Leak Detection in Networks of Pipelines by the Generalized Likelihood Ratio Method

An efficient model-based method to detect, locate, and estimate leaks occurring in networks of pipelines is proposed in this paper. The method is applicable to networks carrying incompressible fluids. The proposed method is an adaptation of the generalized likelihood ratio test which utilizes flow and pressure measurements and a steady state model of the network. It not only can predict leaks and isolate the leaking branch but also can estimate the magnitude of leak and location along the branch. Modified serial compensation strategy is employed to detect multiple leaks. A simulation study of a large network demonstrates the efficacy of the method. Also a bench scale laboratory experiment is performed to corroborate the simulation study.