Three flight simulator experiments examined how a health monitoring system may aid pilots in detecting flight control faults. The first experiment introduced an unexpected fault in the flight control system during an approach to a fictitious airport. The second experiment used a factorial design of (1) presence ?? notof a Fault Meter display and (2) presence ?? not ?? an Alerting System, which could have one or two phased alerts. In half the runs, a fault was triggered at some point, and pilot response was recorded. The next experiment comprised one flight in which pilots were given a false alarm by these systems, testing for automation bias. No consistent pilot response was found to the faults, with pilots sometimes successfully landing the aircraft, sometimes immediately or eventually initiating a go-around, and sometimes loosing aircraft control and crashing. The pilots were not able to identify the fault in 11% of the cases. Tunnel tracking error increased following the faults and the false alarm, suggesting it may be both a manifestation of attempts to diagnose a fault and a cue to pilots of a problem. Finally, the triggering of a false alarm showed the existence of automation bias induced after a small number of interactions with the HMS.

Visualising uncertainty information has been a research area for the past decade or so, and this thesis contains the results of an experiment that examines whether prior research on icon degradation for showing uncertainty can be used in a simulated aircraft cockpit environment. Using icon degradation has been suggested as being effective to combat overconfidence bias, as well as to accurately convey information about uncertainty. Two icon sets using icon degradation were taken from prior research, and one new icon set using shape change and colour change was created for comparison. Subjects flew a flight simulator while reading icons to evaluate the uncertainty displayed, and also evaluating their own confidence in their reading. The results show that shape change leads to much higher accuracy in reading icons, and slightly higher levels of confidence. Furthermore, icon degradation results in a higher variance in reading icons and an increase in errors when no time-pressure or distraction is present. This suggests that the suitability of icon degradation for showing uncertainty is questionable in all situations, and that other design approaches such as shape change should be considered. Furthermore, problems were uncovered in the prior research that the old icons were taken from, and these problems call into question the general approach used in that research. Keywords: Uncertainty visualisation, Naturalistic decision-making, NDM, Aviation, Aircraft cockpit, Decision support, Situation assessment, Threat assessment.

This thesis presents structural analysis of a cockpit of a jet trainer type aircraft and the correlation studies performed by using ground pressurisation test results. For this purpose, first the response of the complex integrated fuselage structure is investigated under the complex type cabin pressure load. Then, cockpit part of the fuselage structure is modelled using commercial finite element software MSC/PATRAN&reg / and MSC/NASTRAN&reg / . The finite element model (FEM) of the cockpit structure is improved by the examination of the ground pressurisation test data and is finalised after achieving a good correlation between the finite element analysis (FEA) and the test results. This final form of the FEM of the cockpit structure serving as a benchmark is proved to be reliable for any future modifications.