Leak detection systems are a vital part of a pipeline integrity management program. For liquid hydrocarbon pipelines, these leak detection systems can take the form of measuring conditions inside the pipeline (internal detection) or by use of hardware installed outside of the pipe (external detection). One internally-based technology is acoustic leak detection, sometimes known as rarefaction-wave monitoring. This technology is based on detecting transient pressure waves that are generated when a sudden leak occurs. Acoustic pressure waves travel in the pipeline at the speed of sound of the fluid that is being transported and can be detected by dynamic pressure sensors. Various filters and algorithms can be used to identify this disturbance and distinguish it from other pressure events on the pipeline. This architecture can even be used for noise and for signal pattern recognition to allow for automatic alarming of potential leak events. Each manufacturer of such technology applies unique algorithms or processing methods to capture and analyze the pressure signals that are used to later predict leaks and their locations.
This paper presents a comprehensive review of the technical basis and methodology employed by acoustic leak detection systems in order to further understand their capabilities and limitations. This work included a vast amount of hydraulic modeling aimed at understanding the physics of wave propagation caused by leak events. Diverse parameters, such as initial pressure wave amplitude, signal attenuation, flow and pressure dependence, speed of sound effects, and sensor locations were evaluated. This modeling was conducted for a variety of simulated fluids. A proportional relationship between leak rate and the initial pressure disturbance caused by a leak was obtained. This linear trend can be used in combination with an attenuation model to calculate sensor location limitations. The work determined that the uncertainty in the speed of sound for a pipeline fluid segment significantly impacts the error bands of leak location. The modeling was used to generate correlations for signal attenuation over distance as a function of pipeline conditions.