Developing a practice-oriented method for the prediction of floor response spectra in buildings

Abstract

A method is developed for predicting seismic demands on secondary and non-structural components in buildings using floor response spectra. The method aims to provide reliable predictions which improve on existing code methods and maintains simplicity to enable adoption in practical design. This work is motivated by recent seismic events which have illustrated the significant costs that can be incurred following damage to these elements within buildings, even where the structural system has performed well. This has prompted increased attention to the seismic performance of non-structural components with questions being raised about the accuracy of design floor acceleration response spectra used in practice.

Floor response spectra are observed to be heavily influenced by the modal characteristics of the lateral load resisting system of a building. The work developed here builds upon previous work to develop a prediction method by superposing demands associated with the dynamic response of the supporting structure. Provisions for low intensity earthquake demands are first developed considering the elastic response of non-structural components and the supporting structural system based upon observations from New Zealand accelerometer-instrumented buildings that recorded earthquake motions in the last decade. The balance between accuracy and simplicity is investigated considering the proposed method and the use of explicit numerical modelling of the supporting structure through time history analysis. The prediction method is then further developed to consider nonlinear non-structural responses, and the development of inelasticity in fixed base reinforced concrete and steel buildings. Seismic demands on non-structural components in buildings with the novel high-performance low-damage engineering technologies of steel con- trolled rocking braced frames and base isolation are investigated, and the prediction method is expanded to consider these advanced systems.

The developed method shows promising predictive performance and has a robust rational basis. It is shown that the modal characteristics of the supporting structure generate amplified peaks in floor response spectra demands. Furthermore, however, the development of nonlinearity in the supporting structure limits the magnitude of such amplified peaks and also broadens the period range over which large demands can be expected. The effect of nonlinear response of the supporting structure is more significant on floor spectra demands associated with the fundamental mode than higher modes. The provision of non-linear deformation capacity in the NS component is seen to significantly reduce the strength requirements, particularly for components possessing periods that are resonant with the supporting structure. The new methodology includes provisions to account for all these factors. In rocking frames, demands associated with the fundamental mode are found to cap and broaden to longer periods. Significant apparent reductions of demands on non-structural components can be achieved through base isolation, and demands in buildings with lead-rubber bearings were better estimated using the developed approach than those with friction slider bearings, which cause large changes in the dynamic properties of the isolated building system after activation.