Loading...

Modeling of Epistemic Uncertainties in Seismic Hazard Analysis Based on Reliability Methods

Askari, Mohammad | 2020

1569 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52720 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Mahsouli, Mojtaba
  7. Abstract:
  8. The primary objective of this thesis is comprehensive treatment of epistemic uncertainties in probabilistic seismic hazard analysis based on structural reliability methods. The classical methods of probabilistic seismic hazard analysis, i.e., PSHA, traditionally consider a limited number of uncertainties. The ones that they do consider comprise the aleatory uncertainty in the magnitude, the distance, and the residuals of ground motion prediction equations and the epistemic uncertainty in the intensity measures predicted by different such equations. However, there are many sources of epistemic uncertainty that, according to past studies, have a significant impact on the upper tail of the probability distribution of ground shaking intensity measures and, consequently, significantly impact the design-level earthquake intensity measures. Identification and modeling of these sources of epistemic uncertainty are also crucial to guide the modeling and data gathering efforts to reduce these uncertainties. These sources include epistemic uncertainty in 1) Geometrical characteristics of seismic sources; 2) Occurrence rates of seismic events; 3) Focal mechanism of seismic sources; 4) Site-specific parameters; 5) Magnitude probability distribution parameters, such as the maximum magnitude; 6) Rupture area of seismic sources; 7) Relative location of hypocenter on the rupture surface; 8) Selection between different seismogenic source models; 9) Selection between different ground motion prediction equations; 10) Regression coefficients of such equations. Uncertainties of discrete nature among these, such as Cases 3, 8, and 9, are typically modeled in classical PSHA through logic trees. In contrast, continuous sources of uncertainty are typically either not considered or discretized due to the nature of classical methods. In addition, classical methods are not capable of modeling the correlation between the sources of uncertainty. In comparison, comprehensive modeling of uncertainties, whether aleatory or epistemic and discrete or continuous, as well as correlations between them, is an inherent ability of structural reliability methods. The present study first introduces a PSHA methodology called “scenario sampling,” which is based on the Monte Carlo sampling reliability method. Then, the probabilistic models needed for the methodology mentioned above are presented, and the software tools developed for PSHA using these models are introduced. Moreover, five of the above-mentioned sources of epistemic uncertainty—which are not generally considered in classical PSHAs—are quantified as random variables. Finally, as a case study, the PSHA of the Tehran province is carried out using the models, the methodology, and the software tools developed in this study. In this application, the importance of each source of epistemic uncertainty on the intensity measure for return periods of 475 and 2475 years is quantified and discussed. It is inferred from the results that the most critical source of epistemic uncertainty among all of the considered sources is the rupture area, which is usually deterministically modeled in classical methods. This source of uncertainty increases the peak ground acceleration, i.e., PGA, up to 15% and 27%, and by an average of 8% and 11%, for return periods of 475 and 2475 years, respectively. The results show that the cumulative effect of the five considered sources of epistemic uncertainty increases the PGA by 7% and 2% at the minimum, 24% and 27% at the maximum, and 12% and 14% on average, for return periods of 475 and 2475 years, respectively. Next, the effect of these sources of uncertainty on the uniform hazard spectra with the exceedance probability of 2%, 5%, 10%, 20%, and 50% in 50 years is discussed. The results indicate the need for revisiting the seismic hazard analysis that underlies the seismic zonation of the national seismic code of Iran
  9. Keywords:
  10. Probabilistic Seismic Hazard Analysis (PSHA) ; Epistemic Uncertainty ; Reliability ; Spectral Acceleration ; Intensity Measure ; Probability Model

 Digital Object List