Rocking structures exhibit significant collapse resistance during earthquakes. In studies of rocking rigid bodies, the collision restitution coefficient is typically determined based on the classical model of the rocking rigid bodies. However, during the rocking process, the collision restitution coefficient, influenced by the uncontrollable error in collision energy dissipation between the rigid body and the ground, indirectly impacts the final results of the equations of motion. Therefore, the rationality and reliability of the collision restitution coefficient are crucial for seismic analysis of rocking rigid bodies and self-centering members. This paper introduces a phasic energy dissipation and kinetic energy redistribution model specifically designed for the rocking rigid body within the system of mass point. This model divides the collision into three distinct stages, incorporating energy dissipation considerations in the first two stages to calculate the total kinetic energy of the rigid body. In the third stage, the remaining kinetic energy is redistributed to precisely determine the analytical solution for the collision restitution coefficient of an ideal, homogeneous rectangular rigid body during collision. Lastly, the validity and reliability of the proposed model are confirmed through comparisons with experimental data.

This study aims to enhance the damage controllability and overall seismic resilience of assembled underground structures under earthquake actions. To achieve this, three types of prefabricated roof–sidewall composite joints are proposed based on the design concepts of cogging for force transfer and local strengthening. These include the high-strength bolt–cogging–grouting sleeve joint (HCG), the prestressed steel strand–cogging–grouting sleeve joint (PCG), and the UHPC–cogging–grouting sleeve joint (UCG). Following the principle of positioning joints in regions of low structural stress, four 1/4-scale reinforced concrete (RC) specimens were designed and fabricated, including one cast-in-place (CIP) reference specimen and three precast RC specimens. Quasi-static tests were carried out to systematically evaluate the seismic behavior and internal force distribution of each specimen. Numerical validation was also performed using ABAQUS. The results show that both UHPC and a reasonable application of prestressing can effectively inhibit crack initiation and damage propagation at the joint seams. When the composite joints are positioned outside the plastic hinge region, they provide a reliable load transfer path for the reinforcement. The HCG and UCG joints significantly enhance the load-bearing capacity and energy dissipation capacity of the specimens. Their ductility and energy dissipation both achieve a seismic performance equivalent to that of the CIP specimen. Furthermore, damage in these specimens is predominantly confined to the designated plastic hinge region of the roof. This effectively mitigates shear damage in the roof–sidewall connection zone (RSC). Although the PCG joint improves the initial stiffness of the specimen, its energy dissipation capacity and ductility are reduced. It also causes damage to be transferred to the RSC. This leads to increased shear deformation and premature shear failure in this zone. Consequently, both UHPC and a reasonable application of prestressing can be used for the prefabrication of underground structures. Positioning the joints outside the roof plastic hinge zone can effectively achieve the seismic design goal of “strong joint, weak component”.

This study investigates the flexural performance of prefabricated reinforced concrete (RC) beams with a novel UHPC–cogging–grouted sleeve composite joint. Five beam specimens—one cast-in-place reference and four prefabricated examples—were tested under four-point bending. The key parameters included the presence of interface shear keys and the configuration of middle connecting reinforcement. The results demonstrate that the proposed joint is generally consistent with the “strong joint, weak member” design philosophy. The specimen with shear keys and dedicated connecting bars (W2) achieved the highest peak load of 224 kN, 7.7% higher than the CIP beam, with the joint crack width limited to 0.2 mm at failure. Replacing dedicated bars with two bent-up bottom bars (W4) maintained acceptable performance (219 kN, 5.3% above CIP), while bending up four bars (W5) reduced the capacity to 200 kN, below the CIP level, inducing a near-under-reinforced flexural failure. The UHPC joint and shear keys effectively delayed cracking and suppressed joint damage but shifted the failure mode from flexural to shear-compression. The stress development in the rebars was fully consistent with the failure modes: the midspan rebar dominated in flexure-dominated specimens (580–596 MPa), whereas the loading-point rebar dominated in shear-compression-dominated ones (585–590 MPa). Plane section assumption was validated before cracking. Finite element models developed in ABAQUS reproduced the experimental behavior with deviations within 10%, and the stiffness degradation distributions accurately captured the damage patterns. Preliminary design recommendations are provided, suggesting that the proportion of bent-up bars should be conservatively controlled, subject to further experimental verification.