Self-adaptive approach for optimisation of passive control systems for seismic resistant buildings

The concept of passive control of the seismic response of structures was introduced to improve the performance of structures by increasing their energy dissipation and reduce or eliminate damage in the structural elements. The key task in the design of passive systems is to determine the forces in the control devices (yield/slip or post-tensioning) at each floor, that will result in best performance (e.g. minimum inter-storey drift). This can be achieved by large parametric studies in which both the maximum control force (e.g. at ground level) and the distribution of forces along the height of the structure are varied. Alternatively, optimum forces in the devices can be achieved by semi-active control, where the structure self-adapts to the earthquake. Both solutions are expensive: the first requires hundreds of non-linear response simulations in the design stage; the second needs a system of sensors, controllers and electromechanical devices. Presented here is a new Self Adaptive Optimisation Approach (SAOA) in which the self-optimisation of a semi-active system is used in the design stage and the resulting distribution of control forces is adopted as a passive system. The new approach was evaluated through comparing the simulated dynamic responses of two relatively simple benchmark structures (braced and post-tensioned) with three sets of control forces: (1) passive system with forces obtained in parametric study, (2) semi-active system with self-adapting control forces, and (3) passive system with SAOA-optimized forces. The results show good performance of the SAOA systems, indicating that SAOA offers a simple and effective solution that can replace the existing optimisation approaches for the design of passively controlled earthquake resistant structures. This study presents a novel idea of using the semi-active control as a tool for optimising a passive control system. The passive control systems can be further improved by a larger study in which the semi-active control algorithms are also optimised.