1. Introduction
The construction of reinforced concrete (RC) bridge components using the in situ construction method is time-consuming, as all the construction processes are completed on-site. Accelerated Bridge Construction (ABC) [
1] aims to reduce on-site construction time by prefabricating most structural elements in a factory. These prefabricated parts are then transported to the construction site ready for use. This technique can reduce the number of workers, control the quality of the structure, and improve the environment [
2]. One promising system to promote the application of ABC in bridges is a self-centering (SC) precast segmental bridge pier system. This system was first introduced in 1999 based on the rocking beam–column connection presented by Priestley and Tao [
3]. The general components of the SC precast column system consist of a series of precast segments pinned together and connected to the cap beam and foundation by unbonded tendons. Previous experimental and numerical studies on the SC precast column system for bridges have shown that the joints between the segments provide an opening and closing mechanism (a rocking mechanism) under lateral loading [
2]. This rocking mechanism can significantly reduce the level of damage caused by earthquakes. The force in the tendons ensures that the column is pulled back into the vertical position, minimizing the residual displacement of the system [
4]. On the other hand, such a system suffers from low energy dissipation (ED) capacity compared to a conventional cast-in-place concrete column.
Several studies have been carried out in the research community to improve the ED capacity by equipping the system with additional energy dissipaters. In this case, the SC is obtained from the force in the unbonded tendons, while the additional energy dissipaters dissipate the energy. These dissipaters can generally be divided into two categories, namely internal ED reinforcing bars and external ED devices. For the first category, various internal ED bars have been investigated in the literature, including bars made of mild steel [
5], high-strength steel [
6], and shape memory alloys [
7]. Among these strategies, using bonded mild steel bars running continuously through the column segments is one of the most common strategies to overcome ED deficiencies. As the system combines the tendon and the ED bars, it is generally called a precast hybrid system. Stone et al. [
4] first presented the precast hybrid system, in which bonded mild steel bars were installed in SC precast beam-to-column connections in a precast moment-resisting frame. Due to the promising performance of the precast hybrid system and the achievement of the required design objectives [
8], it has been applied to various SC structures, including precast shear walls [
9,
10,
11], and precast moment-resisting frames [
12,
13,
14]. Among them, Restrepo and Rahman [
10] carried out an experimental study on SC precast shear walls. The wall units were designed to include mild steel bars placed longitudinally across the joint between the walls and the foundation to enhance the ED capacity and maintain the SC ability. Song et al. [
13] investigated the lateral behavior of a one-bay, one-story SC moment-resisting RC frame. The beam–column connections incorporated two steel bolts to offer an adequate ED capacity. Guo et al. [
14] tested a one-by-two bay, two-story SC moment-resisting RC frame. The adaptable ED capacity is achieved through the employment of web friction devices at the beam ends. More recently, the concept of a hybrid precast concrete system has been extended to concrete bridge piers [
15,
16,
17,
18,
19].
Numerous research papers have proposed to improve the hysteretic ED ability of precast bridge piers by using mild steel bars over the segment joints [
20,
21,
22,
23,
24]. Under seismic loading, the precast hybrid system obtains SC by rocking the precast elements over the unbonded tendons and dissipating energy via mild steel bars. Among these studies, Wang et al. [
17] carried out an experimental investigation on four SC precast segmental long bridge columns under cyclic loading. The system has used various design configurations to increase the lateral strength and hysteretic ED capability of the columns. The addition of bonded mild steel bars across the segment joints is one of these design configurations. The performance of four short large-scale precast SC segmental bridge columns under lateral cyclic loading was investigated by Ou et al. [
18]. The experimental program consisted of one typical precast column (without ED mild steel bars), and three precast columns had different ED mild steel bars ratios running longitudinally across the segment joint and post-tensioning forces. Bu et al. [
21] conducted a study to evaluate the structural performance of various types and arrangements of reinforcement in bridge columns. The study involved testing a monolithic bridge column as a reference specimen, as well as four precast segmental bridge columns with different reinforcement configurations. In one of the specimens, bonded mild steel ED bars were incorporated to enhance the ED capacity of the precast bridge column. According to these studies, compared to a typical precast pier, the hybrid system can ensure a substantial increase in the ED capacity of the structure. In addition, the use of ED bars can increase the column’s load-carrying capacity. However, severe damage characterized by cracking and crushing of the concrete was detected in the bottom segments with ED bars [
17,
18]. The hysteresis curve demonstrated a significant degradation in both the strength and stiffness as a result of the severe spalling at the column toes due to the substantial gap opening between the segments’ joint [
21]. In addition, due to the large displacement of the column and an increase in the opening between the segments, a number of the ED bars fractured [
17,
18]. The test results also showed that the test specimens exhibited a large residual drift after unloading, which exceeded 1% due to the inelastic deformation of the steel bars. According to Japanese code and specifications, it is challenging to rehabilitate bridges that have a residual drift of more than 1% [
25]. Therefore, using precast SC bridge piers with mild steel bars may improve the ED capacity but extreme post-earthquake damage and residual drift may still occur. Therefore, using the hybrid precast system as the main system may not achieve the desired rapid recover ability after severe earthquakes. However, it can be used as an ED source for the main SC system.
