Modelling Regulation Principles of

Human Activities

---

Project A -Human learning, performance and reliability in operating complex systems

The project aims to build models for examining

1. The learning trajectories of how individuals learn to operate a highly complex system (in this case, a chemical production system).
2. Individual performance efficacy in transferring their acquired knowledge and skills to work under normal and extreme conditions (such as under high time pressure).

These models will highlight the situational and individual characteristics that affect human learning, performance and reliability.

Collaborator:

Dr Olaf Kohlisch, Institute of Occupational Physiology, Dortmund, has developed a simulator of chemical production system. Currently, the simulator is used to evaluate a three-year apprenticeship program in the chemical production industry.

Potential areas for industrial exploitation:

We are building an analytical toolkit for aiding system engineers in identifying factors which may have detrimental effects on system productivity and human reliability. The toolkit will automate the data collection for the purpose of facilitating a multi-level analysis of how human activities during operations are regulated. In doing so, issues relating to ergonomics and reliability engineering can be walked through during the design phase for the early mitigating of system failure and the reduction of incidences of system-induced human error.

---

Project B - The effects of mental workload on air-traffic controllers.

We examine the impact of low and heavy air traffic on the tactical operations of air traffic controllers. More specifically, we introduce a class of event history analyses for examining how an air traffic tactical controller manages time under heavy and light workload conditions. The analysis begins by assessing the changing hazard rate, h(t), of a transition (or spell) between the controller's communication and flight progress activities. h(t) is the instantaneous rate of going from one state (or an activity of communication or flight progress) to another in a unit of time.

Our findings indicated that the spells or transitions among activities reported closely followed a Weibull distribution, which supports the validity of using this class of analysis. Further Cox regressions indicated that the spell distribution was more likely to be regulated by time in heavy than in light workload conditions. The regression also suggested that under heavy workload, indirect speech from the planner controller would decrease the hazard rate for communication to flight progress spells. The results suggest that a dynamic model for analysis of air traffic control may be necessary, and that the implications of using modular automation may not be straightforward.

Collaborator: Dr.Peter Goillau, Defense Research Agency.

---

Forthcoming Publications:

Kuk, G., Arnold, M., & Ritter, F. (submitted for publication). Using hazard rate models to examine the impact of workload on an air traffic controller's operations.

Kuk, G., Baxter, G., & Kohlisch, O. (in prep). Towards a framework for task performance analysis.