A Simulation Study on Using Inverse Transient Analysis for Leak Detection in Water Distribution Networks

Leaks from water distribution systems are a major concern to water utilities. Devising efficient and economic leak detection and management strategies is becoming a high priority for almost all water utilities worldwide. During the last two decades, numerous methods for leak detection have been developed and deployed. However, the efficiency and economic viability of many of these methods have impeded their wide scale deployment, and many water utilities still do not have an effective strategy for leak detection and management.

Inverse transient analysis has been used by several researchers since 1994 as a promising and low-cost leak detection technique, and has been shown to produce results in laboratory settings. However, the feasibility and technical limitations of this approach in actual field conditions has yet to be established. This paper is part of a study that aims to assess the applicability and effectiveness of applying inverse transient analysis to the detection of leaks in real water distribution systems.

Essentially, inverse transient analysis involves the establishment of known hydraulic transients at given locations in the network. The pressure signals created by such transients are then recorded at various predetermined locations throughout the system. In parallel, a computer model of the network is coded into a transient analysis software program in which identical transient events are introduced. The model is run for numerous sets of system parameters, a process facilitated by optimization routines such as genetic algorithms, until the best match between measured real system and computer generated transient response is observed. Discrepancies between response signatures (pressure traces at certain locations) can reveal leaks if there is not too much background noise and accurate information regarding system condition and demands is known.

A section of the water distribution network in the City of Regina was chosen for subsequent field-testing. This paper reports on the simulation study conducted on this area. Simulation scenarios of various transient severity and leak sizes are described. Based on the insight gained from the simulation, the paper will conclude by providing a preliminary assessment on the effectiveness and feasibility of using the inverse transient method for leak detection.