Author(s)

S. Mambretti

Abstract

In the paper a method for the optimization of water distributions networks is developed and applied to the case of the water supply network of Milano. This network is very complex and has no suspended reservoirs as the hydraulic head is maintained by the action of 31 pumping stations. Starting from real data, the operations of the entire network and pumping station, with the actual scheduling, are simulated with software using EPANET engine which was developed to this purpose. Afterwards, the operation of the pumping stations is optimized through the use of a simple Genetic Algorithm, in order to reduce the energy consumption maintaining a good service. Results show a significant improvement with the new working logics, thus justifying the study. Site tests show that the assumptions made in building the model are reliable, but they will probably require some adjustments following this research. Keywords: water distribution, Genetic Algorithms, optimization, energy saving. 1 Introduction Water distribution networks are a primary part of water supply systems, and since they represent one of the main infrastructure assets of the society, its management must be effective, efficient and energy saving. To this end, hydraulic constraints are not sufficient to find the best system and economic and energy conditions also have to be taken into account. For this kind of complex and multi-objective optimization problem Evolutionary Computation may be useful. The term Evolutionary Computation (EC) [1] represents a large spectrum of heuristic approaches to simulate evolution. The latter includes: Genetic Algorithms (GAs) [2, 3], Evolutionary Strategies [4, 5], Evolutionary

Keywords

water distribution, Genetic Algorithms, optimization, energy saving