Lukas Moschen
Contributions to the probabilistic seismic assessment of acceleration demands in buildings
In the early days of earthquake engineering less attention has been paid to building contents, often referred to as nonstructural components (NSCs). In the last two decades, however, the potential seismic risk associated with NSCs has been recognized, separated into life safety and economic loss.
It is reasonable to distinguish NSCs with respect to their response behavior. Displacement sensitive (drift sensitive) NSCs such as claddings or partition walls are mounted multiply at different slabs of the load-bearing structure, and thus, are damaged if the inter-story drift ratio exceeds a certain amount. In contrast, NSCs such as boilers or medical equipment are attached at a single wall or a single slab only. The equivalent seismic force is proportional to the total acceleration of the center of mass of the NSC. Thus, these NSCs are denoted as acceleration sensitive.
During the design process usually the seismic behavior of NSCs cannot be considered by the structural engineer. Moreover, relocating an NSC in a facility may lead to amplified responses, and subsequently, to higher damage probabilities. Hence, efficient and sufficiently accurate methods are required to estimate the maximum absolute (peak) responses of NSCs. This dissertation focuses on the probabilistic seismic assessment of peak floor acceleration (PFA) demands. Research effort during the last decade has shown that damage of acceleration sensitive NSCs is well correlated with the PFA demand of the attachment point. Thus, the strategy of this research is to estimate PFA demands of the load-bearing structure in order to draw conclusions of the seismic response behavior of NSCs.
In the first part of this dissertation a stochastic ground motion selection procedure is introduced to identify quickly record sets consistent with the site specific hazard. The selection is based on an optimization procedure such that the first moment and the second central moment of the spectral acceleration of the record set matches a target spectrum and a target dispersion in a certain period (or frequency) range. Information of the structure such as the fundamental period is explicitly not required. Once a record set is found, it can be used for a broad class of buildings at the specified location. This is an enormous benefit compared to common ground motion selection algorithms.
Generic structural models tuned to fundamental properties of real buildings are developed in an effort to the study the seismic PFA demand. The generic formulation of these structures allows sensitivity analyses, and thus, global conclusions for a class of building types such as moment-resisting frames or structural walls can be drawn. The novelty of the presented generic buildings is its extension to spatial structures and the explicit consideration of vertical PFA demands as a consequence of the vertical component of the ground motion. Additionally, the response behavior to multi-directional earthquake input can be studied yielding new insights for seismic assessment of acceleration sensitive nonstructural components.
Nonlinear response history analysis of various steel-, concrete-, and wall structures is conducted, which, nowadays, provides the closest approximation of real structural behavior in a seismic event. Based on an inelastic single-degree-of-freedom system the concept of acceleration ductility is introduced, which is consistent to the definition of the well known displacement ductility. The concept of acceleration ductility can be used to assess in a simplified manner the PFA demand of inelastic multi-story structures. The PFA demands obtained from response history analysis of multi-story structures are consistent with observations of other research groups. It is shown that application of simplified methods provided in building codes may lead to non-conservative estimations of PFA demands, particularly in the vertical direction.
Nonlinear response history analysis methods are, however, computationally expensive and can only be conducted by an experienced seismic engineer. To avoid this effort, response spectrum analyses based on various modal combination rules are widely accepted in research and professional engineering. If correctly applied, a wealth of information can be exploited with minimum computational effort. In this dissertation a response spectrum method for estimation of elastic PFA demands is rigorously derived. Application to various planar and spatial structures shows the superiority of this method compared to approaches available in literature. Additionally, various simplifications of this method are presented and subsequently evaluated.