Loss optimisation seismic design (LOSD) is a performance-based design procedure which aims to introduce a simple, yet reliable, design methodology for the next generation of seismic design codes. A preliminary framework for LOSD based on simplified prediction of various sources of loss has been proposed in the literature. The first part of this thesis aims to further extend and enhance the LOSD framework. To this purpose, three different levels of computational rigor (i.e., rigorous, semi-rigorous and simplified) are defined for LOSD which render it as a comprehensive design methodology that can cater different projects depending on the importance, complexity, regulatory requirements, and available resources. Investigation of the LOSD framework reveals the need for extensive additional research to develop simplified modules of LOSD. Among those is the development of simplified functions/methods to estimate seismic demands for estimating loss, i.e., inter-storey drifts, peak floor accelerations and seismic collapse probability at various hazard levels. Thereby, the second part of the thesis is predominantly focused on the estimation of seismic collapse probability for RC frame buildings.

To develop simplified methods to predict seismic collapse probability, various aspects of the seismic collapse probability prediction of buildings are scrutinised. First, application of an enhanced nonlinear fiber element model for simulation of seismic collapse is examined. The fiber element model utilises advanced nonlinear material models to replicate buckling, cyclic degradation, and fatigue of reinforcing bars. It is demonstrated that the nonlinear fiber-element model enables simulation of sidesway mode of collapse (because of lateral instability) and vertical mode of collapse (as a result of the loss of vertical load carrying capacity of structural components).

Next, the effect of ground motion (GM) selection on the seismic collapse probability estimation of RC frame buildings is investigated. To this purpose, collapse probability assessment is carried out for a case study building using four different contemporary GM selection methods. It is shown that for prediction of collapse probability at a given hazard level, conventional GM selection methods, using a uniform hazard spectrum to match GMs to a target spectrum, offer conservative yet more reliable results for engineering purposes, as they are less dependent on the structural properties and GM selection parameters.

In the next stage of the research, seismic collapse probability assessment is carried out for an extensive range of case study buildings. It is demonstrated that collapse probability of RC frame buildings is highly dependent on design inter-storey drifts and anti-buckling detailing of plastic hinge regions of structural components. Moreover, it is shown that collapse probability of RC frame buildings designed in accordance with NZ standards may become high if buildings are designed with large inter-storey drifts. Based on the results of the case study buildings, simplified equations for prediction of seismic collapse probability at the maximum credible hazard level are proposed. Subsequently, by utilising the proposed equations, a new seismic design approach aiming to limit the collapse probability within an acceptable level is presented.

Finally, height-wise variation of seismic demand parameters for RC frame buildings obtained from nonlinear response history analyses for selected case study buildings are illustrated in the appendix. It is demonstrated that the anti-buckling detailing and confinement of structural components highly affects the height-wise variation of the maximum inter-storey drift demands of RC frames, which is not captured by linear analysis methods. Comparison of inter-storey drift predictions by nonlinear analyses and a linear equivalent static procedure and response spectrum analysis demonstrate that linear procedures provide conservative estimations. Peak floor acceleration demands obtained from nonlinear response history analyses are shown to be close to the peak ground acceleration for buildings that are responding highly in nonlinear range.