AHA: a knowledge based system for automatic hazard identification in chemical plant by multimodel approach

AHA (automatic hazard analyzer), an expert system with new process knowledg e models and inference algorithms for hazard analysis, is developed and tes ted. A multimodel approach is used to build better process models suited to chemical process. Knowledge representation models are composed of a unit k nowledge base, an organizational knowledge base and a material knowledge ba se. Three hazard analysis algorithms (deviation, malfunction and accident a nalysis algorithm) are proposed. AHA is developed using expert system shell G2. The unit knowledge base is devised to model a process unit. It consist s of a unit behavior model and a unit function model. In the unit knowledge base, a process unit is modeled in different terms of variable and functio n. This model represents physical hazards. The organizational knowledge bas e gives information about spatial arrangement of process units and streams. In a material knowledge base, material properties are considered according to NFPA code. This system performs hazard analysis in terms of both functi onal failure and variable deviation and thereby improves the quality of ana lysis and more possible accidents can be identified. The result of analysis provides a pathway leading to an accident, and, therefore, gives not only clear understanding of the accident, but useful information for hazard asse ssment. Using AHA, proposed methodology is applied to the feed section of a n olefin dimerization plant, and performed better than traditional qualitat ive hazard analysis methods such as HAZOP study. (C) 1999 Elsevier Science Ltd. All rights reserved.