The Emergency Shutdown System (ESD) is a type of Safety Instrumented Systems (SIS) used to shut down the system in the event of anomalous conditions, it consists of sensors, logic solvers, and final elements. Reliability databases such as OREDA (Offshore and Onshore Reliability Data) and expert judgment indicate that Emergency Shutdown Valve (ESDV) failures are the most common cause of the unavailability of ESDs. Therefore, proper diagnostic is essential to prevent the failure risks of these valves. ESDV problems are of a different nature and difficult to treat with mathematical models because of their non-linear behavior, the imprecision of information, and the appearance of many failure modes that arise from many failure causes. Therefore, failure diagnostic mechanism based on heuristic knowledge of ESDV parameters must be established. In this paper, the problem of diagnosing ESDV failures is addressed based on the fault-symptom tree model and the Multiple Input-Multiple Output (MIMO) fuzzy inference system. The latter is built on a set of linguistic rules “if-then” provided by the fault-symptom tree model. the proposed approach was applied and verified on an ESDV subsystem in the petrochemical industry.

The Safety Instrumented System (SIS) is an automated system used to implement one or more safety instrumented functions. A SIS, like the Emergency Shutdown (ESD) system, consists of any combination of sensor(s), safety PLC(s) and final element(s) (e.g. ESD valves). ESD valves are the last line of defense against risks, although the ESD valve has high performance, the data (based on expert judgment and OREDA database) indicates that ESD valves failures are the most critical in the ESD systems. In order to improve the reliability and safety of these valves, we applied the FMEDA diagnostic technique. We started with a decomposition of the ESD valve to the subsystems and we identified their functions. Then we described the failure modes, their mechanisms, their sites and their effects. Then we identified the impact of each failure mode according to the criticality classes and identified the failure rates and their class according to the criticality and the detectability by automatic diagnosis of each mode and from the failure rates we calculated the Safe Failure Fraction (SFF) and Safety Integrity Level (SIL) required and we concluded that the actuator subsystem is the most critical system. Finally, we proposed preventive and protective measures to eliminate or reduce the risk of failure.