Keywords

1 Introduction

The prefabricated concrete frame structure has advantages such as low environmental impact, high component quality, short construction period, and low labor requirements, making it the direction of future development in the construction industry. Previous seismic damage investigations have shown that some prefabricated concrete structures suffered severe damage during earthquakes [1, 2] mainly due to the failure of the connection joints between prefabricated components, leading to the collapse of the entire structure. Therefore, the beam-column connection remains a key research subject.

Dry assembly beam-column connections, during installation, do not require on-site wet work, resulting in a higher assembly rate and better alignment with the current industrialized production requirements. Several researchers have conducted extensive experimental and finite element simulation studies to enhance the seismic performance of prefabricated frame beam-column connections. Englekirk et al. [3, 4] and Li Xiangmin et al. [5] proposed a ductile connection system to improve the seismic performance of structures. This system involves embedding ductile links within prefabricated columns, connecting them to prefabricated beams using bolts. During seismic events, the ductile links undergo plastic deformation, dissipating energy and reducing seismic forces, thus preventing damage to other structural components. However, this approach involves the casting of ductile connectors within the column, and after seismic action, the core region of the connection exhibits shear failure characteristics, which can hinder post-earthquake connection repair. Zhao Bin et al. [6] suggested embedding screws at the column node and casting concrete beams with I-shaped steel joints as an integral component. Subsequently, they used screws and rubber bearings for assembly, creating a semi-rigid connection. Experimental results showed that while bolted connections exhibited improved load-carrying capacity compared to cast-in-place connections, their energy dissipation performance was poorer, and significant cracks developed in the concrete core region in the later stages of loading. In summary, although certain existing structural approaches have enhanced the seismic performance of prefabricated beam-column connections, they still suffer from drawbacks such as dispersed damage under rare seismic events, uncontrollable locations for plastic deformation development, and the risk of damage to the core region of column ends and critical connection points. Once structural damage occurs, it becomes challenging to achieve repairs.

In recent years, seismic disaster investigations have revealed [7] that while the overall collapse of buildings due to seismic forces has been effectively controlled, the economic losses and societal impact caused by earthquakes are still significant. The severe damage sustained by buildings after an earthquake makes the repair process challenging, leading to prolonged interruptions in the normal functioning of buildings and hindering daily life and production. To address the need for structures that do not require extensive repair or maintenance after earthquakes, or that can quickly recover their normal functionality with minimal repairs, the concept of resilient structures has emerged [8]. In pursuit of this concept, scholars have proposed structural systems where damage to connecting nodes is controllable, and components are replaceable. This is achieved by placing replaceable connecting components at plastic hinge locations, and inducing damage concentration in these replaceable components. The goal is to achieve either no damage or minimal damage in concrete beam-column structures. After replacement of the damaged components, the structure can rapidly restore its functional use.

Ye Jianfeng et al. [9] proposed a beam-column connection node based on replaceable energy-dissipating hinges. After an earthquake, the seismic performance remains essentially unchanged by replacing the metal damper components, achieving functional recovery. Huang Wei et al. [10] introduced a beam-column node connected by multi-slot energy-dissipating devices. Experimental results indicate that this node can achieve damage concentration, facilitate rapid seismic damage recovery, and exhibit comparable load-bearing capacity and stiffness to conventionally cast nodes. Additionally, it possesses higher ductility and energy-dissipating capability.

Ma Zhehao et al. [11] proposed an assembly-type RC beam-column node with artificial energy-dissipating plastic hinges. By utilizing plastic perforated energy-dissipating steel plates for connection, the structure exhibits higher load-bearing capacity and lower damage, effectively enhancing seismic resistance. Song et al. [12] introduced a self-centering concrete beam-column connection with a web friction device. Research indicates that this structure functions to reduce residual deformations and minimize structural damage. Aninthaneni et al. [13] investigated the hysteric performance of prefabricated frame sub-assemblies with end plate connections. The study results demonstrate that this connection can be used for high-rigidity anti-bending connections and developed a detachable sub-component energy-dissipating system. Wang Meng et al. [14] proposed the use of low-yield-point steel as connecting components, enabling ductile energy dissipation resembling a fuse function. This alteration in the failure mode of the node helps prevent the main structure from entering the plastic phase prematurely.

