考量直航轉運並行之軸輻式航空網路模型

Hub-and-Spoke Airline Network Planning with the Coexistence of Direct and Hub Connected Services

軸輻式網路（hub-and-spoko network）具有集貨之功能，可降低航空公司的單位運輸成本，因此經常被業者採用。一般而言，軸輻式網路規劃大多先決定轉運站位址與數目，再對飛航網路進行規劃。然而轉運站位址與飛航網路彼此影響，須同時考慮，因此分階段的規劃方法並不能保證達到航空公司營運成本的最佳化。此外，實務上常有某些航線可允許直航及轉運兩種路徑並存，但過去的文獻並未曾對此做探討。本研究依據模式延伸性分別建構三個模式，其模式約假設直航與轉運兩種路徑可二擇一亦可同時並存。其中完整模式將轉運站位址、數目與營運網路一併同時納入模式考量，外加考慮轉運站容量限制與設置成本，以更接近實務上之需求。三個模式皆被定式為混合整數規劃問題，本研究利用數學規劃軟體CPLEX結合自行發展的程式來求解。實證部分則以兩岸航空貨物運輸為例，分別對三個模式進行不同的情境及敏感度分析。 Hub-and-spoke network has been wildly used in the airline operations. In the design of a hub-and-spoke network, usually the number of hubs and their locations will be determined first, and then the airline service network is established based on the given hubs. However, as the efficiency of a service network highly depends on hub number and locations, the above two-step procedure cannot guarantee the optimal operation cost. Besides, practically, there are three possible service policies in a hub-and-spoke system, direct flight, hub connected flight, and coexistence of both flights. Earlier studies only addressed the service network with either direct or hub connected services. This study developed three different hub-and-spoke network planning models considering three possible service policies. The third unified model was able to determine the number of hubs, hub locations, and flight service network simultaneously. The models were formulated as mix-integer programs. A mathematical programming software, CPLEX, in conjunction with C programs developed by this study was used to solve the models. Finally, the air freight data between Taiwan and Mainland China was used to test the model performance.

Hub-and-spoke network has been wildly used in the airline operations. In the design of a hub-and-spoke network, usually the number of hubs and their locations will be determined first, and then the airline service network is established based on the given hubs. However, as the efficiency of a service network highly depends on hub number and locations, the above two-step procedure cannot guarantee the optimal operation cost. Besides, practically, there are three possible service policies in a hub-and-spoke system, direct flight, hub connected flight, and coexistence of both flights. Earlier studies only addressed the service network with either direct or hub connected services. This study developed three different hub-and-spoke network planning models considering three possible service policies. The third unified model was able to determine the number of hubs, hub locations, and flight service network simultaneously. The models were formulated as mix-integer programs. A mathematical programming software, CPLEX, in conjunction with C programs developed by this study was used to solve the models. Finally, the air freight data between Taiwan and Mainland China was used to test the model performance.