Seismic performance of prefabricated beam-to-column joint with replaceable energy-dissipating steel hinge

This study presents a novel energy-dissipating prefabricated joint for connecting beam to column in a precast frame structure. The joints are characterized by a replaceable steel hinge and a prefabricated steel tube confined joint core, providing advantages for precast concrete reinforced frames, such as complete assembly, damage control, and maintainability of the structure after an earthquake. The hysteretic behavior of the proposed prefabricated joint was studied through two tests. First, a full-scale prefabricated joint was tested under cyclic loading until failure. On the basis of the initial test, only four weakened dissipaters of the steel hinges in the prefabricated joint were replaced and the second test was conducted to investigate the restorable functional characteristics of the proposed prefabricated joints. For comparison, a reference monolithic joint was also tested. The experimental results demonstrate that the novel prefabricated beam-to-column joint displayed excellent hysteretic performance, and corresponding to the monolithic joint, the load-bearing, energy dissipation, and deformation capacity were improved. The damage of the prefabricated joint was concentrated on the weakened dissipaters of the steel hinges, indicating that the failure mode and damage degree of the prefabricated joint can be controlled. In the second test, the prefabricated joint exhibited similar hysteretic behavior to that of the first test; however, the initial stiffness was slightly lower. Therefore, the prefabricated joint can meet the replaceability requirement and achieve satisfactory beam-to-column joint function recovery after an earthquake.


Introduction
Prefabricated buildings have attracted increased attention in research owing to their advantages of high quality, easy in situ operation, short construction period, and cost savings. Moreover, prefabricated buildings meet the growing demand for building industrialization throughout the world and also satisfy the requirements of environmental friendliness and sustainability. Consequently, prefabricated structures have been extensively applied in building construction (ElliottKim 2002;VanGeem 2006;Jaillon and Poon 2014;Chang et al. 2018;Wang and Bi 2019). Precast concrete frames account for a large proportion of prefabricated buildings owing to their flexibility regarding architectural layout and standardization of the modular components. However, in contrast to conventional cast-in-place structures, a prefabricated building structure demonstrates poor integrity and most prefabricated building structures lose their normal functionality with connection failure due to complex stress conditions, which is important in force transfer between components; in particular, the beam and column. Beam-to-column joints are a vulnerable component and thus it is important to develop good mechanical performance for their use in prefabricated building structures.
Therefore, substantial research has been devoted to developing new connections; particularly, beam-to-column joints (Engström 2008;Ghayeb et al. 2020;Breccolotti et al. 2016;Bahrami et al. 2017;Guan et al. 2016;Nzabonimpa et al. 2018;Fan et al. 2020) with excellent mechanical behavior for application in prefabricated building structures to achieve good structural integrity. Previous studies indicated that connections between precast frame members could be classified into three main classes: dry connection, wet connection, and hybrid connection. The equivalent monolithic reinforced concrete frame system was developed using cast-in-situ joints or fabricated connections (Vidjeapriya and Jaya 2013;Choi et al. 2013;Zhang et al. 2020a, b;Ryu et al. 2007;Joergensen and Hoang 2013). In addition, considerable research has been conducted to better understand the mechanical behavior and evaluate the seismic performance of newly developed connections or precast structures with new connections (Ghayeb et al. 2020). Restrepo et al. (1995) conducted experimental research on four mid-span connections and two beam-to-column connections. It was shown that the connection configurations had a slight effect on the mechanical behavior of the structure. All tested connections exhibited good ductility, energy-dissipation capacity, and bearing resistance and such connections could be constructed to emulate monolithic cast-in-place connections. Zhao et al. (2004) investigated the behavior of a precast beam-to-column joint with high-strength concrete, using a full-scale test, and the structure with this new connection demonstrated equivalently desirable seismic performance, such as failure mode, hysteresis behavior, and ductility. Parastesh et al. (2014) proposed a novel ductile moment-resisting connection for precast RC frames in seismic regions, which could provide excellent structural integrity and rapid construction. The seismic performance was investigated experimentally and the results demonstrated that the proposed connections could develop adequate flexural strength, as well as higher ductility and energy consumption. Moreover, the failure mode could be improved by concentrating damage to the plastic hinge region.
Prestressed steel strands can effectively provide a self-centering capacity for beamcolumn joints and improve the energy dissipation capacity, without serious structural damage (Deng et al. 2013;Song et al. 2014;Koshikawa 2017;Yan et al. 2018;Yang et al. 2020;Li et al. 2008;Wang et al. 2018aWang et al. , b, 2019a. Li et al. (2008) researched precast beam-to-column joint subjected to bidirectional lateral loading to assess precast RC frames under the action of an earthquake. The results demonstrated that the prestressed connection exhibited similar performance to the cast-in-place connection. This column remained damage-free during the loading period and an equivalent viscous damping pattern was established, considering the effects of prestress loss and energy dissipater yield. Wang et al. (2018a) proposed a prefabricated prestressed beam-column joint, which uses replaceable mild steel reinforced bars to provide an energy disappearance capacity and steel strands to provide a self-centering capacity. The effectiveness of the proposed joint was experimentally validated. In addition, the configuration was improved based on the results and a parametric study using a calibrated FEA model was conducted to determine the essential parameters the effect of the parameters of interest on joint performance . Through experiments and numerical research, the design specifications to ensure full utilization of the joints were determined. Wang et al. (2018b developed an all-steel bamboo-shaped energy dissipater and applied it to precast concrete beam-column joints, which played an important role in fusing and protecting the main structure. Through experimental research on five precast concrete connections under cyclic loading, it was found that these connections exhibit good hysteretic performance and self-centering action. Using supplemental energy dissipaters in connection is another research topic aimed at enhancing the energy dissipation capacity and concentrating plastic damage to target members rather than beams or columns, such as top-and-seat angles (Garlock et al. 2005;Christopoulos et al. 2002), reduced flanges (Chou et al. 2006), and friction devices (Rojas et al. 2005;Kim and Christopoulos 2008;Chou and Lai 2009). To improve the failure mode of the frame structure, a beam-column hybrid joint using flange cover connecting plates for energy-consuming and web connection plates for load transferring ) is proposed. The efficiency of the hybrid joint is investigated based on experimental research on PC joints and monolithic control connections. Song et al. (Song et al. 2014) used bolted web friction devices and a self-centering prestressed connection and adopted them in a moment-resisting frame to reduce residual drift under large drifts and improve the energy dissipation ability with friction damping. Experimental and numerical studies were performed to evaluate its effectiveness. The research results demonstrated that this bolted connection has capabilities of energy dissipation and self-centering comparable to welding connections and simultaneously avoids expensive field welding. Li et al. (2020) developed an innovative type of prefabricated beam-to-column joint and investigated the influence of the damper's geometric dimensions on the hysteretic performance of the prefabricated joint. Similarly, another type of damper was proposed by Qi et al. (2021) for prefabricated beam-to-column joint, and the seismic performance of the joint were studied as well as the design procedure was proposed.
Although previous studies have demonstrated that the behavior of prestressed joints or hybrid connections is comparable to that of monolithic connections (Eom et al. 2016;Mou et al. 2019), there are still issues, such as the difficulty in constructing connections between framing components, the feasibility of damage or failure mode control, and the potential for replacing the damaged energy dissipaters. Therefore, according to the present research, a new prefabricated beam-to-column joint is developed to mitigate the unsolved issue. Pseudo-static research was conducted on a novel prefabricated joint and the failure mode, load-bearing capacity, stiffness, ductility, energy dissipation, and earthquake-resilience were analyzed. In addition, the dissipaters of the steel hinges in the prefabricated joint were replaced and a second hysteretic test was conducted to study the restorable characteristics of the proposed prefabricated joints.

