An Evaluation of Workload Model Predictions in a Helicopter Environment

Abstract : This report compares the empirical workload results obtained during a field exercise involving four CH-136 Kiowa crews with results predicted from a task network simulation of the exercise. None of the correlation coefficients is outstanding and only a few of the analytical measures explained greater than 50% of the variance in the empirical workload scores. The correlations between the pilots' empirical workload scores and the analytical workload values were generally greater than that found for the observers, possibly a result of the smaller number of subjects in the observer group or perhaps reflecting greater attention focused on the workload of the pilot by the modeling community. The variation of the workload measures within each flight was substantial, suggesting individual differences between subjects as well as differences in the details of each mission played significant roles in determining the perception of workload rated by the subjects. Of the overall workload measures, the simpler measures were found to capture the greatest portion of the empirical workload variance although these measures provide little detail in what is actually causing the overload and at best only capture 50% of the workload variance. While the multi-dimensional workload measures may provide greater detail about what is causing high workloads, they do not seem to be capturing a great deal of the workload variance to begin with. Although the correlations found in this study are low and only half the workload variance was captured, the models may still be useful.