Abstract:
Chemical process industries involve a lot of hazards like flammable hydrocarbons, toxic chemicals, etc. These hazards have a tendency to get transformed into accidents that often have catastrophic onsequences e.g. Bhopal gas tragedy, Texas refinery accident, etc. To prevent such accidents process plants have layers of protection. One of the most important layers of protection is that of humans. It is interesting that human errors are a root cause of majority of accidents taking place in process industries. However, sadly, limited research has been done in the field of human errors. Determination of human failure probabilities is still heavily dependent on expert judgment which is prone to inconsistencies due to limited or no availability of relevant data. In this thesis, we propose an experimental strategy, based on collection of process data to validate human failure probabilities of control room operators, arrived at by use of expert judgments. We, further, propose a methodology to measure human operator erformance and then improve it so as to reduce human failure probability, and subsequently minimize the risk associated with a plant. The experiments involve a total of 71 graduate and undergraduate students.