack parallel to the interface (concrete cone) and, in one case, also to the formation of concrete struts in the old part. These failures in the long rough interfaces were recorded for an applied shear slip significantly larger than the one corresponding to the attainment of the maximum interface resistance (s_u).

Regarding the short and long smooth interfaces, a difference was observed in their failure mode: the first failed due to a diagonal crack developed in the new part, whereas the latter failed due to the crack along the interface, which caused the fracture of the dowels, for a rather large value of the applied shear slip.

In general, in both short and long interfaces reinforced with concrete screws, the falling branch of the shear force vs. shear slip curves was not steep and, thus, the behavior of the interfaces was not characterized by pronounced brittleness. Figure 8 shows the degradation of the maximum resistance due to cycling. Although the long and short interface specimens were not directly comparable, as they differed in the roughness of the interface (Table 2), it seems that, as a rule, the short interfaces exhibited more significant degradation due to cycling than the long ones (comparing, for example, specimens R-24/G/65/0.1 and L-R-23/G/65/0.1). The same observation was valid for smooth interfaces, as the comparison between specimens S-25/G/65/0.1, S-32/G/65/0.1, and L-S-27/G/65/0.1 showed, even though the difference was not so pronounced.

In Figure 13, the normalized envelopes are plotted for the first cycle of short and long interfaces to better investigate the asymmetry of behavior in the two loading directions. The data plotted in Figure 13 reveal that the envelopes for long interfaces were quite similar in the two loading directions. This characteristic was not observed for the short interfaces, where the asymmetry of resistance between the two loading directions was rather significant.

On the other hand, the data presented in Table 3 show that, although the maximum resistance, τ_u, of long interfaces did not differ significantly from that of the short ones, it was mobilized at smaller shear slip values for the rough interfaces, whereas the long interfaces seemed to exhibit longer falling branches (Figure 14) after the attainment of the maximum resistance until failure occurred and the test was terminated. The difference was much more pronounced in the case of smooth interfaces. This characteristic, paired with the symmetry of hysteresis loops in the two loading directions, showed that the behavior of long interfaces presented more favorable properties than their short counterparts. Furthermore, the model developed by the authors for the calculation of the maximum resistance of interfaces [2], although based on experimental data obtained basically through short interfaces, could quite accurately predict the maximum resistance of long interfaces as well (Table 2). It should be noted that, when calculating the maximum shear resistance of an interface, the relevant equation for concrete breakout [22] was also accounted for. The tensile stress developed in the bars was taken as the smaller of two values, namely, the tensile stress due to pullout, provided by the manufacturer of the screws, and that corresponding to concrete cone failure, according to [22]. In the specimens tested within this work (both short and long interfaces), the two values of the tensile stresses were very close to each other, thus not allowing for a clear identification of a governing failure mechanism (pullout of the screws or breakout cone). This seems to be in agreement with the observed failure mode, presented in Section 5.1 and described in Table 3.

6.2. Distribution of Shear Slips and Steel Strains

Figure 15 shows the distribution of shear slips, crack openings, and steel strains at each dowel position along the long interfaces as the mean values of the two loading directions. It should be remembered that in case of interfaces 500 mm long, crossed by three connectors, the shear slip was measured in two locations along the interface [4], and the obtained values were practically of the same magnitude. In long interfaces, at the early loading stage, the distribution of the relative shear slips was not uniform, and the dowels near the edges were mobilized more, compared to those placed in the middle area of the interfaces. This was attributed to the fact that the shear slips were imposed at the front and at the rear end of the specimens for the two loading directions. This is illustrated in Figure 16, where—for one of the specimens—the shear slips, crack openings, and steel strains are plotted for each loading direction, separately. Indeed, the screws that were located closer to the point where the slip was imposed deformed more than the ones located at a larger distance. However, as the imposed shear slip increased, the distribution of shear slips along the interfaces tended to become uniform on average (Figure 15), as the entire new part slid uniformly relative to the old part. This was valid both for the smooth and the rough interfaces.

As for the steel strains recorded by the strain gauges, it was observed that in the long interfaces, the screws crossing the interface were not constantly in tension, as the prevailing dowel action resulted in positive and negative strains of comparable amplitudes. This observation is consistent with the short interfaces (indicatively, compare Figure 10c and Figure 12e,f with Figure 17 for smooth interfaces).

7. Summary and Conclusions

In this experimental campaign, long interfaces were investigated, crossed by steel screws and post-installed into the existing concrete. The results acquired from testing these interfaces were analyzed and compared with short interfaces of similar characteristics, tested in previous experimental campaigns. The experimental results allowed for the following conclusions to be formulated:

• To record, in detail, the behavior of interfaces, both under monotonic and cyclic actions, adequate experimental setup and instrumentation is needed. The testing procedure adopted in this work was proven to be efficient, as it allowed for shear slips and crack openings to be reliably measured in several locations along the interfaces, thus permitting to closely follow the activation of the interfaces throughout testing.
• Furthermore, strain measurements on all screws, along with the technique developed to calculate the contribution of each distinct shear transfer mechanism to the total measured strain, confirmed the rather limited activation of the clamping effect (due to the small embedment depth of the screws) and, on the contrary, the significant contribution of dowel action.
• Rough long interfaces (in contrast to rough short interfaces) activated their maximum resistance at small values of imposed shear slip (of the order of 0.10 mm), at which point only a crack along the interfaces occurred. After that, the response was shown as smoothly degrading. The primary crack was that developed along the interface, whereas a crack parallel to the interface (concrete cone) and, in one case, also concrete struts in the old part, appeared at an advanced state of imposed shear slip.
• Although the results of only one smooth long interface were available, the well-known dependence of the maximum resistance on the interface roughness was confirmed. Indeed, when roughness was reduced, the maximum resistance was reduced, while the corresponding shear slip tended to increase, and the overall behavior was stable.
• The limited embedment length of the steel screws led to their pronounced pullout, documented by the large values of residual crack openings at zero shear slip. Furthermore, in contrast to interfaces crossed by post-installed steel reinforcing bars, where the governing mechanism was the clamping effect, in interfaces crossed by screws (either smooth or rough), the contribution of the dowel mechanism was enhanced.
• The asymmetry of resistance between the two loading directions, which was observed in short interfaces, was alleviated when long ones were concerned, for both levels of the examined roughness. Furthermore, the scatter of the results was limited among duplicate specimens. These observations served as strong indications of the redistribution capacity inside a system, evidencing that the behavior of groups of anchors was differentiated from that of single anchors, not in terms of maximum resistance, but in terms of overall behavior.
• The design equation developed by the authors, applied to predict the shear resistance of the specimens, was proven to be quite accurate.