Operator Performance in the Control of a Laboratory Process Plant

Abstract

Subjects were required to control a relatively complex and slow-response laboratory process plant for a total of three hours. Output error and control movement scores were measured, and power spectral density and cross correlation functions were computed. The subjects' strategies were assessed by observation and from the responses to a questionnaire given after they had completed the experiment. The results show that when intermediate information from the plant is available, subjects tend to develop an anticipatory model, which enables them to control the plant more effectively than with output error feedback alone. Special instructions about the structure and dynamics of the plant tended to facilitate the development of the anticipatory model. Under the particular conditions of the experiment, a time-history record of the control settings and output error were of little value. It was found that observation of the output of a three-term automatic controller was of limited value as a training aid. In general, the human operator's strategies and information requirements were very different from those of the automatic controller.