Methodology for computer aided fuzzy fault tree analysis

Process facilities are well known for unplanned chemical emission, toxic release, fire and explosion and operational disruption. These incidents have the potential to cause an industrial accident and environmental damage. From the investigation of all major accidents, it is apparent that most industrial accidents can be avoided or restricted with a systematic risk analysis and safety management strategy. An effective risk analysis strategy always gives preference to minimizing the risk of a process facility at its design stages. -- Probabilistic risk assessment (PRA) is a comprehensive, structured and logical method for identifying and assessing risks of complex process systems. It uses fault tree analysis (FTA) as a tool to identify basic causes leading to an undesired event, to represent logical dependency of these basic causes in leading to the event, and finally to calculate the probability of occurrence of this event. Probability data estimation, and large and complex fault trees, are challenging aspects of FTA as applied to process facilities. -- Quantitative analysis of a fault tree for a process system requires a fault tree and the system components (basic events) failure data. Sometimes or always it is difficult to have an exact estimation of the failure rate of individual components or the probability of occurrence of undesired events due to a lack of sufficient data. Further, due to imprecision in basic failure data or the data sufficiency the overall analysis of a fault tree may be questionable. To avoid such conditions, a fuzzy approach may be used with the FTA technique. This reduces the ambiguity and imprecision arising out of the subjectivity of the data. -- Fault tree construction for a process facility must accommodate for a wide variation in components, process operations and control mechanisms. It is more scientific to analyze such a large and complex fault tree through proper sub-divisions of the tree. A proper modularization technique (sub-division) can sub divide a tree into its equivalent sub trees and then analyze it for the process facility. -- This work is focused on developing a methodology of a fuzzy based computer-aided fault tree analysis tool. The central idea of this methodology is to adopt a suitable algorithm for moduling (sub-dividing) a large and complex fault tree and then evaluate it by using the fuzzy approach. This methodology uses a systematic approach of fault tree development, fault tree modularization, minimal cut sets determination, fuzzy probability analysis, and fuzzy based sensitivity analysis of a system for achieving its objectives. Besides developing a methodology for computer- aided FTA, this study also proposes a procedure of fuzzy approach for the uncertainty analysis, which is used for comparing error robustness of fuzzy FTA and conventional FTA