In spite of the advances in hazard identification (HAZID) and fault diagnosis, process system failures and industrial accidents still continue to occur. One important aspect to address this situation is the necessity of improved HAZID approaches and the development of diagnostic methodologies and tools based on HAZID knowledge.

In this work we introduce a novel hazard identification method. The results from this method are used to support multi-agent based diagnostic systems that aid operator performance.

The HAZID method is based on a new conceptual paradigm called the Functional Systems Framework (FSF) developed for understanding a range of important aspects of process system design and operation.

The first feature of the method is that it blends the strengths of HAZOP and FMEA to incorporate both deviation analysis and component failure analysis. Secondly, the results are captured in a structured manner where the analysis language has been represented by a HAZID ontology – a formal representation of HAZID concepts. This structured representation provides the starting point to automate the building of knowledge bases used by multi-agent software systems that aid in performing fault diagnosis. The third feature of the method is an ordered strategy for decomposing a P&ID into logical subsystems which facilitates understanding of the process functionality and aids in the application of the method to diagnosis.

This method has been tested with two major Australian process companies, and the results of these case studies will be presented.