Skip to main content
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Risk Assessment and Management for Interconnected and Interactive Critical Flood Defense Systems


The current State-of-the-Practice relies heavily in the deterministic characterization and assessment of performance of civil engineering infrastructure. In particular, flood defense systems, such as levees, have been evaluated within the context of Factor of Safety where the capacity of the system is compared with the expected demand. Uncertainty associated with the capacity and demand render deterministic modeling inaccurate. In particular, two structures with the same Factor of Safety can have vastly different probabilities of failure. While efforts have been made to assess levee vulnerability, results from these more traditional engineering approaches are questionable because they do not more fully account for uncertainties included in modeling, natural variability, or human and organization factors.

This study develops and documents a probabilistic Risk Assessment Methodology that explicitly addresses levee resilience and sustainability by explicitly incorporating uncertainty in the Capacity and Demand components. In this research, we have categorized uncertainties into four different categories: Type I- Inherent (or aleatory) variability; Type II- Analytical/ Model (epistemic) variability; Type III- Human and Organizational Performance Uncertainty; and Type IV- Knowledge integration uncertainty.

The complete infrastructure system in the Delta is very complex with many components integrally correlated. These include large-scale water supplies that supply over 20 million residents; a flood protection and levee system that past research has shown to be at great risk; an electricity transmission grid key to California and western North America; and a multimodal transportation system (roads, rail and shipping) that extends throughout the Pacific Rim. Delta's levees are among of the most unstable engineering systems, with several major hazards threatening the stability of the approximately 1100 miles of its levees. Flood, sea level rise, and aging infrastructure all contribute to this risk. It is this potential levee failure that could cause the greatest damage, particularly with respect to the security of freshwater exports.

This thesis validates the proposed methodology by evaluating the probability of failure for an interconnected flood defense system in the California Sacramento-San Joaquin Delta. The study focuses on the behavior of the levee system protecting Sherman Island. Sherman island is of critical importance to California because of the critical infrastructures that pass under, on and over it, including: natural gas pipelines: regional and inter-regional electricity transmission lines; two deepwater shipping channels that run alongside the island; and the presence of State Highway 160, a link between major expressways. The work evaluates current (year 2010) and future conditions (year 2100) and incorporates variations in capacity and demand arising from human activities and global climate change. Specifically, the work evaluates the uncertainties for three potential failure modes: underseepage, slope (or levee) instability and overtopping/erosion through the use of Monte Carlo simulations that correctly capture the probability distribution of capacity and demand measures.

Finally, the work incorporates Human and Organizational Factors including interconnections and uncertainties into the Risk Assessment Model as they account for the largest contribution of major engineered system failures.

With this approach, probability of failure was determined and uncertainties were explicitly stated in every step of the method. By doing so decision makers and engineers can quickly identify where the uncertainty lies and decrease the probability of failure by increasing their understanding of the engineered system.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View