3 2 Mechanism of novel prefabricated joint
Based on the combination of load-carrying and energy-dissipation elements, an innovative type of prefabricated beam-to-column joint was developed, according to the schematic illustration in Fig. 1. The new beam-to-column joint is characterized by replaceable steel hinge and prefabricated steel tube confined joint core, providing advantages, such as complete assembly, damage control, and maintainability of the structure after an earthquake.

Replaceable energy-dissipating steel hinge
As described in Fig. 1a, b, replaceable steel hinges are set at the ends of the precast beam adjacent to the column. I-shaped steel is partly embedded at concrete of the precast beam end, and partly exposed for bolted connecting with steel hinge, the I-shaped steel is equal strength designed with the precast reinforced concrete beam, and the flanges are welded with the corresponding longitudinal reinforcements of beams to make sure reliable bending performance of the precast beam. In addition, I-shaped steel is welded at both sides of the joint core steel tube. Therefore, the precast beam can be bolted to the column via a steel hinge. The steel hinge is composed of two types of connected components, replaceable upper and lower energy dissipaters and the pin shaft connection, as shown in Fig. 1c. Lowyield point steel (LYP) is cut into a dog-bone shape to reduce geometry as described in steel design codes, such as FEMA 350-351 (2000a, b) andEC8 (2005), and an arc-shaped stiffener is welded under the LYP plate to avoid premature buckling. The LYP plate, arcshaped stiffener, and end plates formed an energy dissipater, which was used to transfer the bending moment between the beam and column, and also acted as a fuse to dissipate energy and concentrate plasticity under earthquake excitations. Weakened areas are easily formed due to the low yield strength and reduced section of the LYP plate; therefore, the damage position and damage degree can be controlled. Owing to the excellent seismic behavior of LYP steel , the LYP plate will concentrate most of the plasticity and damage, maintain the main structural members, such as beams and columns, and remain elastic. The pin shaft connection of the steel hinge contains left and right lugs, a (a) Precast  high-strength pin, and end-plates. The pin hinges the lugs and supplies the shear capacity in the steel hinge. The steel hinge adequately rotates around the pin, which can outward the plastic hinge away from the beam-column interface and shift into the weakened region of the steel hinge. This can improve the uncertainty of the location of plastic hinges in traditional reinforcement concrete frames owing to their limited rotation ability. Steel hinges are used to connect the precast beam and column with high-strength bolts, which are easy to manufacture, disassemble, and replace. Moreover, endplates of the energy dissipaters and endplates of lugs are independent, as shown in Fig. 1c, allowing for the replacement of the dissipaters instead of the entire steel hinge after an earthquake.

