Situations Saved by the Human Operator when Automation Failed

Abstract

In many systems an increased level of automation implies an altered role for the human operator. Behind the introduction of new automation lies different automation philosophies which stretches from trying to use as much automation as possible to adding automation only as a support to human tasks in specific situations.The Swedish Radiation Safety Authority has assigned AF Infrastructure, Department of Risk Management to describe the current automation philosophies within the nuclear industry. The assignment also includes a survey of events in which human involvement was necessary in order to save situations in which the automation failed.Much of the knowledge of human-machine interaction in process systems is derived by learning from incidents. These incidents, however, only represent a limited set of observations of real life human-machine interactions when the human in most cases did not have a positive effect on the sequence of events. Cases when the human operator did have a positive effect on the sequence of events are typically not reported as they in most cases do not lead to incidents. Thus, much of the available information on human-machine interactions is biased towards cases of human errors, and does not give a true picture of real-life. It is the hypothesis of this paper that the role of the human operator as a vital safety function is more significant than normally concluded by studying incident reports which claim human error as a contributing cause to accidents.In this report two events are described in which human intervention was crucial for the successful outcome of the situation; Vandellos/Spain 1989 and Forsmark/Sweden 2006. These events show that the human operator is one of the most vital parts of the defence in depth at a power station, hence a strong focus should be given to looking after and maintaining the human abilities in order for her to be able to act safely in emergency situations.The events also show the potential improvement of the defence in depth by making the most of the unique human abilities of intuitive and creative thinking and acting without access to external sources of power or prearranged procedures.These abilities are, however, affected by the level of automation, e.g. a too high level of automation can lead to a lack in situation awareness whilst a too low level can lead to too high levels of mental workload for the operators. To avoid degradation in human abilities to safely intervene, changes in automation levels should always be preceded by an analysis of its long term effect on the human operators’ skills and capabilities.To gain more knowledge on the role of the human operator as a safety function in process systems, it is suggested that real-life human machine interaction is actively observed in order to also identify cases when human intervention prevents incidents at the very early stages of the sequence of events. These cases are normally impossible to retrieve from historical records, which mainly cover incident investigations of cases that lead to severe consequences.Many of the theories on automation philosophy used in this paper originate from the aviation industry, but it is shown that they are equally applicable within other sectors as well, i.e. the process industry.