Optimal Design of a Flood Protection Levee

Abstract

In this chapter we illustrate the application of the LQI method to select a structure that is optimal. Dams, dikes, and levees have traditionally been designed to a fuzzy quantity called the PMF (“probable” maximum flood). Proper probabilistic design is preferable, but it requires that hydrological data be translated into a local flood probability distribution. It is inadequate merely to do a conventional estimation of the distribution, since the application is to a unique location (a single sample realization of a time series) rather than a statistical population; the application is monoscopic in the sense of Matheron (1989). Estimation is a process that always introduces additional information, going beyond the facts. A tool to minimize this contaminating information, the method of relative entropy with fractile (or quantile) constraints (REF) has a practical and simple approximation described and illustrated here. The societal capacity to commit resources (SCCR), is used as the design criterion. Details of financing have an important influence on the design of civil engineering facilities by socio-economic optimization, including flood control projects. Since future life risk must be discounted like finances, the interest rate and the amortization period influence designs decisively. These aspects are all brought out in the example of a city protected by a levee.