Learning HAZOP expert system by case-based reasoning and ontology

Safety is an important issue in process design and operation in the chemical process industry (CPI). It is even more critical for modern chemical manufacturing processes, which are either operated under extreme conditions to achieve maximum economic profit, or are highly flexible. The importance of safety analysis in process operation is well recognized after occurrence of several tragic accidents that could have been avoided by adequate process safety analysis. To ensure safe operation, process hazard analysis (PHA) is very important to proactively identify the potential safety problems and recommend feasible mitigation actions. Among the available PHA techniques, hazard and operability (HAZOP) analysis is the most widely used one in the CPI. HAZOP analysis done by human teams has the following shortcomings: time consuming, laborious, expensive and inconsistent. To solve these problems, various model and/or rule-based HAZOP expert systems have been developed during the last one decade [Venkatasubramanian, 2000]. These systems, however, can only address “routine” or process-generic HAZOP analysis. In the CPI, “routine” HAZOP analysis roughly occupies 60-80% while “non-routine” or process-specific HAZOP analysis occupies 20-40%. Due to the lack of selflearning capability of the current HAZOP experts systems, the knowledge of non-routine analysis could not be formulized and reused for similar chemical processes, and the “non-routine” HAZOP analysis still needs to be addressed by human experts. Therefore, the completeness of the existing HAZOP expert systems is discounted. Recently case-based reasoning (CBR) technology has been integrated into HAZOP automation technology by researchers at Purdue University to enhance the selflearning capability of HAZOP expert systems [Zhao, 2005]. However, the case based reasoning they proposed aimed to facilitate modification of the existing models and creation of new models based on the knowledge in the existing models. The “non-routine” HAZOP analysis still replies on the human team. To improve the learning capability of HAZOP expert systems, a new learning HAZOP expert (LHE) system has been developed based on CBR that can help automate “non-routing” HAZOP analysis. In the LHE system, the HAZOP analysis knowledge is represented as cases and stored in a case base (a structured database). The case structure and the indexes chosen for CBR are addressed. A case constructor is designed for effectively creating cases. The case retrieval strategy, which combines DFS (Depth-First Search) and BFS (Breadth-First Search), is also discussed in detail. A new CBRbased HAZOP analysis ontology (CHAO) is created by integration of existing ontologies reported in literatures to enhance the case retrievals. At last the application of the LHE system is demonstrated by industrial processes. References: • Venkatasubramanian, V., Zhao, J., Viswanathan, S.,2000. Intelligent systems for HAZOP analysis of complex process plants, Computers & Chemical Engineering 24: 2291-2302 • Zhao, C., Bhushan, M., Venkatasubramanian, V., 2005. PHASuite: An automated HAZOP analysis tool for chemical processes Part I: knowledge engineering framework, Trans. IChemE, Part B: Process Safety and Environmental Protection, 83(B6):509-532 Mary Kay O’Connor Process Safety Center 2007 International Symposium http://process-safety.tamu.edu