Optimization of the Multi-Level Spring Restrainer for Bridges by Hybrid Particle Swarm and Gravitational Search Algorithm

This paper proposes a new multi-level spring restrainer (MLSR) that exhibits multi stiffness performance in different levels of movement of bridge superstructure to prevent unseating during applied dynamic loads. The analytical model of the proposed MLSR was formulated and the fabricated prototype was tested using dynamic actuator. Based on the developed analytical mode, the function of MLSR device relied on 12 parameters that further complicated the design process to achieve the best performance. However, the conventional optimization techniques utilized only one or a few factors for simple systems. Therefore, a multi-objective optimization method is proposed in this study by introducing the hybridization of Particle Swarm Optimization and Gravitational Search algorithm (PSOGSA) to optimize the restrainer parameters, as well as to improve the seismic performance of bridges using the optimum design. The optimized MLSR was implemented in the bridge subjected to multi-directional ground motion and its multi-level action to prevent unseating of bridge deck when the applied excitation was evaluated. The optimization process revealed girder displacement in three directions and the number of plastic hinges decreased from 44 to 99% for the optimized design. The time history analysis disclosed that the use of optimized MLSR device decreased the structural seismic response, such as the 3D deck movements, from 79 to 90%. Next, the base shear and drift ratio of bridge bent reduced to 75 and 85% in longitudinal direction and to 72 and 90% in transverse direction, correspondingly. The outcomes signify that the proposed MLSR device and the optimization algorithm have successfully improved the bridge structure resistance against severe ground motions.