Document Type


Publication Date





Controller workload, recognized as a significant bottleneck to capacity increase in the future National Airspace System, has been researched extensively in air traffic management. Unfortunately, subjective workload has been an unreliable predictor of a controller’s ability to safely manage the traffic, leading to attempts at replacing workload with more objective metrics, such as task load (e.g. number of clearances) and traffic density (e.g. aircraft count). A significant caveat to substituting these metrics for workload ratings, however, is that their relationships are nonlinear. More specifically, when the objective metrics, such as aircraft count, increase linearly, the controller’s perceived workload remains low until the traffic and associated task load increase to a critical threshold. From this point, the workload increases at a much faster rate with each added task. In a series of informal studies conducted as a precursor to testing Distributed Air Ground Traffic Management (DAG-TM) concepts, researchers at NASA Ames Research Center manipulated aircraft count in real-time human-in-the-loop simulations to determine the maximum traffic levels at which the controllers stated that traffic would no longer be manageable. As hypothesized, traffic scenarios that elicited moderate levels of controller workload quickly became unmanageable when only a few aircraft were added. Feedback from the controllers further supported the non-linear nature of subjective workload. Task load data partially supported the above findings but the results were inconclusive due to individual differences and varying results from different task load metrics. The non-linear relationship between subjective workload and aircraft count has been further examined using data collected from the Free Maneuvering concept feasibility study in June 2004, which shows a step-function relationship between the two. The combined results suggest that any estimation on workload should not be extrapolated linearly from a set of workload measures taken from an experiment since the extrapolated workload is likely to significantly underestimate workload.