Systematic exploration of pipeline network calibration using transients

Accurate information of pipeline and system properties is crucially important for precise computer simulation of pipe networks. Even though inverse transient analysis (ITA) techniques have been widely investigated for leak detection and friction factor calibration, many challenges still remain. One reason for these difficulties is that real water distribution systems invariably have many kinds of uncertainties including pipe diameter, wave speed and the water demand at the time of the tests. This paper investigates quantitatively how inaccuracies in such values invariably deteriorate the performance of calibration approaches. Thus, the paper argues that a systematic calibration should explicitly include these additional uncertainties during the ITA process. Two evolutionary approaches, namely Genetic Algorithms and Particle Swarm Optimization, are applied and are both compared and contrasted during the ITA iterations. The evolutionary algorithms help the search escape from poor local optima in multifaceted and complex problems and thus assist in locating global (or near-global) optima. However, most current approaches are shown to converge poorly as the full scale of the typical field problems is progressively reflected in the search space.