(461a) Patterns in Process Safety

Human error is perhaps the most prevalent cause of process accidents as it is in many other technical domains. “Listing human error as one of the causes of an accident is about as useful as listing gravity as the cause of a fall”  (Kletz, 1993).  Understanding and preventing accidents therefore would require a deeper understanding of human error.  Just as patterns such as trees, spirals, waves commonly occur in nature in different contexts and can be used to explain and understand our world, patterns occur regularly in human behavior and therefore in human organizations. Also, a wide variety of sources that reveal human decisions and behaviors are now becoming available. These can be used to observe various aspects of human and organizational behavior – ranging from sustainability reports at the (macro) corporation level, to news and investigation reports at an event level, all the way to  plant operators gaze information while operating the distributed control system (DCS) at the micro level. These new sources of data and information can through new data analysis techniques like text mining offer insights into micro and macro behaviors of people and organization. In this paper, we report on our research to identify recurrent patterns that correlate with process plant accidents. Examples based on actual accidents and real human eye tracking data will be  presented.