Calibration of Water Distribution Hydraulic Models Considering Different Decision Variables and Consumption Scenarios

In recent years, hydraulic models are widely used to simulate hydraulic performance of water
distribution systems. They play a significant role in evaluation and operational management of these
systems. Calibration is necessary for improvement of model function and reduction of the
uncertainties during different consumption scenarios in the system. At the moment just a few
commercial models are capable of calibration calculation besides the hydraulic performance
evaluation. However, the popular and freely available model, EPANET can not be calibrated. In this
paper, an optimization procedure is proposed using Genetic Algorithm (GA) for calibration of the
hydraulic simulator. For this problem, the pipe roughness (Hazen Williams) coefficients, nodal
demands and pipe diameters are considered as decision variables of the optimization problem. The
procedure is implemented for four different consumption scenarios: normal, minimum, maximum and
fire flow. Finally, to evaluate the proposed methodology, a test network at different consumption
scenarios is calibrated and the results are compared. Comparison of the results shows that the best
result with minimum error has been obtained when the fire consumption with considering both
variables of Hazen-William coefficient and nodal demand was selected.