Prefabricated steel tube confined joint core
For the newly developed prefabricated beam-to-column joint, a steel tube was employed to confine the core area to realize the seismic design concept commonly referred to as a strong connection, as well as to connect the upper and lower precast column segments. As shown in Fig. 2, the proposed prefabricated steel tube confined joint core is comprises a steel tube, joint core concrete, internal diaphragms, connected components, and ducts for the insertion of reinforcements and grout pouring. The joint core concrete is confined by the steel tube, improving its shear resistance. Thus, shear failure in the joint core area can be avoided, and save assembling of stirrups. Two internal diaphragms are welded in the steel tube to transfer the horizontal force in the core area, which is equivalent to the continuous longitudinal reinforced steel bar of the beam passing through the core area utilized in the cast-in-place beam-to-column joint.
It should be noted that, the shear bearing capacity of the joint core increases with the increasing of steel tube thickness, this is because the increasing of shear carried by steel tube as well as improvement of confining efficiency on the core concrete. Therefore, it is necessary to determine the thickness of steel tube through design and calculation to meet the shear transfer requirements in the core area. In addition, using expansive concrete can improve the performance further.
The connection system between precast columns in this research is based on the use of precast columns, with sleeves encased on the lower end and longitudinal reinforcements protruding from the upper end. For the column-to-column connection assembly, as displayed in Fig. 3, the lower precast column is placed in position. Then, the prefabricated joint core is lifted, aligned with the lower column, and lowered to insert the protruding reinforcements of the lower column through the corrugated steel ducts of the joint core. High-strength grout is then poured to fill the ducts and create a layer at the interface between the lower column and the joint core. Finally, the upper precast column is lifted and positioned to insert the protruding reinforcements of the lower precast column into the sleeves of the upper precast column. The sleeves were grouted to achieve continuous longitudinal reinforcement of the column and the interface between the upper precast column and prefabricated joint core was simultaneously grouted. Both ends of the precast column are concrete rough surfaces to ensure the transfer of shear force at the interface.

Test specimens
Two quasi-static tests were conducted to investigate the hysteretic behavior of the novel prefabricated beam-to-column joint. First, a prefabricated joint was tested under cyclic loading until failure. According to this previous test, only four weakened steel hinge dissipaters were replaced and the second test was conducted to investigate the restorable functional characteristics of the proposed prefabricated joints. The specimens of the prefabricated joint under the two loadings were labeled as PJ-1 and PJ-2, respectively. Finally, a  Fig. 3 Connection system between precast columns cast-in-place monolithic joint specimen (labeled MJ) was also tested to estimate the seismic behavior of the proposed prefabricated joint.
This specimen adopts an interior beam-to-column joint idealized from the prototype structure; the beam has a span of 4000 mm, and the length of the column is 3040 mm. This prefabricated joint was composed of precast RC columns, precast RC beams, a steel tube confined joint core, and an energy-dissipating steel hinge with weakened flanges. The detailed configuration of the prefabricated joint and geometrical dimensions of the steel hinge and steel tube confined joint core are presented in Fig. 4. The section of the precast beams was 250 mm × 550 mm (width × height) and the section width of the square precast column was 400 mm. The precast column and beam were reinforced with 12 longitudinal bars 22 mm in diameter and eight bars that were 18 mm in diameter, respectively. In addition, 8 mm diameter hoops were provided and the stirrup spacing was 150 mm. For comparison, as shown in Fig. 5, the geometry and reinforcement of the monolithic joint were identical to those of the prefabricated joint.

Specimen construction
The fabrication procedure of this prefabricated joint began with the production of the upper column, lower column, beams, steel tube confined joint core, and energy-dissipating steel hinges, as shown in Fig. 6. The formwork for the precast columns and precast beams View of the prefabricated specimen series for the a reinforcement and grouted sleeve of the upper precast column, b lower precast column, c precast beams, d prefabricated steel tube confined joint core, and e energy-dissipating steel hinge was prepared using plywood. Then, the reinforcements and grouted sleeves of the columns were installed, and the upper and lower flanges of an I-shaped steel beam with a length of 300 mm were welded with the corresponding longitudinal reinforcements of beams. Subsequently, concrete with a strength grade of C55 was poured for columns and beams and cured for 14 days. For the precast beam, as shown in Fig. 6b, the depth of the embedded I-steel embedded in the concrete is 150 mm to make sure effective welded connection between longitudinal reinforcement and I-shaped steel, and the other depth of 150 mm I-steel was exposed for connecting with steel hinge.
A steel tube was welded using four plates measuring 400 × 500 mm 2 for confining the core area. Two internal diaphragms were welded to the steel tube at the position aligned with longitudinal reinforcements of the precast beam to ensure a reliable bending moment at the beam ends. After two I-shaped connecting beams were welded to the steel tube, concrete was cast into the joint core between two internal diaphragms.
When the concrete strength of the columns, beams, and joint core reached the hoisting allowable strength, the top surfaces of the lower column and bottom surface of the upper column were roughened, as shown in Fig. 6b, to transfer the shear force between the interfaces of the upper column, joint core, and lower column. Then, the prefabricated components were assembled according to the following steps: (1) presented in Fig. 7a, hoist the prefabricated steel tube confined joint core to make the protruding longitudinal reinforcements in the lower precast column through the reserved hole formed by the a b c f d e Fig. 7 Assembly process of the prefabricated joint: a assemble the prefabricated steel tube confined joint core and lower precast column, b duct grouting of the joint core, c install the upper precast column, d grout the sleeves of the upper precast column, e install the energy-dissipating steel hinge, and f connect the precast beams and columns steel syphon bellows in the joint core. Then, the prefabricated joint core was placed on the top of the lower column.
(2) High-performance cement paste was grouted to the duct between the bellows and reinforcements, as shown in Fig. 7b, to ensure the stability of the reinforcements and connect the top surface and joint core. (3) The upper precast column was installed at the top of the joint core (Fig. 7c). (4) The sleeves of the upper precast column were grouted to connect the longitudinal reinforcements of the lower and upper columns and the bottom surface of the upper column and joint core. As shown in Fig. 7d, a high-performance cement paste was grouted from the bottom and flowed out of the top to ensure that the cavity of the sleeve was filled with paste. (5) The steel hinge was connected to the precast beam using high-strength bolts (Fig. 7e). (6) Two precast beams and steel hinges were connected to each side of the joint core (Fig. 7f).