In summary, with the increasing demands on building structures, research on prefabricated nodes has evolved from seismic resistance and isolation to the current development of recoverable functional structures. Recoverable functional structures mainly encompass prestressed systems and replaceable connection component systems. Recent studies have found that prestressed steel bars in prestressed systems suffer from severe elastic losses and complex construction procedures [15]. On the other hand, in replaceable connection component systems, small-scale BRB structures are commonly used as replacement devices [9, 16]. However, this system still faces challenges such as high component replacement costs and intricate manufacturing processes.

In light of the problems encountered in the development of prefabricated concrete structures today, this paper proposes a steel node with replaceable friction energy dissipation capability. Different from conventional prefabricated nodes, this node allows artificial control of strength by adjusting pre-tension at the beam end. Moreover, a friction energy dissipation damper is installed at the connection location. During seismic events, the damper's contact interface begins to slide, exerting energy dissipation effects to safeguard the entire structure from damage. Additionally, it helps confine damage locations within the replaceable steel construction area. This node exhibits excellent energy dissipation capabilities and controllable plastic development, making it of significant research value and importance.

2 Node Structure and Working Principle

The key components of the replaceable friction energy consuming steel joints for assembled RC beams and columns (hereinafter referred to as the steel node) are the friction energy dissipation system and the U-shaped anti-shear steel plate system. The force transmission path of this node is clear, with the friction energy dissipation system primarily bearing the bending moment at the beam end, while the U-shaped anti-shear steel plate system takes on the entire shear force. As illustrated in Fig. 1, during installation, the steel node components are first connected to the precast beams with bolt holes using high-strength bolt assemblies. Subsequently, the precast beam with ductile joints is hoisted to the column end for alignment, and the ductile joints are connected to the core area of the joints through high-strength bolt pairs and T-section steel with flange extension as backing plates to complete the assembly work.

Fig. 1.
figure 1

Replaceable friction energy dissipation steel joint structure of assembled RC beam-column.

Full size image

3 Seismic Performance Analysis of the Replaceable Friction Energy Dissipation Steel Node

3.1 Node Parameter Design

To exploit the damage control and concentrated energy dissipation characteristics of the steel node under seismic loads, a design load capacity coefficient μ between the steel node and the concrete beam has been defined, as shown in Eq. (1). When μ is greater than 1, the reinforcement in the beam-column tends to yield earlier than the steel node, leading to damage in the primary structural elements and a reduction in the recoverable functionality of the node. When μ is less than 1, the steel node can yield and dissipate energy earlier under seismic actions, playing a protective role for non-energy dissipating components such as the reinforcement in the beam-column.

Fig. 2.
figure 2

Force mechanism of beam-column joint.

Full size image
$$ \mu = \frac{{M_{f} }}{{M_{b,r} }} $$
(1)
$$ M_{b,r} = n \cdot f_{y} \cdot A_{s} \cdot h_{s} $$
(2)

In the equation, Mf represents the initial slip moment of the friction damper; Mb,r is the elastic resistance moment of longitudinal steel bars in the concrete beam; n is the number of tension or compression zone steel bars; fy is the yield strength of longitudinal steel bars in the beam; As is the cross-sectional area of a single steel bar; hs is the distance between the upper and lower longitudinal steel bar force couples.

The initial slip moment Mf of the friction damper is obtained by the equivalent force couple formed by the tension and compression of the upper and lower friction dampers, as shown in the force analysis in Fig. 2. Additionally, at this node, the U-shaped shear-resistant steel plate bears the majority of the shear force at the beam end. Therefore, the friction damper can be approximately considered to only bear axial tensile and compressive forces, that is:

$$ M_{f} = F_{A} \cdot h_{A} $$
(3)

In the equation, FA represents the initial sliding force of the friction damper, and hA is the distance between the upper and lower force couples of the friction damper.

