Application of Mixed-Integer Programming for Flood Control in the Sacramento Valley: Insights & Limitations

TitleApplication of Mixed-Integer Programming for Flood Control in the Sacramento Valley: Insights & Limitations
Publication TypeThesis
Year of Publication1999
AuthorsJones, D. J.
Academic DepartmentEngineering
DegreeMS
Number of Pages111
Date Published1999
UniversityUniversity of California
CityDavis
Keywordsfloodplains, Sacramento Valley
Abstract This report presents results from an optimization study of the Sacramento Basin flood control system using the Hydrologic Engineering Center’s flood control optimization software, HEC-FCLP. The objective of this study is to determine whether significant benefits might be realized from an integrated operation of the system. To do this, a deterministic mixed-integer program (MIP) is developed and applied to the 1995 and 1997 flood events. A MIP model, rather than a linear programming (LP) model, is used to allow a more accurate representation of non-convex constraint sets. The objective of the model is to minimize damage throughout the system by deciding what releases should be from each reservoir during each time step of the analysis. For this study a 6-hour time step is used. Penalties are incurred for exceeding certain defined storage and flow levels or for exceeding the change-in-release constraints. Results of this study show that when incremental inflows to the system are high, Shasta Dam has an appreciable effect only as far downstream as the Bend Bridge gaging station and the Feather/Yuba River system consisting of Oroville Dam and New Bullards Bar Dam has an appreciable effect only as far as the Nicolaus gaging station. The results imply that these subsystems could be optimized separately from the complete system under these conditions. This study illustrates that MIP is a useful tool for flood control optimization. However, it is also found that solving complex systems using MIP can lead to excessive computation times. Simplifications must be implemented whenever practical to reduce the number of binary variables used by the model.
URLhttp://cee.engr.ucdavis.edu/faculty/lund/students/DustinJonesThesis.pdf