Abstract
While global travel restrictions have led to a dramatic collapse in air travel, the in dustry is expected to recover and resume, within a few years, growth rates similar to the 4.6% per year that commercial passenger air traffic had attained in the Asia Pa cific region prior to COVID-19. The expansion of market demand is expected to be met, possibly at the cost of increased network congestion. Due to the high interconnec tivity of airline flight schedules, congestion delays from a single point can propagate downstream to other parts of the network. Consequently, we need an effective repre sentation, emulation, and analysis of the observed phenomena in the air traffic system, in order to facilitate dialogues between airlines, airports, ANSPs, and state authori ties regarding the current management of traffic and available resources, and possibly the exploration of and investing in augmented and alternative system infrastructure to cope with expanding demand. This dissertation develops a comprehensive Air Traffic Flow Management (ATFM) model which gives prominence to the dense interconnectivity of airline flights. It does so by considering passenger connections at hub airports on top of flight connections dictated by aircraft rosters. The former is generalised as inter-tail flight connections rep resenting passenger transfers between distinct aircraft, from a “feeding” set of flights to a “collecting” set of flights at a hub airport. The dissertation also presents a decom position approach to the solution of the Traffic Flow Management Problem (TFMP) by column generation and MIP methods. iii A comprehensive ASEAN air traffic network model is then built and tested with the TFMP to realistically characterise traffic flows within the entire Southeast Asian sub continent, through the fine-grained tuning of various ATFM parameters. The results of the experiments show that complex large-scale instances in the ASEAN subcontinent can be solved in reasonably tractable times. The solutions have a small optimality gap and demonstrate that airline collector flights’ postponed gate departures are caused in part by delays in passenger transfers from feeder flights. In our evaluation, we assert that the granular tuning of airline flight parameters will lead to greater levels of testbed realism, with little to no increase in computational complexity. Finally, the dissertation proposes an original method of measuring system performance at a high resolution, by devising a root cause delay analysis methodology that satisfies fundamental axioms. Root causes of delay (congested network elements) are identified through flight paths, aircraft flight rosters, and passenger connections. Experimental results demonstrate the specificity of our methodology, and allow us to observe the downstream propaga tion of congestion delays across the ASEAN air traffic network through airline flight connections