This research presents a novel resilient system for precast concrete bridge columns that can overcome the above limitations of precast SC bridge columns. The proposed system consists of a typical hollow RC precast column and a newly introduced hybrid RC precast column serving as an ED source. The ED core column (ED–CC) combines prestressed tendons and bonded mild steel bars. The precast segmental column and the ED–CC contact each other at the interface between the head faces of the ED–CC and the corresponding inner faces of the concrete segments of the precast column, as shown in
Figure 1a. Under seismic loading, the ED–CC dissipates energy via mild steel bars. Since the main system (precast segmental column) and the damping source (ED–CC) exhibit SC ability, the entire system ensures controlled residual displacement of the proposed system. To investigate the lateral behavior of the proposed system, a finite element model was first developed for a typical precast segmental column that was experimentally tested. Then, the verified model was systematically studied with the proposed ED–CC. The investigated parameters are related to the design details of the ED–CC, including: reinforcement ratio, prestressing force level, and unbonded length of the reinforcing bars. To improve damage tolerance due to large compressive stresses at the toes of the ED–CC, the FRP sheathing on the lower part of the column was also investigated.
3. Resisting Mechanism
The intended seismic resisting system ensures that the bridge’s main substructure system and the ED units work in parallel. The main column resists lateral loading through a rocking mechanism, in which the prestressed force is applied via tendons to integrate all components of the bridge pier (the cap beam, the pier segments, and the foundation), as shown in
Figure 1a. Post-tensioning is also effective in reducing the permanent deformations of the bridge after a seismic action: the rocking system is an SC system. The ED source is a non-emulative unit, consisting of a prefabricated RC column connected to the foundation and contributing to the lateral resistance system of the whole structure. The entire system will receive the necessary damping energy under lateral load from the mild steel bars yielding across the ED–CC and the foundation connection. The ED–CC will also contribute to resisting lateral loads and positively distribute the opening throughout the column segments, thus decreasing localized stress. Additionally, the prestress force in the tendons of the ED–CC can work in parallel with the main post-tension system of the entire structure to limit the residual drift that remains as a result of the mild steel bars’ inelastic nature (controlling the residual drift). The core ED column is located within the main rocking system, and it may not affect its elastic stiffness. Even so, the yielding of the steel bars is responsible for the required ED. The ED–CC column and the main rocking system integrate only in the case of lateral displacement. Under lateral load, the contact between the head surfaces of the ED–CC and the corresponding internal surfaces of the main column segments will activate the lateral resistance of the ED–CC. In this case, the lateral resistance of the system is the summation of the lateral resistance of the main system and the lateral resistance of the added ED–CC. That is, the introduced system is a new resilient system capable of dissipating energy without compromising the seismic demands or the required repairability.
