Conclusions

In this dissertation, the possibility of semi-active control strategies have been explored, and their effectiveness has been verified experimentally for seismically excited buildings. The properties of the hysteretic force-displacement loops produced by semi-actively controlled MR dampers and their seismic performance were investigated.

A review of literature in the area of structural control technologies was first presented with a focus on outrigger damping systems and hybrid base-isolation systems. Literature related to real- time hybrid simulation was reviewed as well. Also, basic background on modern control theories which are necessary to design active and semi-active controllers was introduced, and the servo- hydraulic system model, the MR damper model, and the model-based compensators for RTHS were developed.

The nature of the hysteretic behavior of the active control forces produced by the widely em- ployed LQG-based acceleration feedback control strategies was investigated, revealing the relation- ship between the properties of the control forces and the response. Numerical simulation studies carried out on one-story and three-story buildings with active bracing show that the LQG-based algorithms are quite versatile and can produce controllers with a variety of behaviors. The effec- tiveness of negative stiffness control force was shown through numerical studies. Additionally, the numerical results demonstrated that the presented LQG-based acceleration feedback control had performance comparable to the LQR in the presented SDOF and 3DOF building models.

Following the hysteretic behavior of the active control forces and the seismic responses, hys- teresis loops produced by the semi-actively controlled MR damper were investigated. Two new model-free semi-active control algorithms for controllable dampers were proposed. One of the algo-

rithms needs only the directions of the displacement and the velocity of the damper to decide the input current to the damper. The other one needs the directions of only the displacement and the output force of the damper. Thus, the structure model and a number of sensors are not required to implement the proposed algorithms. Moreover, this research demonstrated that the proposed controllers can produce versatile hysteresis control force loops through numerical simulations on the scaled three-story building model with the MR damper. Also, the effectiveness of hysteresis loops having pseudo-negative stiffness was verified. Additionally, the numerical results showed that the proposed two algorithms producing pseudo-negative stiffness had performance comparable to the LQG-based clipped-optimal controllers, which need the accurate structure model and more sensors.

To show the effectiveness of the semi-active strategies on complicated structures experimentally, the efficacy of the model-based compensator for RTHS on a MDOF structure was verified through a high-rise building model with an outrigger damping system. Through this research, the following general conclusions were drawn: a) RTHS worked when all modes of the structure were lightly damped, demonstrating the robustness of the actuator controller without the need for adding numerical damping; b) the actuator control strategy used in this study demonstrated stable and accurate results in MDOF structural systems; c) RTHS can be employed for validation of structural control algorithms; d) RTHS provides an effective means for assessing the system performance of rate-dependent components in complex structures.

Also, this study showed that MR dampers can be employed effectively in outrigger damping systems using passive-on mode and semi-active controllers. The MR damper’s restoring force can be simulated quite well by the proposed MR damper model for the two earthquake records; however, differences are still present. In particular, discrepancies between simulations and RTHSs were found in the base shear. Moreover, the physical specimen contains no modeling errors, while it is subject to experimental error such as magnitude and time delay. The numerical model provides a good verification tool for RTHS; however it is subject to numerical errors, is only valid within the range of behavior for which the model is calibrated, and cannot fully represent the complex specimen behavior. Thus, the importance of combining RTHS with numerical simulation to ensure accurate results was demonstrated.

A class of isolation systems that use semi-active control devices (i.e., smart base isolation) was investigated through numerical simulation and RTHS as well. In this study, a base-isolated six-story building was considered. The isolation system consisted of linear, low-damping isolators, combined with a MR damper. In RTHS, the smart isolated building was substructured, such that the building was modeled computationally, whereas the MR damper was tested physically. This smart base isolation system is found to reduce base displacements and floor accelerations better than the passive counterparts. Improvements were also demonstrated over the active isolation system, without the need for large external power sources.

On the base-isolated six-story building model, the versatility of the hysteresis loops produced by the LQG-based clipped-optimal controllers and the proposed simple controllers was shown through numerical simulation and RTHS, as well. However, in seismic performance, the proposed simple controllers were not comparable to the LQG-based clipped-optimal controllers, especially in the up- per floors. This result implies that the proposed simple controllers are not suitable for complicated structures.

In conclusion, this dissertation provided the hysteresis behavior and the seismic performance of semi-active control strategies on buildings and showed the strong potential for practical use to mitigate seismic damage. However, some disappointing results were also observed.