An important component of any dam safety program is the dam breach analysis and resulting flood inundation maps critical for assisting emergency management agencies in minimizing loss of life. Although a dam breach analysis follows a prescribed deterministic methodology, it includes a significant degree of uncertainty. This uncertainty can lead to being overly conservative and does not provide an understanding of the critical magnitude and areas of risk. Due to the high degree of uncertainty surrounding the size, shape, and formation time of a hypothetical breach of Camanche Main Dam and Dikes and potential magnitudes of these breaches, East Bay Municipal Utility District (EBMUD) decided to perform a probabilistic dam breach analysis in addition to traditional deterministic methods. Camanche Reservoir is owned and operated by EBMUD and located on the Mokelumne River in the foothills of the Sierra Nevada Mountains near Lodi, California. The reservoir, formed by the main dam on the Mokelumne River and six auxiliary dikes along the northern and southern ends of the reservoir, is primarily operated to provide flood control, meet downstream in-stream flow and water rights obligations, provide recreational services, and generate hydroelectric power. The main dam and dikes consist of zoned compacted earthfill structures with heights ranging from 40 to 170 feet and impounded reservoir volumes of 40,000 to 430,000 acre-feet. This paper discusses the analysis performed for one of the southern dikes, Dike 2, including the methods used for setting up and forming the statistical distributions for the input breach parameters, developing and calibrating the HEC-RAS 2D model, and the process of running the Monte Carlo simulation to determine the distribution and probabilities of peak breach flow rates. Results from the probabilistic analysis of Dike 2 helped provide great insight on several key factors that were overlooked during the deterministic analysis of the dikes. The findings of this robust approach for dam breach analysis highlights a powerful application of the probabilistic approach in being able to inform the results from a deterministic study.
