This thesis addresses a diverse range of topics in the area of probabilistic seismic risk
analysis of engineering facilities. This intentional path of diversity has been followed
primarily because of the relatively new and rapid development of this facet of earthquake
engineering. As such this thesis focuses on the rigorous scrutinization of current, and in
particular, simplified methods of seismic risk assessment; the development of novel aspects
of a risk assessment methodology which provides easily communicated performance
measures and explicit consideration for the many uncertainties in the entire earthquake
problem; and the application of this methodology to case-study examples including structures
supported on pile foundations embedded in liquefiable soils.
The state-of-the-art in seismic risk and loss assessment is discussed via the case study
of a 10 storey New Zealand office building. Particular attention is given to the quality and
quantity of information that such assessment methodologies provide to engineers and
stakeholders for rational decision-making.
Two chapters are devoted to the investigation of the power-law model for representing
the ground motion hazard. Based on the inaccuracy of the power-law model at representing
the seismic hazard over a wide range of exceedance rates, an alternative, more accurate,
parametric hazard model based on a hyperbola in log-log space is developed and applied to
New Zealand peak ground acceleration and spectral acceleration hazard data. A semianalytical
closed-form solution for the demand hazard is also developed using the hyperbolic
hazard model and applied for a case-study performance assessment. The power-law hazard
model is also commonly used to obtain a closed-form solution for the annual rate of structural
collapse (collapse hazard). The magnitude of the error in this closed-form solution due to
errors in the necessary functional forms of its constitutive relations is examined via a
parametric study.
A series of seven chapters are devoted to the further development of various aspects of
a seismic risk assessment methodology. Intensity measures for use in the estimation of
spatially distributed seismic demands and seismic risk assessment which are: easily
predicted; can predict seismic response with little uncertainty; and are unbiased regarding additional properties of the input ground motions are examined. An efficient numerical
integration algorithm which is specifically tailored for the solution of the governing risk
assessment equations is developed and compared against other common methods of
numerical integration. The efficacy of approximate uncertainty propagation in seismic risk
assessment using the so-called First-Order Second-Moment method is investigated.
Particular attention is given to the locations at which the approximate uncertainty propagation
is used, the possible errors for various computed seismic risk measures, and the reductions in
computational demands. Component correlations have to date been not rigorously considered
in seismic loss assessments due to complications in their estimation and tractable
methodologies to account for them. Rigorous and computationally efficient algorithms to
account for component correlations are presented. Particular attention is also given to the
determination of correlations in the case of limited empirical data, and the errors which may
occur in seismic loss assessment computations neglecting proper treatment of correlations are
examined. Trends in magnitude, distribution, and correlation of epistemic uncertainties in
seismic hazard analyses for sites in the San Francisco bay area are examined. The
characteristics of these epistemic uncertainties are then used to compare and contrast three
methods which can be used to propagate such uncertainties to other seismic risk measures.
Causes of epistemic uncertainties in component fragility functions, their evaluation, and
combination are also examined.
A series of three chapters address details regarding the seismic risk assessment of
structures supported on pile foundations embedded in liquefiable soils. A ground motion
prediction equation for spectrum intensity (found to be a desirable intensity measure for
seismic response analysis in liquefiable soils) is developed based on ground motion
prediction equations for spectral accelerations, which are available in abundance in literature.
Determination of intensity measures for the seismic response of pile foundations, which are
invariably located in soil deposits susceptible to liquefaction, is examined. Finally, a rigorous
seismic performance and loss assessment of a case-study bridge structure is examined using
rigorous ground motion selection, seismic effective stress analyses, and professional cost
estimates. Both direct repair and loss of functionality consequences for the bridge structure
are examined.
| Identifer | oai:union.ndltd.org:canterbury.ac.nz/oai:ir.canterbury.ac.nz:10092/2588 |
| Date | January 2009 |
| Creators | Bradley, Brendon Archie |
| Publisher | University of Canterbury. Department of Civil and Natural Resources Engineering |
| Source Sets | University of Canterbury |
| Language | English |
| Detected Language | English |
| Type | Text, Electronic thesis or dissertation |
| Rights | Copyright Brendon Archie Bradley, http://library.canterbury.ac.nz/thesis/etheses_copyright.shtml |
| Relation | NZCU |
Page generated in 0.0063 seconds