Computational Pipeline Monitoring in Challenging Situations

This paper introduces a new, active methodology to modeling and leak detection intended to mitigate the effects of data uncertainty in such challenging situations, and presents three case studies. The American Petroleum Institute (API) has coined the phrase Computational Pipeline Monitoring (CPM) to encompass several methods of leak detection. The use of real-time transient hydraulic simulation tools, driven by data gathered by a Supervisory Control and Data Acquisition (SCADA) system, is one form of CPM system. Such real-time simulations impose SCADA-gathered data (typically pressures, flows, temperatures) onto a characterization of the pipeline (the model) and the fluids in the system. In a tuned CPM system, if the SCADA-gathered data cannot be successfully imposed on the model without transgressing the laws of fluid mechanics, this signifies a pipeline anomaly, which may be a release. However, in reality, many pipeline hydraulic anomalies are due to changing uncertainties in the data presented to the model and if annunciated to the pipeline operators would constitute a “false leak alarm.” While they typically are not large enough to compromise pipeline operations, uncertainties abound in the SCADA-gathered data. Even were the SCADA-gathered pressure and temperature data to contain no uncertainty, the fluid properties might not be sufficiently characterized for the simulation to accurately calculate how the fluid behaves under pressure and/or temperature changes. Measurement failure further complicates the task of the CPM application, as does slack line flow. Uncertainty in the CPM-driving data is not constant, it is ever-changing with variations in the pipeline flow rate, the characterization of the fluids in the line, and the quality of the individual measurement data, to mention only a few. CPM systems use a variety of methodologies to vary their sensitivity according to the uncertainty in the data used for their calculations. However, in general terms, the more uncertainty there is in the data, the lower the resulting system sensitivity becomes. Active features in a CPM leak detection system can mitigate the performance degradation due to varying data uncertainty.