A design methodology for evolutionary air transportation networks
MetadataShow full item record
The air transportation demand at large hubs in the U.S. is anticipated to double in the near future. Current runway construction plans at selected airports can relieve some capacity and delay problems, but many are doubtful that this solution is sufficient to accommodate the anticipated demand growth in the National Airspace System (NAS). With the worsening congestion problem, it is imperative to seek alternative solutions other than costly runway constructions. In this respect, many researchers and organizations have been building models and performing analyses of the NAS. However, the complexity and size of the problem results in an overwhelming task for transportation system modelers. This research seeks to compose an active design algorithm for an evolutionary airline network model so as to include network specific control properties. An airline network designer, referred to as a network architect, can use this tool to assess the possibilities of gaining more capacity by changing the network configuration. Since the Airline Deregulation Act of 1978, the airline service network has evolved from a point-to-point into a distinct hub-and-spoke network. Enplanement demand on the H&S network is the sum of Origin-Destination (O-D) demand and transfer demand. Even though the flight or enplanement demand is a function of O-D demand and passenger routings on the airline network, the distinction between enplanement and O-D demand is not often made. Instead, many demand forecast practices in current days are based on scale-ups from the enplanements, which include the demand to and from transferring network hubs. Based on this research, it was found that the current demand prediction practice can be improved by dissecting enplanements further into smaller pieces of information. As a result, enplanement demand is decomposed into intrinsic and variable parts. The proposed intrinsic demand model is based on the concept of 'true' origin-destination demand which includes the direction of each round trip travel. The result from using true O-D concept reveals the socioeconomic functional roles of airports on the network. Linear trends are observed for both the produced and attracted demand from the data. Therefore, this approach is expected to provide more accurate prediction capability. With the intrinsic demand model in place, the variable part of the demand is modeled on an air transportation network model, which is built with accelerated evolution scheme. The accelerated evolution scheme was introduced to view the air transportation network as an evolutionary one instead of a parametric one. The network model takes in intrinsic demand data before undergoing an evolution path to generate a target network. The results from the network model suggests that air transportation networks can be modeled using evolutionary structure and it was possible to generate the emulated NAS. A dehubbing scenario study of Lambert-St. Louis International Airport demonstrated the prediction capability of the proposed network model. The overall process from intrinsic demand modeling and evolutionary network modeling is a unique and it is highly beneficial for simulating active control of the transportation networks.