Many-Objective Analysis to Optimize Pumping and Releases in Multireservoir Water Supply Network


Abstract

This study contributes a many-objective analysis of water supply in the Tarrant Regional Water District (TRWD) in Texas in the United States, in order to demonstrate how multiobjective evolutionary algorithm (MOEA) decision support can be used with sophisticated water infrastructure models in a highly constrained water-planning environment. A detailed representation of TRWD's network in the river system modeling tool RiverWare (a model actually used by the utility) is coupled with the Borg MOEA to solve a seven-objective problem formulation. Objectives within the formulation include systemwide performance objectives, as well as reliability objectives to balance multiple reservoirs, calculated using two stochastic hydrology scenarios. The results provide valuable information to TRWD that may inform the utility's approach to optimization and decision making in the future, as well as illuminate the benefits of optimizing systemwide and component-specific objectives to reveal hidden system dynamics.