Digraph-based models for automated HAZOP analysis.
Reliab Eng Syst Saf 1995 Jan; 50(1):33-49
The development of digraph based models that can be used to automate hazard and operability (HAZOP) analyses was discussed. HAZOP analysis is a widely used standardized procedure for identifying hazards in chemical process facilities. The basic concept underlying a HAZOP analysis is that hazards arise in a process facility due to deviations from its design or from acceptable or normal operating behavior. A HAZOP analysis is performed by a team of experts who systematically analyze a process's piping and instrumental drawings to identify causes and adverse consequences for every conceivable abnormal behavior that could occur in a process facility. In order to identify all possible malfunctions that could occur, process deviations to be considered in the analysis are generated by systematically applying a set of guide words such as MORE OF, LESS OF, REVERSE, PART OF, NONE, HIGH, LOW, NORMAL, AS WELL AS, or OTHER THAN to the process variables. Process variables typically incorporated into a HAZOP analysis include flow rate, pressure, temperature, level of, and composition. A HAZOP analysis is a very labor and knowledge intensive and expensive process. A proposed method for automating HAZOP analyses to make the process more tractable and less intensive was discussed. The procedure is based on creating digraph models of the process units. The digraphs are based on standard system nodes connected by directed edges, but which also include 'abnormal cause' and 'adverse consequence' nodes. The process variables in the digraph can have the value 'null' which corresponds to the guide word NONE in the HAZOP analysis as well as the values HIGH, NORMAL, and LOW. The digraphs will be incorporated into the HAZOP analytical framework using the expert system shell G2 to create an automated expert version of HAZOP, known as HAZOPExpert. The use of HAZOPExpert was illustrated by applying it to data obtained in a HAZOP analysis of a sour water stripping facility. HAZOPExpert's performance in this analysis was found to generate most of the results obtained by a team of HAZOP experts.
NIOSH-Publication; NIOSH-Grant; Grants-other; Risk-analysis; Industrial-processes; Failure-analysis; Industrial-safety; Industrial-design; Chemical-manufacturing-industry; Case-studies
Chemical Engineering Purdue University West Lafayette, IN 47907-1283
Other Occupational Concerns; Grants-other
Reliability Engineering and System Safety
Purdue University West Lafayette, West Lafayette, Indiana