Material properties
LYP was applied to a replaceable energy dissipater. The other steel components, including the lugs, I-shape steel, steel tube, and internal diaphragms of the beam and column connector, were made of Q355. HRB400 was used as reinforcements in the precast beams and columns. The material properties of the steel plates and bars were obtained by tensile tests that were based on GB/T228.1-2010 (2010). The determined yield strength (f y ), yield strain (ε y ), ultimate strength (f u ), elastic modulus (E s ), and Poisson's ratio (μ s ) values are listed in Table 1. High-strength bolts (Class 10.9) were employed for connections between the prefabricated joint core, steel energy-dissipating hinge, and precast beam. All specimens were cast using concrete with identical mix proportions. The cubic compressive strength (f cu ) and elastic modulus (E c ) of the concrete were measured by a cube with dimensions of 150 mm × 150 mm × 150 mm and a prism with dimensions of 150 mm × 150 mm × 300 mm, respectively. At the testing days, f cu and E c of the concrete were 56 MPa and 35,400 MPa, respectively.

Test setup and loading scheme
Hinge support boundary conditions are set for the column bottom and beam ends to simulate infection points in a frame. To reflect the actual working conditions and account for the second-order influence in the frame, a constant axial load was applied to the free top of the column during the lateral cyclic loading for all specimens. First, an axial compression of 1600 kN (an axial load ratio of approximately 0.3) was applied and maintained constant using a jack with a capability of 2000 kN. This jack was connected to a rigid reaction beam by a rolling support to ensure that the compression on the column was concentric. The cyclic load was provided by a hydraulic actuator with an ability of ± 500 kN. The test setup is shown in Fig. 8.
The tests were conducted under force and displacement control following ACT-24 (1992). In the force control phase, load levels of 0.25 P uc , 0.5 P uc , and 0.7 P uc were selected, where P uc denotes the predicted lateral ultimate strength by finite element analysis. When the yield strain was observed, the displacement control was then obtained at increments of Δ y , 1.5Δ y , 2Δ y , 3Δ y …, where Δ y represents the yielding displacement. The cyclic loading scheme of the lateral load is shown in Fig. 9. Two cycles and three cycles at each level were conducted for load and displacement control phase, respectively.
During the test, the lateral load-displacement (P-Δ) hysteretic curves were automatically recorded by the loading system. Four extensometers (Nos. 1-4 in Fig. 10a) were set at the upper and lower flanges of the steel hinge to indirectly measure the rotation of the steel hinges. The shear drift of the joint core was measured using two extensometers (No. 5 and 6 in Fig. 10a) installed along the diagonal lines. Displacement transducers 1, 2, and 3, 4 were set at the bottom of the upper precast column and the top of the lower precast column, respectively, to evaluate the rotations of the columns. The ranges were ± 200 mm for the displacement transducers and ± 50 mm for the extensometers.
For the prefabricated specimens, as presented in Fig. 10a, b total of 13 strain gauges and two strain rosettes were used to obtain the strains in energy dissipaters and the lugs of  Fig. 10c. For the monolithic specimen, strain gauges were set at a distance of 100 mm from the joint core for the longitudinal reinforcement in the columns and beams.

Initial loading test for the prefabricated specimen (PJ-1)
During the first load step of the prefabricated specimen, specimen PJ-1 experienced an approximate elastic deformation at the force control stage. Yielding occurred on the flange near the maximum weakened position of the replaceable energy dissipaters in steel hinge, at a load of 100 kN (approximately 0.5P uc ), and the corresponding displacement of column top was 9.6 mm. After that, displacement control loading was applied based on Δ y = 10 mm. When the displacement reached 1.5Δ y , four longitudinal shear sliding cracks appeared in the precast beam along the interface between the flanges of the embedded steel and the nearby concrete. The length of the cracks was similar to that of the embedded part of the I-shaped steel. A flexural crack through the bottom of the left beam was observed at the same time, which was approximately 150 mm away from the concrete edge of the precast beam. Under reversed loading and the second cycle of 1.5Δ y , symmetric flexural cracks at the top and bottom of the beams adjacent to the column were observed.
New flexural cracks were observed approximately 150 mm from the initial flexural crack as the displacement increased to 2Δ y and further outward flexural cracks developed at 3Δ y . In addition, the cracks developed from the bottom and top to the sides of the beam and inclined cracks subsequently formed. Slight cracks appeared in the lower precast column near the joint core at a displacement of 4Δ y ; however, no cracks appeared in the upper precast column owing to the reinforcement of the grouted sleeve during the entire loading process.
Slight local buckling on the energy dissipation hinge in the compression flange was observed at an incremental displacement of 5Δ y . Subsequently, the precast beam and  5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Fig.10 Arrangement of the instrumentation the load-bearing capacity of the specimen deteriorated owing to further loading. A slight crack was observed in the upper tension flange of the right beam in the first cycle of 10Δ y , and the fracture developed at the second cycle of 10Δ y . Then, the specimen failed. The final failure appearance of specimen PJ-1 is shown in Fig. 11. During the loading process, the strains for the longitudinal reinforcements in the precast beam and column were in the elastic range.

