To Automate or Not to Automate!

Abstract

Systems are becoming more complex due to various technological developments (such as information and communication technology) and constraints (environmental, social and financial etc.). The first opens the possibility to make links to new domains (e.g. between product quality and profits or financial markets). The constraints encourage the introduction of automated systems designed to keep the process within a certain working range. Both, the technological developments and new constraints, increase the complexity of the system. We investigated the effect of the degree of automation and the effect of the degree of complexity on system and operator performance in various independent experiments. Definitions for the degree of complexity and automation are given and objective and subjective measures for these degrees were applied. The results suggest that an increase in the degree of automation above a certain threshold is not resulting in a proportional increase in system performance. In fact the results indicate that a further increase in the degree of automation is counter effective because it decreases the operator performance during abnormal operations and fault diagnosis. This is due to a loss of situation awareness and internal representation during normal operation before the fault is introduced. Other experiments done in our laboratory show that the operator perceived task complexity reaches its maximum when the operator is controlling in the order of 20 linked sub-systems under normal conditions. We will show that these observations are not only important for operator task allocation and workload assessment but also for the development of alarm handling strategies.