For structures with post-earthquake recovery functionality, their basic working principle under seismic actions involves the concentrated energy dissipation of connectors, while main components such as beams and columns remain essentially in an elastic state. After an earthquake, structural functionality can be restored by simply replacing damaged connectors. To validate the ability of nodes to dissipate concentrated energy and achieve post-earthquake replaceability, six sets of node specimens were designed by adjusting the preload force of high-strength bolts to change the load capacity coefficient. The model numbers are XJ, PT, PT-1 to PT-7, as shown in Table 1. Single-parameter analysis was conducted using the finite element method. Through computation and comparison, the concentrated damage mechanism of steel nodes was further analyzed, and the rapid recovery capability of their post-earthquake functionality was assessed.

Table 1. Model number and node parameters.
Full size table

It is noteworthy that in the previous context, the bearing capacity factor μ was defined by the ratio of the moment resistance at the node at the initial sliding moment of the friction damper to the elastic moment resistance of the longitudinal steel bars in the concrete beam. However, as displacement is gradually applied, the U-shaped steel plate shear-resistant system also begins to contribute to the moment resistance of the node, and these contributions cannot be ignored. This results in the actual moment resistance Mr being greater than the initial sliding moment of the friction damper Mf in Eq. (3) thereby causing the bearing capacity factor μ to increase. To account for the impact of this situation, the concept of “enhancement ratio-μl,r” is introduced, namely:

$$ \mu_{l,r} = \frac{{M_{r} - M_{f} }}{{M_{b,r} }} $$
(4)

As only the U-shaped steel plate provides additional moment resistance after the sliding of the friction damper at this node, Eq. (4) can also be expressed as:

$$ \mu_{l,r} = \frac{{M_{U} }}{{M_{b,r} }} $$
(5)

In the equation, MU represents the moment resistance provided by the U-shaped steel plate after the sliding of the friction damper.

The beam-column section dimensions were designed according to the requirements of reference [11], with PT serving as the standard model. The specific parameters for PT and XJ can be found in Fig. 3. PT-1 to PT-7 only differ in the preload force applied to the high-strength bolts, while the other characteristics of the models remain consistent with PT.

Fig. 3.
figure 3

Reinforcement size and component details of beam-column joints

Full size image

In the connection components, the material constitutive model for the 10.9-grade M22 high-strength bolts adopts a two-linear model, with a yield strength (fy) of 995 MPa and an ultimate strength (fu) of 1160 MPa. The bolt hole sizes, positions, and spacing comply with the design specifications for high-strength bolts. T-shaped connectors and U-shaped shear-resistant steel plates are made of Q235 steel (yield strength of 235 MPa, and post-yield stiffness is 0.01 times the initial stiffness). The remaining steel components are constructed from Q345 steel (yield strength of 345 MPa). The elastic modulus (E) for all steel materials is 21000 MPa, following the kinematic hardening criterion. The concrete constitutive model employs the C30 plastic damage model, and specific parameter relationships are referenced from experimental results in literature [11].

Seismic data is referenced to a specific engineering project, with a seismic design intensity of 7° in the strong motion zone. The design basic seismic acceleration is 0.10 g, and the foundation is of Type II soil without liquefiable layers. The design seismic grouping is the first group with a damping ratio of 0.05. To facilitate quasi-static loading, the model adopts the following loading Protocol: Prior to component yielding, each load increment is 5 kN, with 1 cycle per increment. After specimen yielding, increments are multiples of the yield displacement (Δy), with 2 cycles per increment. All models follow the same loading protocol, as shown in Fig. 4. The total nodal rotation is defined as the displacement at the loading point divided by the total beam length. The maximum applied displacement is 80 mm, corresponding to a rotation of 0.035 radians.

Fig. 4.
figure 4

Loading regime

Full size image

3.2 Finite Element Model

A refined beam-column joint model was established using Abaqus finite element analysis software. To ensure the accuracy of the computational results, 8-node linear reduced integration solid elements (C3D8R) were employed for the concrete beams, columns, bolts, and connection components. Local mesh refinement and thickness direction refinement were applied to critical force-bearing regions. The reinforcing bars were modeled using T3D2 elements, and the Embed command was used to incorporate them into the concrete without considering bond slip between the reinforcement and concrete.

The contact interactions between components were simulated using the “Face to Face” command, with Coulomb friction employed in the tangential direction. The friction coefficient between brass plates and steel components was set at 0.2, while for other components, it was set at 0.3 [17]. The normal direction was defined as “Hard Contact”. The contact between steel components and concrete beams utilized the Tie command. The Bolt Load command was used to apply pre-tension to the bolt shank's mid-plane. Subsequent analysis steps simulated the loss of bolt pre-tension by setting “Fixed Current Length.”