Second loading test for the prefabricated specimen (PJ-2)
Based on the previous test, four damaged dissipaters of the steel hinges were replaced, forming specimen PJ-2 and the second loading test for the prefabricated specimen was (a) Finial  conducted under the same scheme as the first loading test. The replacement process is shown in Fig. 12. During the initial stage of loading, there was no new crack development and specimen PJ-2 was in the elastic stage. New cracks at the end of the shear cracks on the beam sides were first observed at 2Δ y , following which new crack propagation was rare. The crack development in the precast beams of specimen PJ-2 is marked by the red line, as shown in Fig. 13a. Similar to the first test, slight local buckling on the compression dissipaters of the steel hinge was observed at 5Δ y . Subsequently, the precast beam and column developed no   further cracks when the plastic deformations were centralized in the dissipaters. In addition, the local buckling of the steel hinge became more obvious upon further loading. Specimen PJ-2 reached its peak load at 7Δ y . The values were 225.70 kN and − 233.41 kN for the push and pull directions, respectively, which is similar to the load-bearing capacity of specimen PJ-1. The failure area of specimen PJ-2 was in the energy dissipaters of the steel hinge as well as that of specimen PJ-1. At the third cycle of 10Δ y , the lower tension dissipater of the right steel hinge fractured and the test was terminated. The deformation development of the steel hinges is shown in Fig. 13b, c. During the two tests for the prefabricated specimens, no other visible buckling or changing phenomenon was observed. The confined steel tube was removed after testing specimen PJ-2 for direct observation of the core concrete. As shown in Fig. 14, only a slight crack was found in the core area, indicating that the novel prefabricated joint met the seismic design concept commonly referred to as "strong connection".

Monolithic specimen (MJ)
In the monolithic specimen, the first cracks occurred near the beam and column interface at a load of 80 kN (0.5P uc ) and some flexural cracks were located at the bottom and top of the beams. The corresponding displacement was 7.6 mm. Then, displacement control loading was applied based on Δ y = 7 mm.
During the displacement control phase, the flexural cracks in the beams developed from the position close to the core area to the end; then, the inclined crack developed into a through-crack at a displacement of 1.5Δ y . When the displacement attained 2Δ y , a horizontal flexural crack was found in the column at a position close to the joint core. Upon further loading, no cracks developed in the beam. The occurrence and development of cracks were concentrated in the joint core area. Furthermore, oblique shear cracks were formed at an angle of approximately 38-45° in the horizontal direction in the joint core area at 3Δ y and specimen MJ attained maximum strength. Some concrete gradually began to crush and fall off within the four joint core area corners at 5Δ y and this became more serious and extended to the upper column at incremental displacements of 5Δ y -7Δ y , which caused exposure of the reinforcements. After the peak load, as shown in Fig. 15, at 10Δ y , the lateral load reduced to 85% of the peak load, the monolithic joint failed by shearing at the joint core, and the test was terminated.
During testing, the relationship between the precast columns and precast beams was reliable and the prefabricated specimens and monolithic specimen exhibited good overall mechanical performance. Compared with the monolithic joint, the plastic hinge was forced outward from the column face, and damage was controllable in the prefabricated joint owing to the steel hinge. The replaceable energy dissipater made of low-yield-point steel entered the plastic stage first and dissipated the energy. Therefore, the crack development and damage of the precast RC beams, precast RC columns, and joint core area were efficiently controlled. There were no significant differences in the failure mode and loadbearing capacity of the prefabricated specimens under the first and second loading times (specimens PJ-1 and PJ-2). Thus, the hysteretic performance of the proposed novel prefabricated joint can be restored by replacing only the energy dissipater.

Lateral load-displacement (P-Δ) hysteretic curves
The P-Δ hysteretic curves of all specimens are shown in Fig. 16. The hysteretic curves of all specimens approached straight lines and the specimens experienced elastic deformation during the initial loading stage. For prefabricated specimens PJ-1 and PJ-2, the slope of the hysteretic curves decreased slightly as energy dissipater in the steel hinge yielded while the strength increased consistently owing to the stress strengthening of the dissipater, and residual displacement was observed when unloading. The P-Δ hysteretic curves of the prefabricated joints are plump in shape and have no obvious pinching effect, as shown in Fig. 16a, indicating excellent energy dissipation capacity. When the lateral displacement reached 80 mm-i.e., that the drift ratio reached 2.5%-the load capacity decreased owing to the local buckling of the dissipaters. When the prefabricated specimens failed, the drift ratio reached 3.33%. Moreover, the P-Δ hysteretic curves of PJ-1 and PJ-2 coincided, indicating that the seismic behavior of the novel prefabricated joint can be restored by replacing only the energy dissipater.
The P-Δ hysteretic curves of the monolithic joint exhibited significant pinching in the middle, as shown in Fig. 16b, because specimen MJ failed in a shear-dominant mode in the Fig.15 Failure mode of specimen MJ joint core area. After the peak load, the strength decreased gradually owing to the concrete crushing and spalling. A comparison of all the P-Δ hysteretic curves revealed that the seismic performance, such as the load-bearing capacity, energy consumption, and ductility, of the designed novel prefabricated joint was better than that of the monolithic joint. Figure 17 shows the P-Δ envelope curves of all specimens. The yield load (P y ) and yield displacement (Δ y ) were determined by the geometrography method, as presented in Fig. 18 in (Han et al. 2009). The load and displacement at the peak point of the P-Δ envelope curve were labeled as P max and Δ max , respectively, and the failure load (P u ) and corresponding displacement (Δ u ) were determined when the load decreased to 85% of P max . The displacement and load at the points of yield, peak, and failure are listed in Table 2 for all specimens.

