Seismic Response Control Of Structures Using Superelastic Shape Memory Alloy Wires

Superelastic shape memory alloy (SMA) is a perfect material for passive control of structures. Numerous studies have been conducted in using SMA for structural vibration control, and some developments have been attained. However, there still exist some problems in modeling and application of SMA. Further investigations are carried out in this thesis, concering constitutive model, damper design, vibration analysis and optimization design of damped structures.(1) Experiments on the mechanical behaviors of the superelastic NiTi wires are performed. Cyclic number, loading rate, strain amplitude, and temperature are considered to investigate their effects on the hysteretic curves and the mechanical parameters of SMA (such as critical stress, energy dissipation per cycle, modulus and residual strain). Based on the experimental results, the Lagoudas' multilinear model is modified to reflect the dynamic hysteretic hehaviors or the cyclic mechanical behaviors of SMA, considering loading rate or cyclic number's effects on the parameters of the model, respectively. In addition, the Graesser's model is improved to reflect the variations of the hysteretic behaviors of SMA during loading and unloading.(2) An innovative recentering damping device (RDD) is designed and manufactured, which provides a perfect energy dissipation compatible with a negligible residual displacement, and has such advantages as the simple configuration and the uniform stresses of the wires. Extensive experiments are carried out to investigate the influences of cyclic number, displacement amplitude and loading frequency on the damper's hysteretic curves and the mechanical behaviors (such as maximum output force, residual deflection, secant stiffness, dissipated energy per cycle and equivalent viscous damping ratio). By analyzing the working mechanism of the RDD, a theoretical model is set up on the basis of the multilinear dynamic model of SMA. Some numerical simulations are taken to compare the mechanical behaviors of the damper with different ratio of cross-sectional area of recentering wires to dissipating wires.(3) Seismic nonlinear response time histories and energy analyses of a symmetric structure with the RDD are simulated. Taken a three-story RC frame as an example, some comparisons of the seismic response of the controlled structure and the amount of gross SMA wires are executed, given the same amount of gross wires and the same maximum inter-story displacement ratio of the controlled structure, respectively, and a combination (ratio of cross-sectional area of recentering wires to dissipating wires) of 0.67～2 is suggested to satisfy the desired seismic response of the RDD structure and the economical efficiency. A two-story steel frame model with the RDD is tested on the shaking table, and the results approve that the RDD can decrease the structural vibration and mitigate the residual displacement effectively, which agrees with the theoretical results.(4) Seismic response time histories and energy analyses of an eccentric structure with the RDD excited by bi-directional earthquakes are simulated. Taken an eccentric single-story structure as an example, some comparisons of the seismic response of the controlled structure and the amount of gross wires are executed, given the same amount of gross wires and the same maximum inter-story displacement ratio of the controlled structure, respectively. The results show that the damper with a combination of 0.67～2 can eliminate the coupled lateral-torsional response of the structure at a low cost. Moreover, the seismic response of a three-story eccentric structure controlled by the RDD with a combination of 2 is simulated, considering different earthquakes.(5) To improve computational efficiency of multiobjective optimization algorithms for optimal design of damped structures, a novel one-round tournament genetic algorithm is advanced on the basis of the two-branch tournament genetic algorithm. Using stochastic tournament selection without replacement, the individuals are copied according to the Pareto relationships, and selections in each generation are accomplished in one-round repetition. In order to transfer a constrainted problem into an unconstrainted one, each objective function is revised as a constrainted objective function, according to the precedence rules. Test functions testify that the novel algorithm can search more Pareto optima stretching along the Pareto front in a shorter time. Based on the Pareto optimization theory, the multiobjective optimization models of controlled symmetric and eccentric structures are established to achieve the Pareto optima of the number and the placement of dampers and the seismic response of controlled structures.