The moment capacity of the proposed system,
My, is calculated using Equation (1) as a function of the moment capacity of the main SC segmental column,
MSC, and the moment capacity of the ED–CC,
MED. As shown in
Figure 2a, the main SC system behaves linearly with the initial stiffness
K1 until the rocking joints open at point
Y1. From point
Y1, the tendons elongate elastically and form a post-yielding stiffness
K2. With the introduction of ED–CC, the system shows a flag-shaped hysteretic response described by the yield moment of the system
MED, the flag height
MF, and two stiffnesses (
K3 and
K4), as shown in
Figure 2b. Both the moment capacity of
MSC and
MED can be designed based on the available recommendations and guidelines of the precast SC system [
2,
18]. For the main SC system, the moment capacity of the main segmental column can be calculated based on the simplified analytical model developed by Hewes and Priestley [
2] for typical SC precast segmental bridge columns. Hewes’ work was based on the concept of the “monolithic beam analogy”. For the ED–CC system, the moment capacity can be calculated using the “moment–interaction diagram” developed by Ou et al. [
18]. The moment–interaction diagram can be constructed using curvature analyzes of a proposed cross-section and reinforcement details under consideration with varying axial force from the deformation of the unbonded tendon. When the main SC system is combined with the ED–CC system, the hysteretic moment–drift behavior is shown in
Figure 2c. The moment–drift response of the proposed system is the superposition of the main SC system response and the ED–CC system response, as will be proved through the FE results. When the post-yielding stiffness of the SC system,
K2, is relatively small compared to the initial stiffness,
K1, and the initial stiffness of the ED–CC unit,
K3, Equation (1) can be simplified as shown in Equation (2), neglecting the effect of the additional elongation of the tendon between the opening of the rocking interfaces and the yielding of the reinforcement in the ED–CC unit. For the unbonded tendon, neglecting this additional elongation is a simplifying assumption that allows for a more manageable analysis of the overall structural response due to the relatively small magnitude of this elongation compared to other deformations and the complexity it introduces to the analysis.
4. Finite Element Modeling
To investigate the lateral behavior of the new system, a precise and comprehensive 3D nonlinear finite element (FE) model of prefabricated bridge piers was generated by the ABAQUS program [
28]. First, a comparison was made between the outcomes of the 3D FE and the existing test findings. Subsequently, the ED–CC was simulated and integrated into the previously validated model. The following section provides a brief overview of the precast test specimens provided in previous research [
17], identifies the specifics of the models, and explains the stages of the modeling procedure.
4.1. Description of the Test Specimens
According to the available test specimens of precast concrete bridge columns conducted by Wang et al. [
17], a detailed 3D FE model was developed. The model included three large-scale columns (P1, P2, and P3). The columns consisted of hollow precast concrete segments with cross-section dimensions of 1800 × 1200 mm and wall thickness of 300 mm, as shown in
Figure 3. Each column had an RC cap beam and an RC foundation. The axial and prestressing forces on the three columns amounted to 4000 kN.
For the test specimen P1, nine precast RC segments with a hollow cross-section were assembled, each one meter high. These segments were connected with eight unbonded tendons. The transverse reinforcement consisted of D13 mm mild steel bars, while the longitudinal reinforcement had D22 mm mild steel bars. The reinforcement of each segment did not continue from one segment to another. Consequently, the tendons acted as the only continuous longitudinal reinforcement, supporting the entire height of the column. The P1 test specimen was specifically intended for a regular SC precast concrete bridge pier.
P2 was designed with eight high-strength ED bars with a diameter of 36 mm at the foundation-S1 critical connection. To ensure a secure connection, T-headed threaded couplers were utilized in the foundation to fasten these bars. Additionally, four of these bars extended through grouted steel corrugated ducts up to S2, where they were firmly fastened with steel bolts and plates. The remaining four bars were also securely fastened and extended to the top of segment S5. Moreover, the prestressing tendons of specimen P2 were typically pressure-grouted during stress exposure to prevent corrosion and enhance the lateral strength of the column.
A total of eight RC precast segments were managed for the P3 test specimen. Furthermore, the height of the S1 segment was increased to 2 m. Six bonded tendons were positioned in the P3 specimen to ensure a more central distribution. In contrast to the P2 specimen, the whole circumference of the cross-section was equipped with twenty ED bars, which had a lower yield strength. The twenty ED bars extended from the RC foundation to the first segment. Additionally, twelve bars of the twenty ED bars extended to the third segment, while eight ED bars of the twenty bars extended to the fifth segment. Additional steel bars were used to improve the connection between the first segment and the concrete foundation.
4.2. Modeling Procedure
All models created used an eight-node 3D brick element (C3D8R) to construct all concrete parts, such as the cap-beam block, precast RC columns, RC foundation, and ED–CC. The concrete damage plasticity model, also known as the CDP model, is commonly used in the ABAQUS program to define the materials of the concrete parts. This study utilized CDP input data, using a soft CDP model from reference [
29].