P-Δ envelope curves
In Fig. 17, the initial stiffness of specimen PJ-2 was slightly lower than that of PJ-1; this is because of the existence of cracks in the PJ-2 specimen precast beams and columns from the beginning, as only the dissipaters were replaced after the first load. Thus, Δ y of specimen PJ-2 was slightly higher than that of specimen PJ-1. After the yield point, the difference between the P and Δ envelope curves of specimens PJ-1 and PJ-2 can be considered negligible. The load-bearing, energy dissipation, and deformation capacities can be fully restored by replacing the dissipaters of the steel hinges. As can be seen from Table 2, the load-bearing capacities of prefabricated specimens PJ-1 and PJ-2 were 48% and 49% higher than that of the monolithic specimen MJ, respectively. This is because specimen MJ failed in a shear-dominant pattern in the joint core area while the core area was enhanced by a steel tube for the prefabricated specimens. Moreover, plastic hinges were bound to occur at the beam ends, which protected the precast beam and column from damage.

Stiffness degradation
The average secant stiffness is used as the stiffness of the specimens under different loading levels, and the relative stiffness of the i-th average secant stiffness K i is defined as (): (1) where P i and Δ i are the peak load and lateral displacement, respectively, under the i-th Δ/ Δ y and ' + ' and '−' represent the positive and negative directions, respectively. Figure 19 compares the K i -Δ/Δ y curves of all specimens. For all specimens, the stiffness decreases as the lateral displacement increases. The stiffness under the first loading cycle for the cast-in-place monolithic joint is approximately 23.1% higher than that of prefabricated specimen PJ-1, demonstrating that the steel hinge connection results in a slight deterioration of the integrity of the beam. However, specimen MJ showed a more severe rate of stiffness degradation. Therefore, the stiffnesses of specimens MJ and PJ-1 were similar at 1.5Δ y . Subsequently, the stiffness of specimen PJ-1 was higher. This is because the cracks occurred and extended continuously during the entire loading process for specimen MJ. In the prefabricated joint, after a certain displacement increment, the damage was concentrated in the steel hinge dispersers, which led to no further cracks in the precast column and beam.
From Fig. 19, the initial stiffness of the restored prefabricated specimen PJ-2 is 30% lower than that of PJ-1, as mentioned in Sect. 4.3; this is due to the existence of cracks in the precast beams before loading. The two prefabricated specimens, PJ-1 and PJ-2, experienced similar stiffness degradations because the degradation was approximately due to the yield of the dissipaters. Figure 20 shows the strength degradation of λ 2 and λ 3 of the specimens as a function of the lateral displacement, where λ 2 and λ 3 are the strength degradation coefficients of the second and third cycles, respectively, at the same loading level. For prefabricated specimens PJ-1 and PJ-2, λ 2 and λ 3 were stable at approximately 1.0, with a small jitter before fracturing of the dissipater, indicating that the novel prefabricated joint had an excellent load-bearing capacity under cyclic loading. The strength degradation of the monolithic joint occurred earlier than that of the prefabricated joint and the rate of strength degradation was faster; λ 2 and λ 3 were approximately 0.85 and 0.90, respectively, for specimen MJ. In summary, the strength degradation of MJ is significant.

