Urban water networks usually follow a highly interconnected scheme, which allows very variable flows. Effect of this scheme is a high topological redundancy, enabling the system to overcome various failure conditions. Looped networks are though more difficult to control; thus, it is necessary to subdivide the network into smaller areas, called District Metered Areas (DMAs). The DMAs are realized through a series of valves and flow meters, with the aim to obtain a precise control volume in a delimited area. Furthermore, the DMAs also allow pressure regulation, usually through a rescheduling of pump pattern. Historically, DMA design has followed a heuristic approach. Recently, new procedures have been proposed using elements from genetic algorithms, graph theory and even multi-objective approaches. In this thesis a new methodology aimed at designing DMAs in highly interconnected supply WDN with multiple sources will be proposed. It is based on 2 steps: first, the particle tracing analysis in order to divide the network into elementary DMAs (eDMAs), which are independent from each other and supplied by a station each. Since eDMAs are usually not feasible in real networks, due to economical and performance issues, they need to be grouped into bigger districts called macro-DMAs (mDMAs). For this reason, we will then present a new grouping procedure based on a set of preliminary analysis, with the aim of characterizing the hydraulic behavior of the network. The creation of this mDMAs also requires a pump rescheduling, which in general allows to reduce both energy consumption of the pumps and average pressure. Finally, the algorithms are applied to the real water distribution system of Milan, Italy. Results show that the proposed methodology gives very positive outcomes, with a good reduction of energy consumption and leakage rates without compromising the general performance of the network.
Le reti idriche urbane seguono tipicamente uno schema interconnesso, con tubi in cui la portata può fluire in più di una direzione e nodi che possono essere raggiunti per mezzo di svariate combinazioni di percorsi. Tale schema aumenta la resilienza del sistema, in modo che possa sopportare agevolmente condizioni di emergenza. Lo svantaggio di tali reti è che non permettono un elevato grado di controllo: per questo motivo è nata la necessità di suddividerle in parti più piccole, le Aree Distrettualizzate Misurate (DMA), create mediante l’installazione di saracinesche o la disconnessione dei tubi. Lo scopo è identificare un volume di controllo preciso per identificare più facilmente le perdite. Inoltre, le DMA permettono il controllo delle pressioni, solitamente attuato per mezzo di una riprogrammazione dei pattern delle pompe. Storicamente, la progettazione di DMA seguiva un approccio empirico. Di recente, sono stati introdotti metodi basati su algoritmi genetici, teoria dei grafi e algoritmi multi-obiettivi. In questa tesi verrà presentata una nuova metodologia per la creazione di DMA in reti interconnesse e con sorgenti multiple. Essa è basata su due step: innanzitutto, il tracciamento delle particelle d'acqua immesse in rete, per creare una serie di aree elementari (eDMA) completamente indipendenti l'una dall'altra e rifornite ognuna da una singola fonte. Siccome però la realizzazione nella rete reale di tali eDMA non è conveniente, a causa degli elevati costi di realizzazione e di un significativo calo delle performance della rete, le eDMA devono essere unite per formare macro-distretti (mDMA). Viene quindi successivamente presentata una nuova procedura di unione, basata su una serie di analisi preliminari aventi lo scopo di caratterizzare il comportamento idraulico della rete. La creazione dei mDMA ha richiesto inoltre una riprogrammazione dei pattern delle pompe. Per dimostrare l'efficacia degli algoritmi proposti, essi sono stati applicati alla rete di Milano. I risultati sono molto incoraggianti, poiché consentono una riduzione sia dell'energia consumata sia delle perdite senza però compromettere la resilienza totale della rete.
District metered areas identification in water distribution networks by coupling of graph theory and heuristics optimization
SCUFFI, ALESSANDRO
2015/2016
Abstract
Urban water networks usually follow a highly interconnected scheme, which allows very variable flows. Effect of this scheme is a high topological redundancy, enabling the system to overcome various failure conditions. Looped networks are though more difficult to control; thus, it is necessary to subdivide the network into smaller areas, called District Metered Areas (DMAs). The DMAs are realized through a series of valves and flow meters, with the aim to obtain a precise control volume in a delimited area. Furthermore, the DMAs also allow pressure regulation, usually through a rescheduling of pump pattern. Historically, DMA design has followed a heuristic approach. Recently, new procedures have been proposed using elements from genetic algorithms, graph theory and even multi-objective approaches. In this thesis a new methodology aimed at designing DMAs in highly interconnected supply WDN with multiple sources will be proposed. It is based on 2 steps: first, the particle tracing analysis in order to divide the network into elementary DMAs (eDMAs), which are independent from each other and supplied by a station each. Since eDMAs are usually not feasible in real networks, due to economical and performance issues, they need to be grouped into bigger districts called macro-DMAs (mDMAs). For this reason, we will then present a new grouping procedure based on a set of preliminary analysis, with the aim of characterizing the hydraulic behavior of the network. The creation of this mDMAs also requires a pump rescheduling, which in general allows to reduce both energy consumption of the pumps and average pressure. Finally, the algorithms are applied to the real water distribution system of Milan, Italy. Results show that the proposed methodology gives very positive outcomes, with a good reduction of energy consumption and leakage rates without compromising the general performance of the network.File | Dimensione | Formato | |
---|---|---|---|
2016_12_Scuffi.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
77.87 MB
Formato
Adobe PDF
|
77.87 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/131468