Date of Graduation


Document Type



A multi-phased research program consisted of performance of both laboratory and field test programs, development of hydraulic conductivity probability density functions, and development of a first generation risk model to assess relative risk among groups of levees to determine rank for systemic allocation of scarce rehabilitation resources. The laboratory test program consisted of determining basic soil properties at four USACE WES REMR demonstration sites: Talulla, MS, Sny Island, IL, Skagit, WA, and Elkhorn, CA. Additionally, flexible wall hydraulic conductivity tests were performed with subsequent development of hydraulic conductivity probability density functions (PDF's). The field testing program consisted of conducting in situ two-stage borehole (TSB) tests and developing PDF's. Goodness of fit tests were performed against 21 distributions to develop rank order of best fit distributions. Results varied for the sites with inverse gaussian, lognormal, pearson, weibull, and beta consistently providing the highest ranking fits for hydraulic conductivity. Best fit distributions for the log of hydraulic conductivity were evaluated as extreme value, weibull, normal, logistic, beta, and gamma. Analysis and design of levees is a complex procedure in which multi-modal failures should be addressed. Development of a risk tree model for levees allows for the performance of the decision analysis to describe the chance of failure while incorporating the uncertainty in the analysis parameters. A first generation risk based model was developed to assess uncertainties in a quantitative manner. The analysis risk model presented combines the traditional probability analysis and the risk tree approach to evaluate risk of failure. The uncertain risk of failure is modeled through estimating the probability of deviation from design assumptions. Items contributing to uncertainty and describing the multi-modal aspects in the design of levees are grouped into four main categories: embankment instability, seepage instability, geometric uncertainty, and flood event. Accordingly, the relative risk of groups of levees can be evaluated for a rational analysis/design/rehabilitation decision process. The risk model was applied to six case scenarios for demonstration purposes. Results of the analysis yielded relative risk levels from which rehabilitation decisions may be systemically developed.