In order to prevent the impact of stress concentration on the results, reference points (RP1, RP2, RP3) were established at the column and beam ends. The column-end face and the loading point region were coupled to these reference points.

3.3 Simulation Results Comparison and Analysis

Stress Comparison and Analysis between Cast-in-Place Nodes and Steel Nodes

In the final loading stage (Δ = 80 mm), the stress distribution comparison between the cast-in-place nodes (XJ) and the steel nodes (PT) is illustrated in Fig. 5. From the figure, it can be observed that after loading completion, the XJ node experiences severe tensile damage at the intersection of the beam and column, extending towards the core region of the concrete node. The compressive damage extends from the beam end in an approximately 60° triangular region towards the loading end, and at the bottom of the beam root, the damage has penetrated the entire cross-section. This situation corresponds to the compressive failure of the concrete at the bottom of the beam observed in the experiment [11], accompanied by the phenomenon of longitudinal and shear reinforcement leakage. The PT node exhibits horizontal tensile damage along the beam direction, presenting a pattern of fine and short cracks with a relatively uniform distribution. Compressive damage is only slightly evident at the casting end of the PEC joint where it connects with the concrete.

Fig. 5.
figure 5

Cast-in-place node and steel node cloud diagram

Full size image

Examining the stress contour of the reinforcement in Fig. 5(a,d) reveals that after loading completion, all tensile reinforcements in the core region and at the beam-column intersection in the XJ node have yielded. The maximum stress, σmax, is 620 MPa, corresponding to a strain εmax of 0.13375, approximately 60 times the elastic strain. This indicates severe structural damage to the building under seismic action. In contrast, the PT node exhibits its maximum reinforcement stress at the casting end of the PEC, σmax = 350 MPa, where both longitudinal and transverse reinforcements in the beam-column joint remain in the elastic working stage. Consequently, compared to the cast-in-place node, this steel node can effectively form a plastic hinge mechanism at the beam end, inducing damage concentration in the steel components. After an earthquake, the primary structure remains within the elastic range, safeguarding the main components and achieving a “fuse” effect.

Comparative Analysis of Hysteresis Energy Dissipation Performance

As shown in Fig. 6, comparing the hysteresis curves and skeleton curves of nodes XJ and PT, it can be observed that the hysteresis curve of XJ node exhibits a “spindle shape,” while the hysteresis curve of PT node shows a “parallelogram,” indicating that the steel PT node possesses a stronger energy dissipation capability. In the early loading stage, both the load and initial stiffness of XJ node are higher than those of PT node. XJ node undergoes yielding at a positive displacement of 15.9 mm (yield displacement selected based on the method proposed by Park [18]), and the component's load reaches its peak value at a displacement of 23.8 mm. Subsequently, the load continuously decreases until the end of the loading process.

For the PT node in this study, the moment of initiation of sliding in the frictional energy dissipation system is defined as the yielding point, with a yielding load of 20.2 kN. In the subsequent loading phase, due to the yielding of the shear-resistant system with U-shaped steel plates, the skeleton curve shows a slow initial increase in load, followed by a trend towards a stable state, providing sufficient strength for the structure in the later stages of loading. Clearly, for this steel node, when the load coefficient μ = 0.47, the peak load capacity reaches 31.9 kN, which is comparable to the peak load capacity of a cast-in-place node of the same size (Fmax = 32.6 kN).

Fig. 6.
figure 6

Comparison of hysteretic curves between XJ node and PT node

Full size image

Post-earthquake Rapid Functional Recovery Performance Analysis

To verify the post-earthquake functional recovery performance of the structure, this study utilized the energy balance functionality in ABAQUS to extract the energy dissipation ratios of various components and beams-columns at the prefabricated node. In this node, the overall model energy dissipation can be divided into two parts: the first part is frictional energy dissipation (ALLFD), and the second part is plastic energy dissipation (ALLPD), as illustrated in Fig. 7 The left vertical axis in the figure compares the energy dissipation of different components, including the total energy dissipation of the frictional energy system (F-eds), the U-shaped shear-resistant steel plate system(U-sps), and the concrete beam-column construction system (C-b&c). The right vertical axis represents the ratio of the energy dissipated by the steel components to the total energy dissipation of the entire node (Te-r), with the horizontal axis being the load-bearing capacity coefficient.

