Fault Detection in Pipeline Networks

Operation, supervision, and control of pipeline networks is a highly complex activity, due to the existence of a large number of output nodes with varying demand, and a commonly wide geographical area where different types of risk factors may affect the performance or integrity of the pipeline. Monitoring of unwanted operating conditions is commonly performed as point-based or variable-based deviation analysis, leading to large detection times of events such as leaks, valve shutdown, metering units malfunctioning. In recent years two different types of approaches have been studied, implemented: first-principle model-based fault detection and statistical model-based fault detection. The seminar will present a discussion of the key differences between these approaches, the research that has been conducted in statistical model-based approach, a very recent implementation case in a natural gas pipeline network in Colombia, and challenges to be addressed in future research in fault detection and diagnostics for pipeline systems.

Date: Tuesday, Jun 18, 2013
Time: 2:00 pm - 3:30 pm
Where: Hudson Hall 212
Contact: Thompson, Michele
Phone: 660-5321
Email: mthomp@duke.edu