Computational Pipeline Monitoring (CPM) leak detection in liquid pipelines can be unreliable or have a high number of false alarms during transient hydraulic events. This paper presents techniques for avoiding these false alarms and for improving leak detectability during steady state operations. Application of this approach has improved both the sensitivity and reliability of CPM-based leak detection systems. When a pipeline is in steady-state operation, the leak detection thresholds can be reduced. When a transient is detected, the thresholds are increased to a higher value. Because of known errors in data measurement and the time at which the measurement was made, large uncertainties in the line-pack prediction can exist during these transient events. In this paper, CPM reliability is improved using dual thresholds; tight threshold for near steady state and relaxed threshold for transient operation. In the presented study, different transient triggers in a pipeline have been implemented to the dual threshold method in order to show the effect of transient thresholds on the reliability of the CPM models or the reduction in the number of false alarms. When the CPM model determines that a transient event is occurring, the imbalance threshold is increased to a pre-set higher value. The transient threshold will remain high until the transient event diminishes, then the thresholds will return to their original steady-state value after a pre-set wait period. This applies to each leak detection window and to each flow meter-to-flow meter section in the CPM model. Detailed information about the dual thresholds approach and the transient triggers is provided in the full paper. Results from the implementation of this method and the quantitative improvement of the CPM reliability and sensitivity will also be presented.

This content is only available via PDF.
You do not currently have access to this content.