Document Type



Doctor of Philosophy


Civil Engineering

First Adviser

Paolo Bocchini


During a natural extreme event, such as an earthquake or a hurricane, the amount of socioeconomic losses due to inefficient disaster response or the losses due to long-term reduction of functionality of infrastructure systems is comparable, if not higher, to the immediate losses due to the extreme event itself.

Therefore, the scientific community has recognized the need to be able to accurately predict the performance of the lifelines and infrastructure when a extreme event occurs and also being able to build structures and infrastructures which are able to successfully withstand extreme events, effectively satisfy the needs of the post-event emergency response, and restore its functionality as soon as possible.

One of the first challenges in performing such tasks is to accurately assess the load of an extreme event to the geographic region the infrastructure or lifeline belongs in.

Hazard maps and probability of exceedance curves are very popular tools used initially for the probabilistic seismic hazard analysis and expanded later to other hazards such as hurricanes.

These tools provide the probability of exceeding any given value of an Intensity Measure (IM) of choice (e.g., 1-minute sustained wind speed ) at any location.

These tools are an integral part of the performance-based design approach and are essential for the probabilistic analysis of individual structures.

However, these tools are not appropriate for the analysis of distributed infrastructure systems because they do not account for the correlation information among the values of the IM at different locations.

Engineers have recognized that the various network components cannot be studied independently because the performance of the entire network depends on the combination of the conditions of all members. Thus, considering joint probabilities of having certain values of the IM at the locations of interest if required.

The most widely accepted approach by the scientific community to address these issues is through simulation-based techniques.

Based on this idea, a set of representative extreme scenarios is selected and then the impact on the network components and, in turn, the performance of the network itself is predicted.

The main issue in this approach is the computational cost, which constraints the number of extreme event scenarios to be as small as possible, while the set still captures the probabilistic characteristics of the intensity of the investigated natural extreme event over a region.

A new framework is presented for the selection of an optimal set of stochastic intensity measure maps representing the regional hazard over a geographic area.

This set of IM maps can subsequently be used for the analysis of spatially distributed infrastructure systems.

The proposed methodology results in a versatile multihazard tool that accounts for the spatial correlation through the optimal sampling of IM maps.

Its key characteristic is that it embraces the nature of the regional IM maps as two-dimensional random fields.

The representation of the regional hazard is supported by proofs of optimality, ensuring mean-square convergence of the ensemble of representative IM maps to the complete portfolio of possible hazard events, which is a particularly important property for risk analysis.

A detailed comparison of the proposed technique with other popular methodologies in the same filed is presented.

Before applying the proposed technique or any other hazard representation technique, it is necessary to accurately study and characterize the regional hazard in a probabilistic way.

Two types of natural phenomena were considered in the conducted research for the regional hazard analysis: the earthquake and the hurricane hazard.

In the case of earthquakes, the seismic characterization of the Charleston South Carolina region was studied and a seismic modeling procedure was developed which includes spatial and temporal information and descriptions of fault geometry and style as well as other parameters.

With the complete probabilistic description of the regional seismic hazard, the ground-motion prediction equations (GMPE) are implemented.

The GMPE account for the in between-earthquake and within-earthquake variability, and result the ground shaking acceleration over the region.

In the case of hurricanes, a more holistic approach was adopted by stochastically modeling the hurricane's track.

Historical hurricane events, originated either in the Atlantic basin, Caribbean Sea or the Gulf of Mexico, have shown to significantly affect geographic regions located in the South and Eastern U.S.

Therefore, a simulation framework is developed for the prediction of hurricane wind intensity and direction over any geographic region in the Southern and East U.S.

The proposed framework generates synthetic hurricane directional wind speeds, and does so by utilizing historical data, simulating the hurricane's track and intensity, simulating key characteristic parameters such as the central pressure and the radius to maximum wind among others both for offshore and overland locations of the track, performing a wind field analysis, calculating the 10-meter wind intensity by utilizing oceanic- and land-based boundary layer models and finally simulating offshore and overland wind directions.

Available for download on Thursday, August 26, 2021