Associate Chair for Graduate Programs.

School of Civil and Environmental Engineering

Ph.D. Thesis Defense Announcement

Leveraging Post-Disaster Data for Rapid Damage Assessment of Infrastructure Systems

By: Jorge Mario Lozano

Advisor:

Dr. Iris Tien

Committee Members:  Dr. J. David Frost (CEE), Dr. Mathieu Dahan (ISYE),
Dr. Jack W. Baker (Stanford University), David J. Wald (USGS)

Date and Time:  July 18th, 2023 11:00 am (ET)

Location: SEB 122 & Zoom: https://gatech.zoom.us/j/91863828744

Recent impacts to infrastructure systems after disasters are evidence of their extreme vulnerability to large earthquakes and hurricanes. Systems such as buildings, roads, and pipelines, enable other aspects of society to operate effectively, making post-disaster damage assessment one of the most critical emergency tasks in the aftermath of a disaster. However, the process of performing detailed and comprehensive damage assessment is still expensive and laborious. More specifically, there are three specific problems identified and addressed in this work. First, post-disaster damage assessment is often analyzed at a single point in time, rather than as a dynamically evolving process, limiting the ability to provide a temporal assessment of the evolution of data and post-disaster information availability over time. Second, evolving seismic uncertainty used in building damage assessments is typically treated as a static value instead of a dynamic variable with estimates that change throughout the post-disaster response depending on the level of information. And third, there is the lack of implementation and understanding of the benefits of multi-sensor networks for rapid flood damage assessment. 

As a response to these problems, this work is distributed in three sections that tackle each of these three problems. First, is a comprehensive analysis of the post-disaster data collection process, leading to findings on the availability of tools and datasets for building and lifeline damage assessment after disasters. A detailed analysis across disaster events is presented that identifies three stages of the data collection process, corresponding characteristics of the data available, and essential metadata features for damage data to increase its integrability and shareability. Second, is a new methodology to incorporate the evolution of seismic uncertainty values in the building damage estimates of two methods: the Thiel Zsutty method (TZR) and FEMA’s Hazus. This simulation-based methodology produces an updated damage state distribution for a building in less than 3 seconds, demonstrating its usefulness for rapid building damage assessment accounting for evolving seismic uncertainty parameters. Third, is a methodology to use real time data from a sea level sensor network for rapid flood damage assessment. The data from the sensors is used for estimating damage to critical infrastructure and as the basis for recommending new locations of sensors to expand the network. Compared to previous work focusing only on network coverage and uncertainty, this work results in potential locations that also consider novel performance metrics related to flood risk, including distance to critical infrastructure, flood likelihood, and community vulnerability. The contributions of this work help entities interested in post-disaster damage assessment to increase data usability and shareability, decrease the uncertainty of build