Empirical Evaluation of an Industrial Application of Ecological Interface Design

Abstract

Abnormal events in production plants cost the petrochemical industry billions of dollars annually. In part, these events are difficult to deal with because current interfaces do not adequately inform operators about the state of the process. Ecological human-machine interfaces aim to provide information about higher-level process functions. Several laboratory simulator studies have shown that, in comparison with contemporary process interfaces, ecological interfaces can lead to faster fault detection, better root-cause diagnosis, and more effective control responses. However, an empirical evaluation of these findings for professional operators in more realistic plant settings has been absent from the literature. In this study, two ecological interfaces were created for a representative petrochemical refining process. One was a traditional ecological interface based on a system-based analysis and the other was an ecological interface augmented with additional task-based information. Professional operators used the novel interfaces in an industrial simulator to monitor for, diagnose, and respond to several types of process events. In comparison to operators using the current process interface, participants in both ecological interface conditions showed better control performance, while the participants using the augmented ecological interface provided more accurate fault diagnoses than either of the other two groups. The results shed light on practical implications for the use of ecological interfaces in the process industries.