(166c) Identification and Location Estimation of Multiple Leaks from a Natural Gas Pipeline

Pipelines are the most important transportation methods for natural gas and oil in the booming energy industry. However, the leakage accidents from the pipelines have generated significant safety concerns.  There are two types of leak-detection methods, which have been proposed and implemented, these include hardware-based and software-based methods[1]. Software-based methods only use pressure and flowrate measurement data at the boundary of the pipeline without requiring extra instrumentations[2].  Natural gas pipeline is a more complicated problem comparing to water and oil pipeline because of the possible phase change of the natural gas corresponding to the changing thermal properties.

Identification and isolation of multiple faults is a challenging task. Multiple faults could be mistakenly treated as one fault, especially for a software-based method. Detection of multiple leaks from a water pipeline have been investigated[3, 4]. However, to the best of our knowledge, identification of multiple leaks from a natural gas pipeline has not been studied yet.  In this paper, we are proposing a new method for identifying and isolating multiple leaks from a natural gas pipeline with consumer connections. Both subsequent and simultaneous multiple leaks are considered. For the identification of subsequent multiple leaks, an unknown input observer is designed to estimate the boundary flowrate of the pipeline[5]. The unknown input observer takes ground temperature and boundary pressure oscillation as unknown inputs and it is able to distinguish the real leak events from the disturbances. An adaptive linear model was applied to incorporate the disturbance information and detect multiple leaks.   

In order to identify and isolate the simultaneous multiple leaks, a robust optimization algorithm is applied. The optimization is based on the isothermal model of the natural gas flow in a pipeline. However, the parameter in the model may vary depending on the inputs (leaks), and it remains unknown and to be determined. A robust algorithm is adapted to deal with the uncertainty of the parameter.  The objective function of the optimization algorithm is to minimize the difference between the measurement and estimation. An adaptive search algorithm is applied to identify the inputs (locations of leaks) and reduce the computational time.

Reference

[1] Geiger G, Werner T, Matko D. Leak Detection and Locating–A Survey.  35th Annual PSIG Meeting2003.

[2] Murvay P-S, Silea I. A survey on gas leak detection and localization techniques. Journal of Loss Prevention in the Process Industries. 2012;25:966-73.

[3] Verde C, Visairo N, Gentil S. Two leaks isolation in a pipeline by transient response. Advances in water resources. 2007;30:1711-21.

[4] Verde C. Accommodation of multi-leak location in a pipeline. Control Engineering Practice. 2005;13:1071-8.

[5] Sename O. Unknown input robust observers for time-delay systems.  Decision and Control, 1997, Proceedings of the 36th IEEE Conference on: IEEE; 1997. p. 1629-30.