Study of calibration of water supply networks through genetic algorithms

Author(s)

Fernando das Graças Braga da Silva

• Document type: Doctoral Thesis
• Press: São Carlos.
• Institution: Universidade de São Paulo (USP). Escola de Engenharia de São Carlos (EESC/SBD)
• Defense date: 2003-03-27

Examining board members

Luisa Fernanda Ribeiro Reis; Hans George Arens; Paulo Sérgio Franco Barbosa; Mario Thadeu Leme de Barros; Edevar Luvizotto Junior

• Advisor: Luisa Fernanda Ribeiro Reis

Abstract

Continuous physical changes that occur in water supply networks along the years require their calibration in order to guarantee realistic forecasts through hydraulic models. There are many difficulties involved in water supply network calibration in terms of their parameters as results of many factors, amongst which the lack or uncertainty on data like location and status of valve openings in the network. Several calibration methods have been proposed in the literature, usually based on many analytical equations and optimization techniques. However, it was known the need of estimative of network parameters in proper form of roughnesses and local parameters of relation between pressure and leakage, for systems submitted to expressive leakages. This work proposes a method of calibration in two phases in which the GAs (genetic algorithms) are used as tools in the resolution of corresponding inverse problems, being the studies performed for sectors of the network of the city of São Carlos, SP, about which extensive field work had been done. It was developed, also, studies investigating of the best representation of problem through GAs and the localization of optimal points for monitoring. The results have showed themselves robust in the identification of roughnesses and sections having valves, as well as the parameters of leakage model, besides to indicate possibilities of improvements in GA\'s efficiency. The results that have been obtained from considerations aboutlocalization techniques of optimal points for monitoring were considered good as well. (AU)