Planning and Operation of a Multi-Reservoir Water Distribution System

A water supply distribution system planning model (DSPM) is developed using a directed graph algorithm as its pre- and post-processors and a linear programming (LP) procedure as an `intelligent' system operator at its core. The directed graph is used to produce a mathematical representation of the distribution system. It is also used to check the consistency and correctness of the network topology before an input file is created for the LP procedure. In the post-processing mode, the directed graph generates files for graphical displays of the model results and performs graph analyses such as depth- and breadth-first traversals (i.e., water quality blend analysis, tracking of source-to-demand contributions, and others). Unlike the conventional formulation of a resource allocation problem, the demand requirements are removed from the constraint set and translated into shortages (to be minimized) as a component of a multiobjective function. Other components include uneven distribution of shortages (minimized), system storage (maximized), and total spillage (minimized). This formulation not only circumvents the unnecessary traps of infeasible solutions, but also allows the use of differential weights to control system performances so that they are reflective of the preference of the various objective components. The methodology was successfully applied to the Metropolitan Water District (MWD) of Southern California's water distribution system.