Sensitivity Study of a Computer Model Based Leak Detection System in Liquid Pipelines

Abstract

Computer model based leak detection technique is extensively applied in the energy pipeline industry. However, its effectiveness is often limited by the ability of the computer models to accurately reproduce the complex real-world pipeline systems and by inaccurate and insufficient measured data. Therefore, it is important to evaluate the impact of various factors and identify those which have the largest effect on leak detection. This sensitivity study investigated the effect of leak rate, the R factor (which includes pipeline, instrument and operating variables), leak location, transient type and severity, pressure and flow noise level in instruments on the response of the leak detection system. Datasets of simulated leaks on a virtual pipeline were created and used to test this sensitivity for various leak scenarios and data without and with instrument noise. The output variables of the tests revealed the leak detection system is more sensitive to R factor than to flow state, noise of instruments, and the transient severity. Leak location seems to have a smaller impact on the leak detection system. Leaks are easier to detect in pipeline systems with low R factor or during flow decrease transient events. These results are valuable to prioritize future improvement on the computer model, instruments, or the SCADA system in pipeline systems with low R factor or flow decrease transient events.