Dynamic Capacity Allocation for Airlines with Multi-Channel Distribution

In 2017, China's online e-commerce sales has already reached 29.16 trillion Yuan. While gaining huge benefits, it also poses great challenges for each industry. One of the biggest challenge is the change of sales channels. Of course, there are also huge opportunities between them. Among them, it is a huge impact on the perishable products industry such as the airline industry and the hotel industry. Because of the perishable of the product, both hotels and airlines want to be able to sell the product for a limited period of time and gain considerable benefits. Therefore, at the beginning of the industry, airlines and hotels hoped to sell their products through more channels and attracted channels to sell products by paying their commission fee. With the rapid development of e-commerce, more and more online channels are replacing traditional offline channels. The change of channels has brought great challenges to airline management and costs. For example, although online channels absorb more customer demand, the commission costs of airlines have increased significantly. In addition to the cost pressures imposed on airlines, the increase in channels has brought conflicts between channels and between channels and airlines. Some of the channels' behaviour has caused great losses to the airlines. For example: change the condition of retreat fee, increase ticket or room price, maliciously reduce the price to compete with the airline and so on. These behaviours have affected the airline's reputation and have also brought losses to the airlines. In order to deal with the challenges of online agents, the airline has also taken some corresponding measures, such as the opening of online direct marketing websites, direct sales APP and so on. However, the effect has not been very good and it is difficult to compete with online agents who have customer volume. At the same time, we also see that the airline industry and hotels are also facing great competitive pressure. For example, the high-speed rail increases the competitive of civil transport markets. High-speed trains generally have higher on-time rates than aircraft and also high-speed rail stations are generally more convenient for customers in the city. Therefore, for passengers, high-speed rail has advantages in short trips. In addition, the emergence of low-cost airlines has also intensified competition in the civil aviation industry such as China's Spring Airlines, Europe's Easyjet and Ryanair. Therefore, recently reducing channel distribution costs has been concerned for many airlines which are facing fierce competition in airline markets. In a long period since the 1970s, capacity control has always played a pivotal role in defining airlines market strategy. However, when airlines select distribution channels and make capacity allocation decisions, they still separately make different decisions. Hence, when a customer purchases a ticket from a channel with an appropriate fare class, the channel might not be an optimal channel from the airlines' perspective. When the airline sells a ticket in a right channel, the ticket price is probably not a right fare. Therefore, how to establish a better channel and fare class capacity control model has become the key for airlines to increase revenue. This thesis is a study based on the above issues. The main work includes the following aspects: At first, we studied the single-leg capacity allocation problem that considers the channel factor. Although the network revenue management has a lot of academic research and has been applied in international routes, for many domestic routes airlines still basically use single-leg revenue management system. In addition, from the historical development of revenue management, the single-leg revenue management model is the basic model of all revenue management models. Therefore, it is important to first establish a single-leg revenue management model that considers the channel issues. In this study, we will integrate channel distribution into dynamic capacity control model. The model can make channel decisions in conjunction with inventory and this is similar to the procedure shown in pure capacity allocation. The study has proposed an optimal policy basing on bid price that incorporates commission fee, price, and capacity. The numerical experiment results illustrate that introducing the channel distribution into airline revenue system can significantly improve the revenues and efficiently reduce the channel distribution cost for airlines. The numerical experiments demonstrate that airline revenues will increase more than 3% in a simple integrated system with two channels compared to the independent model. This study also analyses the reasons for improvements in different situations (such as multi-channels have better improvements than a single-channel and the model has a better match of channels and fare classes) so that management insights are obtained for airlines. Secondly, we analyse customer demand behaviours and we find that customers will experience demand transfer behaviours when facing channels. In the Internet age, due to more transparent information, the customer's transfer behaviour has been continuously expanded. For customers, the transfer of channels is more likely to occur than the transfer of fare classes because they do not need to pay for it. Therefore, it is necessary to establish a better revenue management model to consider the customer's channel transfer behaviour.In this part, we added customer channel transfer behaviour based on the original single-leg dynamic capacity allocation model that considers channel issues. We also developed the optimal policy for this model and made some numerical experiments. The numerical experiments demonstrate that the customer shift behaviour can influence the results of the model and subsequently the decisions of airlines. In the general numerical result, the new model can increase 1.23% than the above channel model. At the same time, through the analysis of the results, the airlines are provided with corresponding suggestions to face the customer's choice behaviour. For example, the airline needs to increase the customer's transfer rate through some methods, such as joining a price comparison network and increasing policy incentives. Thirdly, on the base of single-leg model, we propose a new network dynamic model to integrate network revenue management and channel distribution. To take a network structure airline, the airlines can make more revenue benefits comparing the single-leg method. Although the network dynamic model can make more improvements, the exact optimization is impossible for practical purposes because of the curse of dimensionality. Therefore, we use determined linear programming method for approximating to dynamic model. The numerical experiments demonstrate that the airline revenues can increase more than 3% in a simple network when the commission rate is 15% compared to the traditional network model. In addition to the above studies, the paper also summarizes the original literature on revenue management and channel issues and proposes future research directions.