英文摘要 |
Although fault tree analysis has been applied to nuclear, aerospace, and chemical industries for some time abroad, its application in the chemical process industry in Taiwan has been limited. The reasons for this lag are the complexity and uniqueness of most chemical plants, the limited availability of large mathematical process models, the time-consuming and costly process of generating a fault tree, and the limited availability of failure data for process equipment. Undoubtedly, fault tree analysis is an invaluable tool to make safety analysis more systematic and logical and less dependent on rules-of-thumb and arbitrary judgment in the design phases of processes. One major drawback of fault tree analysis is that it is discrete. A component is either in a normal or a failed state; a process is either in a safe or an unsafe condition. Analysis along these lines can be very valuable, but it leaves out the very important time factor. In reality, a process exhibits a wide range of dynamic behavior. A change (sudden failure or gradual deterioration) in a component will lead to a time-dependent change in the state of the process. The process may move toward safe conditions, which may or may not also be suitable operating conditions, or it may move toward unsafe conditions. It may move rapidly or it may move slowly. It does not shift from a safe state to an unsafe one instantaneously. The present study will combine fault tree analysis as well as a realistic dynamic modeling - fault-dynamic modeling to analyze a reactor for cumene oxidation. The more traditional fault tree analysis is used for hazard analysis for process upsets. Then, fault-dynamic modeling is employed to predict the time factor for runaway reactions. |