Development of Multiple Leakage Detection Method for a Reservoir Pipeline Valve System

Abstract

The detection of multiple leakages in pipeline systems has been one of the challenging issues for the control of water loss in water distribution systems. Inverse transient analysis can be a useful principle for predicting leakage through the calibration of location and leakage quantity, based on the pressure reflection that originates from an abnormal boundary condition. In this study, an innovative leakage detection method is proposed to address unknown conditions on multiple leakage dimensions through introduction of revised leakage expressions based on a frequency domain approximation. A multiple leakage function was modified for an efficient representation of multiple abnormalities at a reservoir pipeline valve system. An iterative metaheuristic scheme (IMS) was designed to handle an optimization scheme for multiple leakages using a pressure response for a discharge impulse introduced through value manipulation. In order to address unsteady friction in hydraulic transients combined with multiple leakages, both one-dimensional and two-dimensional models were used to derive leakage expressions for turbulent and laminar flow conditions. An isolated multiple leakage function (IMLF) was proposed to exclusively encapsulate the impact of leakages and unsteady fiction. Considering uncertainties in the hydraulic transient propagation, data noise, and multiple local optima issues in large parameter calibrations, three advanced schemes were modularized to improve detectability of IMS. Several hypothetical examples were presented to show the potential of IMS, validity of three advanced schemes, and robustness in multiple leakage prediction compared to existing approaches.