Fig. 7.
figure 7

Total energy consumption of each node component

Full size image

As shown in Fig. 7, a comparison of the total energy dissipation of different components under various load coefficient values reveals that when the load coefficient is less than 0.81, the total energy dissipation of the node exhibits an increasing trend with the increase of the load coefficient. However, when the load coefficient exceeds 0.81, the total energy dissipation of the node shows a negative growth phenomenon. The total energy dissipation proportion of the node consistently decreases with an increase in the load coefficient. When μ is less than 0.81, the total energy dissipation proportion gradually decreases, and after μ exceeds 0.81, the energy dissipation proportion decreases rapidly. When μ = 1.06, compared to μ = 0.96, the energy dissipation proportion decreases from 96% to 86%. This decline in energy dissipation proportion is primarily attributed to the fact that, with a load coefficient greater than μ > 0.81, main structural components such as beams and columns start to yield and participate in energy dissipation.

It is evident that for the concrete partial frame nodes selected in this paper, the critical point where the performance of the steel nodes undergoes a significant change is when the load-bearing capacity coefficient μ = 0.81. This differs from the consideration in previous studies, which used μ = 1 as the dividing point. In reference [19], when μ < 1, the replaceable components of the steel nodes yield earlier than non-dissipative components, thereby achieving the design objective of concentrated energy dissipation. However, in this paper, to maintain the main structure in an elastic phase throughout the entire stage, the later increase in the load-bearing capacity coefficient caused by the strengthening effect of U-shaped steel plates should be considered, as shown in Table 1. As all U-shaped steel plates use the same material and dimensions, the consistent increase ratio is 0.19. Hence, when the load-bearing capacity coefficient μ + 0.19 > 1, the device's ability to transmit bending moments has already exceeded the yield load of the steel reinforcement, resulting in the transfer of energy dissipation from the connection components to the main structure. At this point, the connection components are unable to protect the main structure, which also elucidates that μ = 0.81 is a specific parameter for the steel nodes in this paper.

As indicated in the preceding text, under the action of low-frequency cyclic loads, the steel node effectively dissipates energy and induces plastic hinges to appear in the steel node region. When the load coefficient is in the range of μ = 0.47 to 0.81, it ensures that the steel node has sufficient bearing capacity and energy dissipation performance while also ensuring minimal or low damage to the main structural components. After an earthquake, the structure can quickly regain its functional use by simply replacing friction pads, bolt assemblies, and U-shaped steel plate components as needed.

For such resilient connection nodes, the load coefficient is a critical parameter influencing both seismic performance and recoverability of the structure. The minimum load coefficient can be determined by designing the resilient node to have a bearing capacity comparable to that of cast-in-place concrete nodes. The maximum load coefficient is determined based on the increase in bearing capacity due to the strengthening of some components after node yield and the deformation-induced rise in structural capacity. Finally, values within this range can be selected based on actual engineering requirements to meet varying seismic demands.

4 Conclusion

This paper, through numerical simulation and parametric analysis, draws the following conclusions:

  1. (1)

    Under the same bearing capacity, the assembled beam-column joints proposed in this paper are significantly better than the cast-in-place joints in terms of ductility and energy dissipation capacity, showing excellent seismic performance.

  2. (2)

    The structural load-bearing coefficient significantly affects the seismic and recoverable performance of this steel node. When the load-bearing coefficient is not greater than 0.81, the total energy dissipation of the node increases with an increase in the load-bearing coefficient. Within the range of coefficients from 0.47 to 0.81, the node exhibits the best energy dissipation capability and recoverability.

  3. (3)

    This structural system departs from the traditional concept of ‘equivalent cast-in-place.’ Under moderate or severe seismic effects, structural damage and seismic energy are concentrated and dissipated within the steel nodes, thereby safeguarding the main structures such as concrete beams, columns, and the core area of the joints. Post-earthquake, only a partial replacement of detachable components is required for the restoration of structural functionality, forming a resilient structural system.