Table 1 summarizes the maximum concrete compressive capacity for each specimen. In defining the parameters of the CDP model, the values of 30, 0.1, 1.16, 0.667, and 0.001 were assumed. These parameters correspond to the dilation angle, the flow potential eccentricity, the ratio of initial equibiaxial compressive yield stress to initial uniaxial compressive yield stress, the ratio of the second stress invariant on the tensile meridian to that on the compressive meridian, and the viscosity parameter, respectively. These input data are consistent with the suggestions of the software documents [
28] and available existing studies by Dawood et al. [
30] and Li et al. [
22].
The post-tensioned tendons, the cage reinforcement of the segments, and the longitudinal ED bars of the ED–CC were simulated with the truss element (T3D2). In order to anchor the unbonded tendons, a specified distance of the tendons was embedded in the RC cap beam and the RC foundation. On the other hand, the bond tendons, ED steel bars, and cage reinforcement bars of the segments were embedded in the solid concrete parts. The behavior of all truss elements was described with a bilinear elastic–plastic relationship based on the properties listed in
Table 1. Surface contact elements were used to simulate the rocking manners between the joints of the column segments and the contact between the ED–CC head’s outer surfaces and the column segments’ inner surface. Friction with a coefficient of 0.5 governed the behavior of the tangential interaction between the surfaces of the solid parts [
22]. To prevent interaction between the components through connection, the normal contact behavior of the interfaces was defined using hard contact. The BFRP confinement for the lower part of the ED–CC is simulated with the shell element (S4R) available in the ABAQUS program version 6.2. A thickness of 2mm, 4mm, and 6mm was adopted for the BFRP shell element. The lower part of the foundation was fully restrained. In addition, the extended ED bars of the ED–CC were embedded in the foundation.
In order to assess the impact of different mesh dimensions on the outcomes of the model and enhance the efficiency of the program analysis, a sensitivity analysis was performed. The analysis focused on examining how the size of the mesh affected the results of the finite element model. Three specific mesh sizes were implemented for the P1 specimen: 50 × 50 mm, 100 × 100 mm, and 150 × 150 mm. Afterward, the numerical results were compared to the experimental results obtained from load–displacement curves. The findings indicated that the numerical curves were nearly indistinguishable, particularly for columns with mesh sizes of 50 mm and 100 mm. Consequently, three distinct mesh volumes were chosen to uniformly divide the column elements, as depicted in
Figure 4. These mesh dimensions consisted of 100 mm, 150 mm, and 200 mm for the upper, middle, and lower segments of the column, respectively.
The loading process of the developed models was carried out in three stages, corresponding to the loading order of the experimental test. The tendons were post-tensioned in the first step. The initial stress available in ABAQUS software version 6.2 was used to apply the post-tensioned force to the tendons in the first step. A gravity load of 4000 kN was applied to the top surface of the upper solid block of the column segments. Finally, the lateral load was applied using displacement control to drift ratio of 0.25, 0.375, 0.5, 0.75, 1.0, 1.5, 2.0, 3.0, and 4.0%. Additionally, the P3 test specimen experienced an additional 5% drift level.
4.3. Model Validation
In this section, a comparison is made between the FE results and test outputs of precast bridge pier columns subjected to cyclic loads. The comparison is divided into two stages, focusing on the damage mode and the hysteretic response.
4.3.1. Damage Pattern
Figure 5 illustrates the extent of damage observed in the FE model for specimen P1 and specimen P2 at a displacement ratio of 4%, and the damage pattern for specimen P3 at a displacement ratio of 5%. The figure also presents the damage pattern from experimental tests. The damage pattern in the FE outputs was assessed by selecting the strain limit. The dark elements of the model indicate that the pressure in these elements has exceeded the maximum compressive strength. This color was observed on the loading side of the first segment of sample P1, indicating that the stress in this part exceeded the maximum compressive strength. This observation is consistent with the experimental test findings of sample P1, where spalling was observed on the cover of the concrete in the lower part of segment S1, as shown in
Figure 5a. Similarly, in sample P2, which utilized high-strength reinforcement bars, tension cracks were distributed along the lower part of the column. The same observation was noted in the test results, in which various cracks were observed in the bottom segments of the column, as shown in
Figure 5b.
Figure 5c demonstrates that the FE model results accurately captured the damage pattern of sample P3 compared with the test sample. There is a significant increase in the tensile stress at the bottom of the model, and blackening can be observed between the first and second segments of the model.
