Multi-level conditional spectrum-based record selection for IDA

Incremental dynamic analysis (IDA) is widespread to evaluate the seismic response of structures. It employs one records set scaled to multiple intensity levels (IMLs) to estimate the structural response distribution. However, the dominating earthquake scenarios differ with the intensity, and a single set introduces bias that increases with scaling. The response is, however, better estimated using multiple stripe analysis (MSA) by selecting multiple hazard consistent record sets at different IMLs via, for example, conditional spectrum (CS) method. Still, IDA remains a popular tool, and would benefit from a single "good" set that would be more amenable to scaling with minimal bias. We explore alternatives of a "multi-level" CS (CSML) scheme, whereby the seismic properties of multiple IMLs are combined to derive a single set. Combined with advanced IMs, CSML provides a viable trade-off between the more accurate and complex method of CS-MSA versus the conceptually and practically simpler IDA.