Return
An Efficient Methodology for Multivariate Return Period-Based Target Spectrum Construction Using Copula Function
DOI:10.1002/eqe.70172.png)
Abstract
En 中文
In performance-based earthquake engineering, selecting hazard consistent ground motions is critical for seismic demand assessment. Traditional ground motion selection based on a single conditional intensity measure (IM) cannot capture the joint hazard of vector-valued IMs. Although the concept of a multivariate return period (MRP) target spectrum addresses this limitation, its practical application has been restricted by high computational costs especially for high-dimensional IM vectors or long return periods. This paper presents a computationally efficient framework for constructing MRP-based target spectra through two key innovations. First, the direct integration of the high-dimensional Gaussian mixture distribution (GMD) is replaced with a fitted Gaussian Copula-based evaluation. Second, A pre-screening strategy based on Fréchet bounds is introduced to filter more than 94% of irrelevant Monte Carlo samples, substantially improving computational efficiency across different MRP definitions. The hazard consistency of the resulting Kendall-based MRP target spectrum is evaluated at both scalar and vector levels. A case study of an eight-story reinforced concrete frame designed according to Chinese codes demonstrates the applicability of the method to both intensity- and risk-based assessments. The calculated engineering demand hazard curve results confirm that the multi-period MRP target spectrum eliminates potential biases associated with a single conditioning period at low annual exceedance probability. It leads to more stable assessments for both drift- and acceleration-sensitive engineering demand parameters. This research offers an efficient and robust solution for MRP-based hazard-consistent ground motion selection in practice.
Keywords:
ground motion selection
hazard consistency
multivariate return period
seismic demand assessment
seismic hazard analysis
Journal
E
IF:
5
Papers:
149
Citations:
0