4.3.2. Load–Displacement Hysteretic Curves
According to the findings of the experiments and the FE numerical calculations, the hysteric curves of the three prefabricated columns are shown in
Figure 6. In the figure, the red dots represent the experimental test curves. In contrast, the continuous black lines represent the numerical model output. Regarding the initial stiffness, the stiffness after yielding, and the ultimate strength, the numerical findings shown in
Figure 6 are consistent with the experimental results. Moreover, the stiffness degradation observed in the tests is well reflected by the hysteresis curves of the FE models. The numerical FE model’s hysteretic curves faithfully reproduce the residual displacement measured from the experimental test.
Table 2 provides a quantitative analysis of the FE results and associated errors compared to the experimental test data for the lateral force at each drift level. It can be noted that the maximum average error between the predicted and experimental results for columns P1, P2, and P3 is below 8%, which is considered satisfactory. In general, the numerical models show considerable accuracy in predicting the lateral response of the precast segmental bridge columns.
5. Parametric Study
Table 3 summarizes the four design parameters that were considered for the ED–CC. These parameters are the reinforcement ratio, the ratio of prestressing forces, the FRP wrap thickness, and the ED bars’ unbonded length. Fifteen cases were included in the parametric analysis. Specimen P1 was selected to investigate the lateral behavior of the precast segmental column with ED–CC. The ED–CC consists of two prefabricated elements with a circular cross-section (see
Figure 4), and its vertical length ratio is 40% of the main column length. This ratio was recommended by [
31] to ensure stable lateral behavior. The circular cross-section was chosen to ensure priority in the confinement. Each fuse element is designed with a diameter of 600 mm. As shown in
Figure 4, a solid block (the head of the ED–CC) connects the two fuse elements and acts as a cap beam for the ED–CC. The steel reinforcement of the ED–CC has a yield strength capacity of 235 MPa. The concrete of the ED–CC has a compressive strength capacity of 37 MPa, similar to that of the P1 specimen.
To increase the friction between the ED–CC and the foundation, it is necessary to increase the force in the tendons during the design. This ensures the maintenance of shear strength in the precast column and resistance to shear forces resulting from seismic actions [
2]. However, the loads in the tendons can lead to high stress on the precast column during the rocking mechanism, decreasing the ED–CC’s ductility [
20]. Three prestressing force levels were carefully selected to investigate the effects of force ratio in ED–CC tendons on the response of the proposed system. These force ratios correspond to 5%, 10%, and 15% of the axial compressive strength capacity of the ED–CC (
Ag. f′c), where (
Ag) represents the gross section area of the ED–CC and (
f′c) represents the concrete compressive strength.
The prestressing force in the tendons of the ED–CC can be denoted as PT followed by the respective percentage; e.g., PT10% means that the prestressing force in the tendons is 10% of the axial compressive strength capacity of the column. The study investigates different reinforcement levels of 2%, 4%, and 6% for each prestressing force level. The parameter indicating the reinforcement ratio of the ED–CC is symbolized as R. For clarification, R4% means that the ratio of the reinforcement of the ED–CC is 4%. It is important to note that the different reinforcement ratios were achieved by modification of the cross-section area of the bar. At the same time, the number and arrangement of rebars remained unchanged.
To prevent early fatigue failure at low cycles, the unbonded length of the ED bars in the ED–CC at the rocking connection was adopted. The unbonded lengths of zero, 100, 200, and 300 mm were chosen to investigate how the ED bars in the ED–CC’s unbonded length affected the proposed column’s performance (as listed in
Table 3). Among the listed cases in
Table 3, a length of 200 mm was left unbonded. The letter U expresses this parameter; for instance, U2 indicates that the ED bars in the ED–CC have an unbonded length of 200 mm.
Under lateral loads, the ED–CC allows a rocking mechanism. During the rocking of the column and at large displacements, the lower part of the ED–CC may suffer great damage due to the large compressive stress at the rocking interfaces. Therefore, to improve the maximum lateral strength capacity and control the damage of the ED–CC, the effects of three BFRP sheets’ volumetric confinement ratios of 1.33%, 2.7%, and 4% were studied. These volumetric confinement ratios correspond to a thickness of 2 mm, 4 mm, and 6 mm of BFRP sheets. The maximum tensile strength and elastic modulus of the BFRP sheet were 2100 MPa and 91 GPa, respectively [
32]. The height of the externally confined zone was 1000 mm from the top surface of the foundation, as shown in
Figure 4b.