Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Pays Bas → 2017 → Model-based optimization of water transmission and distribution networks [Case study : Ahero, Kenya]

UNESCO-IHE Institute for Water Education, Delft (2017)

Model-based optimization of water transmission and distribution networks [Case study : Ahero, Kenya]

Wambugu, Simon Kariuki

Titre : Model-based optimization of water transmission and distribution networks [Case study : Ahero, Kenya]

Auteur : Wambugu, Simon Kariuki

Université de soutenance : UNESCO-IHE Institute for Water Education, Delft

Grade : Master of Science (MS) 2017

Résumé
WDN design is an important part of any project investment. Water distribution systems are normally designed to satisfy consumer demands while being able to meet some performance criteria for instance ensuring minimum pressures and keeping water quality within the permitted limits. For many water engineers, however, developing an optimized water distribution network model has usually involved traditional methods of trial and error which have been found to be cumbersome and computationally long. Nowadays, many state of the art evolutionary algorithms, some which are freely available have been developed. These are designed to handle complex non-linear optimization problems much faster than humans, and yield near optimal solutions. This research presents two approaches to an optimal design of a new WDN using genetic algorithms. The case study is Ahero town network, which is a new design extension to an existing water distribution network of Kisumu city, Kenya. The main objective of the study was to provide a least cost alternative design of water transmission and distribution network between Kisumu city and Ahero town using EPANET coupled with GA and compare the results with WaterGEMS based optimal design. The efficiency of the two algorithms was then compared and overall, some knowledge and skills on GA-based optimization was gained from the study. For the EPANET-GA set up, however, cloud computing was utilized using SURFsara’s HPC environment due to the longer computational time which was required. To ensure flexibility in the design, four hydraulic design alternatives were developed and optimized separately for a least cost objective using two approaches. The first approach involved using EPANET coupled with Matlab-based GA for an optimal design of the pipe network and storage tank sizing. The result was a continuous diameter solution for pipe sizes. A comparison was then made using a fast messy GA which comes embedded as a tool (Darwin designer) in WaterGEMS modelling package. In this setup, the pipe network was optimized while taking the optimal tank sizes from EPANET-GA as fixed input boundary conditions.

Sujets  : water distribution networks hydraulic modelling genetic algorithm Kenya

Présentation

Page publiée le 2 avril 2021