ABSTRACT Eurocode 8 (EC8) allows the use of dynamic analysis for the design and assessment of structures and provides some constraints for the selection of acceleration time-histories as seismic input. However, when it comes to the selection of two-horizontal-component natural ground motion records, parts 1 and 2 of EC8, stipulate apparently different criteria. The first aim of this paper is to build up on previous studies and investigate whether this difference in provisions translates into the selection of systematically different sets of records. A series of record selection case-studies presented in this study, corroborate the preliminary findings of previous work, and show that this is not the case, that is, record sets chosen according to one group of criteria tend to satisfy the other by default. A second aim is to investigate the different options for selecting multi-component ground motion records in part 1, which turn out to be equivalent. Finally, a third issue tackled is the effect that different definitions of spectral acceleration, that the design spectrum refers to, can have on spectrum-compatible record selection, when two horizontal components are involved. The results indicate that, for some of these alternative definitions, such as maximum or random component spectral acceleration, sets obtained via direct spectrum compatibility may not always agree with a simple application of Eurocode 8 provisions. On the other hand, when spectral acceleration is defined as the geometric mean of the two components, consolidated record selection algorithms appear to guarantee spectrum compatibility.
Read full abstract