Ductility and energy dissipation
Following the definition by Han et al. (Han et al. 2009), the displacement ductility coefficient (µ) of all specimens is determined by μ = Δ u /Δ y and listed in Table 2. From Table 2, the average of the active and passive failure displacements for specimens PJ-1 and PJ-2 are 96.51 mm and 100.82 mm, respectively, and the corresponding drift ratios are 3.22% and 3.36%, respectively, indicating the excellent deformation ability of the prefabricated joint. The displacement ductility coefficient of PJ-2 decreased by approximately 31% compared to that of specimen PJ-1, as previously mentioned. This is because the Δ y of PJ-2 was slightly higher than that of PJ-1. For specimen MJ, which is cast-in-place, Δ y and Δ u are 13.85 mm and 78.72 mm, respectively; i.e., significantly lower than those of specimen PJ-1 and PJ-2. In addition, the μ value of specimen PJ-1 was higher than that of specimen MJ. Thus, the deformation capacity of the prefabricated joint was improved. The cumulative hysteretic energy (E p ), calculated based on the area enclosed by the hysteretic hoops from the P-Δ hysteretic curves, and the equivalent hysteretic damping coefficient ( eq ), determined according to Fig. 21 shown in (), was employed to estimate the energy consumption capacity of the joints. The equivalent hysteretic damping coefficient can be expressed as: where S ABC+CDA is the hysteresis loop area, and S OBF+ODE is the area of triangle OBF and ODE. Figures 22 and 23 show the E p -Δ curves and eq -Δ curves for each specimen. The calculated values of the cumulative hysteretic energy and equivalent hysteretic damping coefficient once the lateral load reduced to 85% of the ultimate strength-i.e., E p,u and eq,u , respectively-are listed in Table 3. Evidently, E p increases with increasing displacement. In contrast to monolithic joint MJ, energy dissipation capacity of prefabricated joints PJ-1 and PJ-2 clearly increased owing to the superior plastic energy dissipation capacity of the steel hinge. Here, E p,u of joints PJ-1 and PJ-2 were higher than that of joint MJ by a factor  of approximately 2 and 1.8, respectively. The equivalent hysteretic damping coefficient of the joint MJ developed rapidly during the initial loading stage owing to the serious cracking of the joint core area. However, after the displacement reached 30 mm, the eq of the prefabricated joints were higher than that of monolithic joint. The values of eq,u for joint PJ-1 and PJ-2 increased by 138% and 100%, respectively, compared to joint MJ. From the above, it can be seen that the prefabricated joint provides excellent energy consumption ability through the use of steel hinges. In addition, as shown in Figs. 22 and 23, the agreement between the cumulative hysteretic energy curves of joint PJ-1 and PJ-2, as well as the similar equivalent hysteretic damping coefficient curves between joint PJ-1 and PJ-2, reveals that the energy dissipation capacity can be recovered for the prefabricated joint by replacing the dissipater.

Shear deformation of joint core
As mentioned in Sect. 3.4, the shear drift of the joint core was gauged by extensometers set along the diagonals. As shown in Fig. 24, the shear drift angle (γ) of the joint core can be calculated as: where α 1 and α 2 are the shear drift angles along the height and width direction of the joint core, respectively; h and D are the height and width of the joint core, respectively; δ 1 and δ 2 are the deformations along the diagonals. As shown in Fig. 25, the lateral load-shear drift angle (P-γ) hysteretic curves of all specimens almost linearly cycled and no obvious residual deformation was observed during the initial loading period. For the prefabricated joints, the development of the shear drift angle under varied loading was between − 0.0005 rad and 0.0005 rad, indicating that the joint core area was in the stage of elastic deformation during the entire process, which will also be verified by the measured main strain of the confined steel tube at the joint core in Sect. 4.8. Moreover, the shapes of the P-γ curves for joints PJ-1 and PJ-2 are similar, demonstrating that the damage was controlled to take place at the dissipaters of the steel hinges, which could protect the joint core. For the cast-in-place monolithic joint MJ, the cracks at the joint core developed rapidly, leading to a rapid increase in residual shear deformation. After the main diagonal cracks were formed, the shear drift angle reached 0.006 rad. When the concrete was spalled and crushed, the joint failed due to the joint core damage.

Strain distribution and developement
The lateral load-strain (P-ε) envelope curves for specimens were obtained by sequentially connecting the extreme point of each loading level on the corresponding P-ε hysteretic curves. The longitudinal strains of the longitudinal reinforcements in the beam and column (ε bar ), the longitudinal strains of dissipaters in the steel hinge (ε hinge ), and the main strains of the confined tube in the joint core (ε tube ) are presented in Figs. 26, 27, and 28, respectiv ely. As shown in Fig. 26, the longitudinal reinforcement experienced uniform strain development for the prefabricated joints PJ-1 and PJ-2. The longitudinal strains did not exceed 2000 µε during the entire testing, that were less than the yield strains, as listed in Table 1, were 2446 µε and 2527 µε for the beam and column reinforcements, respectively. In contrast, as shown in Fig. 27, the strains of the dissipaters (ε hinge ) developed rapidly. After the peak, the strains of the dissipater increased remarkably, the maximum value exceeding 0.02, leading to the fracture of the dissipater. These results illustrate that the damage was reinforcements of the beam yielded before the peak point. With crack development, the strain developed rapidly. It can be seen from Fig. 27 that the strain distribution and development on the dissipater of the steel hinge for prefabricated joints PJ-1 and PJ-2 were similar. For each specimen, only the strains of the upper dissipater of the right beam were presented owing to symmetry. The strain gauges were numbered 1-7 from the end connected to the joint core to the end connected to the precast beam, as shown in Fig. 10b. Strain gauge No. 4 yielded first and the strain was maximum during the entire process as it was installed on the weakest section. With the increase in lateral displacement, the plasticity extended from the center to the ends of the dissipater, which led to excellent energy dissipation. Stress strengthening allowed the joint to maintain load-bearing. In addition, strain gauges No. 1 and No. 7 did not yield, limiting stress in the less ductile region near the face of the column and the plastic development was concentrated in the weakened area of the dissipater. Consequently, with the concept of the reduced section of the dissipater, the failure mode and damage position of the proposed prefabricated joint can be controlled and the plastic hinge can be outward from the beam-to-column interface. Although the energy dissipating elements are symmetrically weakened, the strain measured by strain gauges No. 2 and No. 3 are larger than those of strain gauges No. 5 and No. 6, as the bending moment close to the joint core is higher. Figure 28 shows the principal stress and direction of a typical measuring point in the steel tube at the joint core area for the prefabricated joints. The principal stresses calculated by the measuring stains are less than the yield stress of the steel tube, indicating that the core area of the prefabricated joint is confined and effectively protected by the steel tube. Thus, the seismic design concept is commonly referred to as a strong connection. The direction of the principal stress in the steel tube, as shown in Fig. 28b, d, demonstrates that the steel tube was mainly subjected to a shear force.

