Integrating knowledge-based and mathematical programming approaches for process safety verification

Hazard and Operability analysis (HAZOP) is a popular method for performing hazards analysis of chemical plants. It is labour- and knowledge-intensive and could benefit from automation. Recently, a knowledge based framework for automating HAZOP analysis, called HAZOPExpert, was proposed. Dimitriadis et al. (1995) proposed a quantitative model based approach for hazard evaluation. This approach used a dynamic model of the plant and bounds on the process disturbances (including failure modes) and parameters to identify possible unsafe situations. The qualitative analysis performed by HAZOPExpert is thorough and computationally efficient. However, in some situations it suffers from ambiguity. The quantitative analysis has the capability to perform an exact analysis without ambiguities, but a complete quantitative analysis can be computationally prohibitive. In this paper, we present an integrated qualitative-quantitative approach for hazard identification and evaluation which overcomes the shortcomings of qualitative and quantitative methods. In the integrated framework, the broad details of a particular hazardous scenario are extracted by inexpensive qualitative analyses. A detailed quantitative analysis is then performed if needed and only on those parts of the plant identified by the qualitative analysis to contribute to the hazard. This framework is illustrated using an industrial case study.