Estimation of Hydraulic Pressure in Water Networks Using Artificial Neural Networks and Fuzzy Logic

Hydraulic pressure is one of the most significant parameters in optimizing water distribution networks. Its simulation and estimation are essential tools in water distribution network management due to the significant effects it has on certain parameters of these networks. As a result of water losses to leakage, not all the inflow to urban water networks is consumed. Water leakage results in losses of supply, pressure, and capital investment. It also has adverse effects on water transfer capability, water treatment, and other elements in the distribution process. Furthermore, water quality problems could result from pollution at leak points. It is, therefore, necessary to estimate the amount of leakage at each point as a function of pressure.In this paper, artificial neural network as a powerful and flexible mathematical tool is used to model pressure estimation based on reservoir head, node elevation, water consumption, and the amount of leakage at a given point. Part of Tehran metropolitan water distribution system is modeled and the EPANET2.0 software is used to estimate the pressure variations in the network. Two different artificial neural network models, namely, a multi-layered ANN and a fuzzy logic neural network (ANFIS) are used for this purpose. The results are analyzed and compared with those from EPANET.