Water Network Model Calibration Based on Grouping Pipes with Similar Leakage and Roughness Estimates

A simple integrated approach to calibration of water network model and leakage distribution is proposed. The distribution of water losses is studied as the first step in the calibration procedure of the mathematical model. In order to calculate the leakage from pipe an empirical formula which takes into account the length of the pipe and mean pressure, and also contains certain coefficients, is used. For determination of leakage the pressure measured in control nodes is used. For defining pressure in other nodes an iteration process is applied which determines the leakage and pressure intermittently. Following that process the roughness values of the pipes are varied to minimize the differences between the calculated and measured pressures. Pipes with almost similar roughness are grouped together and corrections of roughness for each group are set up. This is followed by adjusting roughnesses with linear optimization process, where the increments of roughnesses are determined on condition that the sum of the squares of the differences between the simulated and measured pressure in control nodes will be minimal. A trial-and-error approach has been used for the comparison as well. The method was used by elaborating a software package for water distribution network of the City of Tallinn.