School of Civil and Environmental Engineering

Ph.D. Thesis Defense Announcement

Stochastic Approaches for Quantification of Near- and Long-Term Structural Reliability

By

Ajay Saini

Advisor:

Dr. Iris Tien (CEE)

Committee Members:

Dr. Lauren Stewart (CEE), Dr. Yang Wang (CEE), Dr. Lauren Linderman (Civil Engineering, University of Minnesota), Dr. Kamran Paynabar (ISYE)

Date & Time: Friday, November 1st, at 4:30 pm

Location: Mason Education, Room 2119

The quantification of structural reliability is an important problem in civil engineering, affecting decisions in design, maintenance, retrofit, and rehabilitation of structures. As structural monitoring data increases, there is the desire to use this data to better estimate and predict the performance of structures, both under extreme loadings such as earthquakes, and over longer time horizons. The first part of this thesis concentrates on estimating and predicting the near-term reliability of structures under earthquake loading. We propose a methodology based on dynamic Bayesian networks and Kalman filter estimators to utilize building-mounted accelerometer data in real time to estimate the maximum nonlinear response of a structure and quantify the reliability in terms of the distribution of the maximum response under the earthquake. We extend this methodology from real-time estimation to prediction to predict the maximum structural response of a structure for an impending earthquake. We develop a time-sensitive, computationally efficient, and sufficiently accurate methodology to predict the maximum response of the structure. We verify the methodologies for the estimation and prediction of maximum response based on experimental data from laboratory tests and show close correspondence between the experimental and theoretical results.
The second part of this thesis focuses on the quantification of long-term structural reliability. We propose a stochastic degradation model for structures and evaluate the effect of environmental and climate parameters on structural reliability based on changes in the rate of degradation and occurrence frequency and intensity of loadings. Finally, we propose a method to predict reliability to optimize the repair of a structure over its lifetime. We use the reliability and repair projections to quantify the structural resilience over longer time horizons. The projected resilience is used to analyze the performance of both an individual structure and a network of structures after a shock event.