Experience with an expert system for automated HAZOP analysis.
Computers Chem Eng 1996; 20:S1589-S1594
The experience with an automated version of the hazard and operability (HAZOP) analysis system was discussed. HAZOP which is a widely used system for analyzing potential hazards in the chemical processing industry is a very labor intensive, time consuming, expensive process. An expert system for automating the HAZOP system (HAZOPExpert) has been developed to deal with these problems. The HAZOPExpert system is based on separating the knowledge required to perform a HAZOP analysis into process/specific and process/general knowledge and using generic HAZOP models of the process units. The process/specific knowledge component consists of information about the materials used in the process, their properties, and the piping and instrumentation drawings of the facility. The process/general knowledge consists of the HAZOP digraph models of the process units, qualitative causal models that have been developed specifically for hazard identification. The HAZOPExpert procedure has been applied to identifying potential hazards in a sour water stripping facility, a hydrotreator facility, and an ethylene production facility. The results agreed with those obtained by teams performing conventional HAZOP analyses of the facilities. The HAZOPExpert analyses, however, found more hazards, defined as abnormal causes and adverse consequences, than the HAZOP analyses did. For example, the HAZOPExpert analysis of the ethylene production facility found 21 abnormal causes that could result in 50 adverse consequences in five process deviations that were considered. Some of these seemed unlikely. The HAZOP analysis identified 11 causes and 11 consequences. A semiquantitative filtering technique based on quantitative data and specifications of the process units and materials when incorporated into the analysis reduced the number of causes found by the HAZOPExpert procedure to 21 and the number of consequences to 24. The causes and consequences found by the HAZOP team were included in those identified by HAZOPExpert after filtering. The authors conclude that evaluating potential hazards in chemical process facilities using intelligent systems such as HAZOPExpert has been validated and is now ready for industrial applications.
NIOSH-Publication; NIOSH-Grant; Grants-other; Risk-analysis; Industrial-processes; Failure-analysis; Chemical-manufacturing-industry; Occupational-hazards; Industrial-design
Chemical Engineering Purdue University West Lafayette, IN 47907-1283
Other Occupational Concerns; Grants-other
Computers in Chemical Engineering
Purdue University West Lafayette, West Lafayette, Indiana