Diagnosis of pipe systems by means of a stochastic successive linear estimator

Environmental concerns and legislation of the water industry have recently drawn the attention of many researchers to the management, calibration and power cost reduction of water pipeline systems. Effective water system management rests upon the knowledge of the current state of a water pipeline system. For example, leakages, unknown status of in-line valves, and partial blockages are often a source of management costs. These anomalies must be detected and corrected as early as possible. In this paper a diagnosis tool is presented able to detect leaks, partially closed in-line valves and partial blockages by means of transient head measurements. The algorithm introduced is a Stochastic successive Linear Estimator [SLE] (Yeh et al. Water Resour Res 32: 2757–2766, 1996a) which provides statistically best unbiased estimate of these anomalies and quantifies the uncertainty associated with these estimates via assimilation of available information. Therefore, the information from common diagnosis techniques e.g., inspection methods, and head measurements available at different locations of the system, obtained by different transient tests are fused to provide the most accurate and reliable diagnosis.