Analysis of simplification errors for water distribution models

Information, data and maps required to analyze water distribution networks are commonly available nowadays. Furthermore, the use of information technology and geographical information systems has become part of normal mapping and management tasks in the water industry. In particular, improved descriptions of both network topology (number and position of the property connections along water mains) and its boundary conditions (e.g. demands, leakage, etc.) permit a more detailed analysis for different management purposes. These facts increase considerably the complexity of the network analysis and the size of the system hydraulic models and related mathematical problems. For example, the mathematical description of the actual demand pattern along a water main could be potentially achieved by adding one interior node for each connection point, but this will normally increase the size of the hydraulic model by more than one order of magnitude. Thus, simplification of the network model is vital for achieving an acceptable computational efficiency level that allows effective investigation. In this paper, network simplification is achieved by using the recently developed Enhanced Global Gradient Algorithm (EGGA). EGGA simplifies the network topology by automatically removing interior/serial nodes, which are those adjacent to two nodes only, thus reducing the total number of demand nodes and pipes in the analyzed hydraulic model whilst preserving original mass and energy balance. The EGGA formulation shows clearly that the sensitivity of the algorithm results to the uncertainties of the boundary conditions is unchanged with respect to classical GGA. Thus, this paper analyses two simplification strategies based on EGGA and applying them to a case study.