Propuesta para una metodología de sectorización de redes de abastecimiento de agua potable
[EN] Drinking water is an indispensable resource for all processes involved in life. The problems of water supply networks (WSN) can be summarized in four broad areas: leakage and non-revenue water; physical integrity of the network; water quality and reliability, and quality of the databases associ...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2013 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | español |
| OAI Identifier: | oai:riunet.upv.es:10251/39139 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/39139 |
| Access Level: | acceso abierto |
| Palabra clave: | Sectorizacion Fugas Algoritmo Sectorization Leakage Algorithm MATEMATICA APLICADA INGENIERIA HIDRAULICA Máster Universitario en Ingeniería Hidráulica y Medio Ambiente-Màster Universitari en Enginyeria Hidràulica i Medi Ambient |
| Sumario: | [EN] Drinking water is an indispensable resource for all processes involved in life. The problems of water supply networks (WSN) can be summarized in four broad areas: leakage and non-revenue water; physical integrity of the network; water quality and reliability, and quality of the databases associated to water distribution systems. In particular, leakage may become of the order of 50% of water injected in a WSN. Today, networks without any water losses would be considered utopian, given the technical and economic implications that this would imply. However, there have been important advances in the understanding and development of equipment and techniques that allow better comprehension and management of leakages. Among these techniques, it is important to highlight the segmentation, which is considered a strategic option which involves the subdivision of small sub networks by closing valves and installing flow meters. One of the major benefits of its implementation is the increased ease at which any abnormality is detected within the subsector due to the dimensional reduction. In most cases, sectorization processes are implemented with great lack of scientific rigor. On the contrary, sectorization is usually based on trial-and-error procedures. In smaller networks, this type of approach is usually straightforward. The problem is the proper definition of sectors in larger networks, where, given the large amount of information associated with the WSN, the implementation of a process of this nature, without the support of tools, would not be feasible. The main objective of the thesis is to establish a computational procedure for obtaining a layout of sectorization of a WSN. In this layout, the network is split into trunk and distribution network. The process not only uses traditional hydraulic or geographical features of RDAP as criteria, but also takes into consideration the leaks into the network. The process involves the next steps: (1). Define within the WSN the trunk and the lines that form the distribution network. This step is completed based on the diameter of the pipes, setting a minimum diameter range to distinguish between trunk and distribution network; (2). Estimate the number of sectors that should have the distribution network in question, by means of a computer technique of cluster analysis; (3). Define the number of microsectors calculated in the previous step, using spectral clustering; (4). Define the entrance (of water) to the sectors based on energy criteria; (5). Validation of the sectorization obtained (through mathematical modeling). Segregation of the distribution network and the trunk helps to conserve the flexibility of the network in the event that, in the future, the sectoring scheme changes, and also to save costs by installing valves and flow meters of large diameter. This work demonstrates the applicability of Machine Learning to tackle the proposed task. By using hierarchical clustering a number of sectors with the highest degree of homogeneity regarding their characteristics are found. Then spectral clustering is able to improve the results of hierarchical clustering, finding a partition while maintaining consistency of the characteristics of each of the sectors, and minimizing the number of valves to be used to make the partition. The use of indicators of energy dissipation through the network enables to find water entrances to each sector so as to minimize the energy dissipated by the network, and ensures the highest possible pressure at the consumers’ nodes. Following this proposal, It was managed to get an example network sectioned (100 km network divided into three sectors) that maintains the pressure within the ranges established as appropriated, and decreases the level of leakage just by its implementation |
|---|