Ground motion selection for generalised conditional intensity measure targets and the target construction
File(s)
Author(s)
Shi, Yuan
Type
Thesis
Abstract
A record selection algorithm aims to select a set of ground motion records whose statistics match a target distribution which is consistent with seismological properties of interest for a given site. Thus, the selected records are implicitly consistent with the seismological properties of interest and the subsequent estimations of responses are decided by the seismological properties of interest. This thesis addresses several issues related to ground motion selection in terms of selection algorithms and the definition and construction of target distributions. Two new algorithms for efficiently selecting suites of ground motions that match a Generalized Conditional Intensity Measure (GCIM) target are firstly generated. These algorithms utilize advance sampling methods which allows for a high chance of generating samples consistent with a target continuous distribution and/or allows for additional constrains are priorities for some important properties of interest. A new algorithm for efficiently selecting a suite of records matching to multiple targets as closely as possible is then developed. This algorithm is able to be adopted in the Incremental Dynamic Analysis to reduce biases to a minimum level when there are the significant changes in terms of the distribution of causal parameters among different intensity measure levels. The investigation is subsequently conducted for whether all features of interest for a give site can be well represented by the use of one target distribution constructed following the “Exact” Conditional Spectrum (CS) approach which is commonly regarded as an exact approach within the community. A framework to establish a CS-based target with the goal of accounting for spectral accelerations at multiple damping ratios is finally introduced.
Version
Open Access
Date Issued
2019-04
Date Awarded
2019-09
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Stafford, Peter
Publisher Department
Civil and Environmental Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)