More than 200,000 water main breaks occur throughout the US each year, which results in approximately 20% to 30% of treated drinking water being lost during transmission from the treatment sites to the end-users. Without proper devices and technology, 90% of leakages are undetectable. Acoustic leakage detection, including inline and external acoustic detection technologies, has been identified as an effective technology for detecting and localizing existing leakages in water distribution networks. The acoustic leak detection technologies can be used on different pipe sizes and materials. Classical signal processing and machine learning techniques are designed to extract acoustic signals related to pipe failures and pinpoint leakages in surveyed pipes. This study explains the application of acoustic leakage detection by introducing hydrophone-based external acoustic leakage detection. The real-world applications of this technology to detect leakages are shown in some utilities. We present the cross-correlation-based detection algorithm, which is used in almost all technologies in the market to detect leakages between a pair of hydrophones or noise-logger sensors. New development in processing acoustic data, such as wave speed calibration and multi-spectrum analysis, is also presented. A statistical and Markov Chain Monte Carlo method is used to increase the probability of detecting several leakages per survey and extends the capacity of this technology to extract more information from the surveyed pipes.
