Because conventional seismic force resisting systems rely on yielding of key structural members to limit seismic forces, structural damage is expected after a design-level earthquake. Repairing this damage can be very expensive, if it is possible at all. Researchers have been developing a new family of self-centring systems that avoid structural damage. One such system is a controlled rocking steel frame, which is the subject of this thesis.

In a controlled rocking steel frame, the columns of a frame are permitted to uplift from the foundation, and the response is controlled by using a combination of post-tensioning and energy dissipation. Although previous studies have confirmed the viability of this system, they have also shown that rocking does not fully limit the peak seismic forces because of higher mode effects. If a structure is designed to account for these effects, it may be uneconomical, but if it is not designed to account for them, it may be unsafe.

The purpose of this thesis is to develop recommendations for the design of controlled rocking steel frames, particularly with regard to higher mode effects. A theoretical framework for understanding higher mode effects is developed, and large-scale shake table testing is used to study the behaviour of a controlled rocking steel frame. Two mechanisms are proposed to mitigate the increase in structural forces due to higher mode effects, and these mechanisms are validated by shake table testing. Numerical modelling of controlled rocking steel frames is shown to become more reliable when higher mode mitigation mechanisms are used to limit the seismic response. In the final chapters, the thesis proposes and validates a new methodology for the limit states design of controlled rocking steel frames.