An efficient stratified-based ground motion selection for cloud analysis
Cloud analysis is numerical method to estimate seismic demand model by performing a number of time-history analyses using unscaled (or minimally scaled) ground motion records. Therefore, an important issue is to reduce the required number of analysis while maintaining the method’s accuracy. This research proposes an adaptive sampling scheme, where at each iteration seismic demand model is updated based on a Gaussian clustering algorithm. A case study example showed that this method led to the same results as a previously established benchmark using a significantly smaller number of analyses (60 records versus 5000 records).