Application of Genetic Algorithm for Determining the Existent Friction Coefficients in the Real-life Pipe Networks as an Inverse Problem

An integrated approach for determining the existent friction coefficients in water networks is
proposed. As pressurized systems age, the carrying capacity of network decreases because the internal
roughness increases with the aging of pipes. This can lead to loss of satisfactory performance and
uneconomic operation. Therefore, rehabilitation of an existing network becomes very important
problem in water industry. Determining the real physical characteristics of pipes for network analysis
is a regular component of the rehabilitation process. Pipe friction coefficients cannot be determined
explicitly by direct measurement, they are determined implicitly, as an inverse problem, from
measured model outputs (pressures). Values of friction coefficients are determined in a way that they
should yield a reasonable match between measured and predicated pressures in the network. One
problem associated with the Re-calibration of real-life pipe networks is the lack of field measurements,
which can sometimes, lead to the formulation of an ill-posed inverse problem. In this study certain
methods have been utilized to tackle this problem. The hydraulic analysis of steady and quasi-steady
flow and optimization process are combined to develop a program. By selecting a proper optimization
method (Genetic Algorithm) the inverse model was developed and verified successfully in a case
study.