Bayesian belief networks for fault detection and diagnostics of a three-phase separator

Vileiniskis, Marius; Remenyte-Prescott, Rasa; Rama, Dovile; Andrews, John D.

Abstract

A three-phase separator (TPS) is one of the key components of offshore oil processing facili-ties. Oil is separated from gas, water and solid impurities by the TPS before it can be further processed. Fail-ures of the TPS can lead to unplanned shutdowns and reduction of the efficiency of the whole oil processing facility as well as posing hazards to safety of personnel. A novel fault detection and diagnostic (FDD) meth-odology for the TPS is proposed in this paper. The core of the methodology is based on Bayesian Belief Net-works (BBN). A BBN model is built to replicate the operation of the TPS: when the system is fault free or operating with single or multiple failed components. Results of the capabilities of the BBN model to detect and diagnose single and multiple faults of the TPS components are reported in this paper.

Citation

Vileiniskis, M., Remenyte-Prescott, R., Rama, D., & Andrews, J. D. (2016). Bayesian belief networks for fault detection and diagnostics of a three-phase separator.

• Conference Name: European Safety and Reliability Conference ESREL 2016
• Start Date: Sep 25, 2016
• End Date: Sep 29, 2016
• Acceptance Date: Jun 13, 2016
• Publication Date: Sep 29, 2016
• Deposit Date: Jun 30, 2016
• Publicly Available Date: Sep 29, 2016
• Peer Reviewed: Peer Reviewed
• Public URL: https://nottingham-repository.worktribe.com/output/795460
• Related Public URLs: http://esrel2016.org/