Moment and rotation curves and deformation capacity
As shown in Fig. 29, the moment-resistant (M) and the rotation (φ) of the energy-dissipation steel hinge can be calculated as: where P is the lateral load applied at the column top, H is the distance from the loading point to the center of the joint core, l is the distance between the steel hinge and the center of the joint core, and L is the distance between the pin support of the beam end and the center of the joint core; Δ c and Δ t are the axial deformations of the compressed and tensioned energy dissipaters, respectively; h is the height of the steel hinge section, which is defined as the distance between the centroids of two energy dissipaters.
The M-φ curves are listed in Fig. 30 for all of the specimens. As shown in Fig. 30a, b, the hysteretic curves of the energy-dissipating steel hinge of the prefabricated joint are plump in shape and show sufficient flexibility. Moreover, the M-φ curves of the left and right steel hinge were almost the same for the prefabricated joint (PJ-1), as well as for PJ-2. This demonstrates that the steel hinge can rotate around the pin axis. The bending moment and rotation at the points of yield, limit, and failure for the steel hinge of the prefabricated joint and plastic hinge for the monolithic joint are listed in Table 4. The average values were calculated as the M-φ curves were similar for the left and right hinge. Again, the characteristic values for the steel hinge in the prefabricated joint under the two tests are close to each other, indicating that the function of the steel hinge can be restored. Furthermore, the failure rotations were 0.048 and − 0.055 for prefabricated specimen PJ-1 and 0.051 and − 0.052 for PJ-2; the average value reached 0.0515, which is much higher than that of the monolithic specimen MJ (the average failure rotation was 0.26) and demonstrated excellent rotation capacity for the steel hinge in the prefabricated joint.

Energy consumption ration
The calculated values of the cumulative hysteretic energy (E p,u ) were calculated for the hinges of the beam-to-column joint once the lateral load was reduced to 85% of the ultimate strength and listed in Table 5. For the prefabricated beam-to-column joint, the energy dissipated by the replaceable steel hinges accounted for 67.8% during the first loading test, indicating that the energy absorption was concentrated on the steel hinges. In addition, for the second loading test, there was no obvious development of cracks, which led to slightly lower energy dissipation and the energy consumption ratio increased to 77.0%. The energy consumption ratio of plastic hinges for a monolithic joint is 45.8%.

Conclusions
This experimental study investigated the mechanical behavior of an innovative type of prefabricated beam-to-column joint subjected to cyclic loading. Based on this study, the conclusions are drawn as follows: The connection between the precast columns and precast beams was reliable and the new prefabricated beam-to-column joint exhibited good overall mechanical performance. Controllable plastic hinges were formed at the ends of the precast beams  because of the prefabricated steel tube confined joint core and energy-dissipating steel hinge, which meet the seismic design concept commonly referred to as a strong connection. The failure of the specimens was concentrated on the flange of the steel hinge, primarily from the buckling of the compressed flange or from fractures forming on the tensioned flange. Only minor flexural cracks occurred near the beam and no other visible damage was observed in the prefabricated beam-to-column joints. Even under repeated tests, the failure mode and damage degree of the prefabricated joint can be controllable. The novel prefabricated beam-to-column joints exhibited excellent hysteretic behavior with generally plump lateral load-displacement hysteretic curves; in contrast, the hysteretic curve of the cast-in-place joint has an obvious pinching effect. Compared with the monolithic joint, the load-bearing capacities of the prefabricated joint are approximately 50% higher and the deformation capacity and ductility are improved. The energy absorption capacity, in terms of the equivalent hysteretic damping coefficient, was approximately two times higher than that of the monolithic joint when the lateral load decreased to 85% of the peak load.
The initial stiffness of the monolithic joint (MJ) was slightly higher than that of the prefabricated specimens; however, the stiffness degradation of specimen MJ was significant. Thus, the stiffness of the monolithic and prefabricated joints was similar at 1.5 Δ y . The prefabricated joints exhibited no obvious strength degradation. The strength degradation coefficients of the monolithic joint were approximately 0.85-0.90.
In prefabricated specimen PJ-2, which only replaced the damaged dissipaters in the steel hinges and was tested under repeated loading, the failure mode and P-Δ hysteretic curve were similar to those of the basic test, specimen PJ-1. However, the initial stiffness was slightly lower, demonstrating that the prefabricated joint sustained a similar hysteretic performance after restoration and the function of the prefabricated joints can be restored after earthquake damage.
The M-φ curves of the steel hinge in prefabricated joint are also in pump shape and exhibited sufficient flexibility. The steel hinge can rotate around the pin axis, and the rotation at the failure point reached 0.0515 rad. In the prefabricated beam-to-column joints, the energy dissipated by the replaceable steel hinges accounted for 67.8% and 77.0%, respectively, indicating that the energy absorption was concentrated on the steel hinges.
Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request.