An experimental verification of the inverse transient technique for leak detection

Abstract

Traditionally, leak detection in pipe networks has been performed using mass balance, steady state and acoustic methods. These methods have problems with accuracy and determinacy of leaks, are intrusive or are expensive to apply. In 1994 Liggett and Chen proposed the use of transients for the determination and location of leakage in a pipe network. In their method, the inverse transient method, the pressure head is measured at a number of locations in a pipe network that is experiencing a transient event. Then, using a transient model that incorporates leakage, different leak candidates are tested at various locations in the network. A least-squares minimisation is performed to fit the measured pressure heads to the modelled pressure heads and determine if a suspected leak is a true leak. This paper presents laboratory-based verification of the inverse transient method using the Levenberg-Marquardt and SCE minimisation algorithms.