Simulation and component analysis of Water Distribution Networks

New tasks and paradigms are emerging in water system analysis, because of the availability of information technology and increased amount of support data for building water distribution system models. Water network models require the topological representation of the hydraulic system, which is represented by a graph linking pipes (arcs) and nodes (edges). With the tendency to create all-mains models from asset data and GIS, the dimension of the topological representation of the water network increases as a detailed description of the hydraulic system (e.g. the exact location of demands) is readily available. For such large water systems, the dimension of the topological representation could increase by at least one order of magnitude while considering customer connections (i.e., demands) along water mains. That topological dimension is very important for efficiency and efficacy of simulation models. For example, the global gradient algorithm, the most commonly used in today’s software packages, is based on the iterative solution of a sparse symmetrical linear system whose size is equal to the number of nodes in the water system’s topological representation. The increased topological dimension has two main disadvantages: (i) the numerical problem related to the solution of the linear system grows in size; (ii) the computational efficiency and efficacy decreases, e.g., which is critical for real-time analyses, optimal design and rehabilitation purposes. This work proposes a reduction strategy that when applied to the topological representation enables a detailed hydraulic analysis of the water system without forfeiting the internal energy and mass balance. This makes it easier to examine the main properties of the network topology and simplifies the system analysis.