Hydraulic modeling of water supply network of green spaces using EPANET and prediction of hydraulic characteristics using artificial intelligence

The occurrence of a problem in each of the water supply network sections due to pressure or velocity fluctuations can cause disruptions in consumers’ regular life. To help avoid these problems, proper design and optimal management of the network is very important. In this study, control of water pressure and velocity to prevent problems in the water supply network is investigated, and hydraulic flow characteristics in pipes are predicted by artificial neural network. In this regard, first, by zoning of Kangavar city in Kermanshah province (as a case study), six zones were identified, based on distribution parameters, the water supply network for the green space of the city, for 10-year plan and target population of 95000, according to working pattern of 22 hours per day and per capita green space of 29.6 m2 at the end of design period was drawn. Then, EPANET software was used to analyze the pressure, velocity and flow in the pipe network. Based on the results, maximum pressure occurred in the 3-3 joint in the third pressure zone, which was about 100 m of water, and maximum velocity in the network was about 1.4 m/s. Also, results showed that the flow rate used for the network is due to the diameter of the pipes and selected paths in different zones in the appropriate range. Subsequently, artificial neural network was trained using the available quantities and the optimal network was selected with a correlation coefficient of 0.87 and 0.85, respectively, for training and testing phases, respectively. Then, the flow velocity and pipe friction-loss were predicted by the optimal network. Results indicated high potential of artificial neural network in analyzing and predicting hydraulic characteristics of water pipe networks.