Methods for probabilistic seismic levee system reliability analysis

Risk assessment of levee systems involves analysis of the probability of whether the demand (e.g., high water events, earthquakes, waves) exceeds the capacity (e.g., freeboard, erodibility, liquefaction susceptibility) for at least one portion of the system. We present a seismic levee system reliability analysis framework, drawing upon related research experiences from the U.S. and the Netherlands. We first suggest definitions for some of the terminology utilized in levee system analysis, including "system", "reach", "section", and "characteristic length". These definitions bear specific meaning in the context of system risk analysis. We then consider two methods for evaluating spatial correlations and distributions of demand and capacity within reaches: Monte-Carlo simulation and a first order approximation. We present results evaluated using both methods for an example levee system subjected to spatially variable ground motions arising from a scenario earthquake event. Monte-Carlo simulation can consider varying demands within a reach, but the calculation is computationally intense for large system. On the other hand, the first order approximation is efficient to deal with a large system but is less accurate when the limit state function varies over short distances.