Programme  OS1k IWRM and water allocation  abstract 721

The Use of Multi-objective Optimization for Reservoir’s System Operation with an Evolutionary Algorithm: The case of the Metropolitan Region of Fortaleza, Brazil

Author(s): Francisco Venícius Fernandes Barros, Luiz Sérgio Vasconcelos do Nascimento, Eduardo Sávio Passos Rodrigues Martins, Dirceu Silveira Reis Junior
Researcher of Fundação Cearense de Meteorologia e Recursos Hídricos-FUNCEME (luizsergiovn@gmail.com), Researcher of FUNCEME (venicius@gmail.com), President of FUNCEME (esm9@secrel.com.br, Researcher of FUNCEME (dirceu.reis@gmail.com, phone: +55 85 3101-10

Keyword(s): reservoir operation, multi-objective optimization, evolutionary algorithms

Article: abs721_article.pdf
Poster:
Get Adobe Reader

Session: OS1k IWRM and water allocation
AbstractThe Metropolitan Region of Fortaleza, with its 2.5

million inhabitants, is located in the state of Ceara, a semi-arid region of Brazil, and obtains water for domestic and

industrial purposes through the use of a complex system of reservoirs, which are linked by canals and water pump

stations, bringing water from different basins within the state. The system consists of 5 reservoirs, 5 pumping stations,

and a long canal (102 km) that diverts water from the Jaguaribe River basin, a large and agricultural basin in which

the largest reservoirs of the State, with inter-annual storage capacity, are located. The system is operated to meet

domestic and industrial demands of the Metropolitan Region of Fortaleza. Current operating policy employed by the

water management agency of the state, responsible for the system operation, is based on a relatively simple set of

rules bused upon current water storage in each reservoir. This paper focuses on the optimization of the reservoir’s

system operation, based upon a multi-objective approach, to derive new operating policies for the system. The

multi-objective optimization procedure employs goals that are often considered in the water management process,

namely, minimizing both the pumping cost and the amount of water losses through evaporation. The latter is justified

by the extremely large potential evaporation rate observed in the region and the relatively different area-volume

curves of these reservoirs. The paper develops and employs a new multi-objective version, based on the Pareto

dominance concept, of the single-objective evolutionary algorithm Honey-Bee Mating Optimization (HBMO) and

the Multi-objective Particle Swarm Optimization (MOPSO). Results based on a 25 year-validation period show that

the use of a new operation policy derived in this study, based on a possible solution of the Pareto front, provides an

economy of up to 8% in pumping costs (minimum cost) and a reduction of water losses through evaporation of 16%

at most (minimum evaporation loss). Moreover, the methodology provides an approximation of the Pareto front of

both objectives, which permits water managers to think more deeply about the value of water that is evaporated and

the costs of trying to avoid these losses.

  Return up