Risk-based fault diagnosis and safety management for process systems

Today, plants in chemical and process industry are becoming larger and more complex. Corollary of this trend implies that each hour of down time is more expensive. As industrial systems enlarge, the total amount of energy and material being handled increases, making fault diagnosis and safety management considerably important both from the viewpoint of process safety as well as economic loss. Therefore, seeking an effective approach to perform fault diagnosis and implement safety management is important and imperative. An innovative methodology of risk-based SPC fault diagnosis and its integration with Safety Instrumented System (SIS) is proposed in this thesis to assure the process safety.;The proposed methodology of risk-based SPC fault diagnosis and its integration with safety instrumented systems is implemented using G2 development environment. To test and verify this methodology, a tank filling system and a steam power plant system with SIS1s and SIS2s are developed in G2 environment. A technique breakthrough, from univariate monitoring to multivariate monitoring for SPC fault diagnosis has been made in the verification in the steam power plant system.;Unlike any existing fault diagnosis and safety management approaches, the proposed methodology pioneers a brand new pathway for the fault diagnosis and safety management in process industry. This proposed methodology neither depends on any process model as model-based methods, nor depends on large amount of historical process data as conventional process history based method. Control chart technique is used to distinguish abnormal situation from normal operation based on three-sigma rule and linear trend forecast. Time series and moving average techniques are used to perform real time monitoring and noise filtering in fault diagnosis process. Furthermore, risk indicators are used to identify and determine potential fault(s) to minimize the number of false alarms.