Airline companies make a huge effort to maximize their revenue while keeping their costs at a minimum. This is not an easy task and, because of that, airline companies invest in tools that allow to optimize their operational schedule. Unfortunately, any operational plan has a strong probability of being affected, not only by large disruptions like the one that happened in April 2010 due to the eruption of the Iceland Eyjafjallajökull volcano but, more frequently, by smaller daily disruptions caused by bad weather, aircraft malfunctions and crew absenteeism, for example. These disruptions affect the original schedule plan, delaying the flights, and cause what is called an Irregular Operation. Studies have estimated that irregular operations can cost between 2% and 3% of the airlines annual revenues and that a better recovery process could result in cost reductions of at least 20% of its irregular operations. In this thesis, we have studied the AOCC of TAP Portugal as well as the work of other researchers in this field in order to propose a distributed and decentralized general approach to integrated and dynamic disruption management in airline operations control, based on the Multi-Agent System (MAS) paradigm. This is one of the main contributions of our work.The approach is distributed because it allows the functional, spatial and physical distribution of the intervening agent roles and the environment (i.e., resources available); it is decentralized because some decisions are made in different nodes of the agents network; it is integrated because it includes the main dimensions of the problem: aircraft, crew and passengers; and it is dynamic because, in real time, several agents are performing in the environment, reacting to constant change. Although we use the TAP Portugals AOCC as a case study, our work was performed with an out-of-the-box thinking. By out-of-the-box thinking we mean trying to think beyond the requirements of the specific problem we are solving as of this moment in time, exploring divergent directions and involving a variety of aspects that, at this moment in time might not be relevant but, in the future, might be an asset. We propose (and this is another one of the main contributions of our work) a negotiation protocol called Generic Q-Negotiation (GQN) to be used as a decision mechanism in the disruption management process. This protocol has characteristics that are beyond what we need to solve the problem as it is today. However, these characteristics, besides solving the problem as we know it today, allow it to be applied to different scenarios that may arise in the future, either for the same application domain or to different application domains that share similar characteristics. The same line of thinking was followed in the other contributions of our work, including the proposal of an Agent-Oriented Software Engineering (AOSE) methodology called PORTO, which resulted from the use and improvement of other methodologies. Based on our MAS-based approach and on the GQN protocol, we have performed several experiments, comparing the results with those of the manual process followed by the AOCC operators at TAP and with an automated process that implements the typical sequential approach followed by the AOCCs. The results show that our proposal, not only corroborates existing studies regarding the possible cost reductions that could result from a better disruption management process but, also, gives the possibility of reaching solutions that balance the utility of the three dimensions of the problem: aircraft, crew and passengers.
| Identifer | oai:union.ndltd.org:up.pt/oai:repositorio-aberto.up.pt:10216/69040 |
| Date | January 2013 |
| Creators | António J. M. Castro |
| Source Sets | Universidade do Porto |
| Language | English |
| Detected Language | English |
| Type | Tese |
| Rights | open access |
Page generated in 0.0026 seconds