SGER: Air-Cargo Revenue Management in the Presence of Intermediaries

该拨款为开发数学模型提供资金，以最大限度地提高航空公司货运业务的长期收入。将开发两种类型的模型。短期模型将侧重于航空公司是否接受直运客户的每一个预订请求的决定。预订决策是根据当前的容量承诺、历史预订概况和出发前的剩余时间动态做出的。这些决定还必须明确考虑到包裹的重量和体积以及对货运代理人的能力承诺，货运代理人作为独立的中间人，预先购买货物能力，并通过进行提货、储存、清关、包装和在当地向收货人交付而产生额外的需求流。中期模型将侧重于航空公司关于向货运代理商提供销售的容量和价格组合的决定。短期和中期的问题将制定为马尔可夫决策过程和最佳解决方案的结构特性将被确定，在可能的情况下。算法产生良好的启发式解决方案和界限的航空公司的最佳预期收入将开发和其准确性将通过广泛的数值实验进行测试。如果成功，这项研究的结果可能会影响航空公司思考和管理货运能力的方式。由于航空货运收益管理是学术研究人员的一个未知领域，拟议的工作也将为未来的研究开辟新的途径。它将导致新的模型和计算技术的发展，为边界和测试大规模随机优化问题的启发式解决方案。本研究的另一个目标是展示使用基于随机优化的解决方案管理货运能力的潜在好处，并产生新的管理见解。最后，建议的工作将有助于培养研究生和本科生。

This grant provides funding for the development of mathematical models for maximizing airlines' long-run revenue from cargo operations. Two types of models will be developed. The short-term models will focus on the airlines' decisions about whether or not to accept each booking request from direct-ship customers. Booking decisions are made dynamically based on current capacity commitments, historical booking profile, and time remaining until departure. Such decisions must also explicitly consider weight and volume dimensions of the package and capacity commitments to freight forwarders who, acting as independent intermediaries, pre-purchase cargo capacity and generate additional streams of demand by performing pick-up, storage, customs' clearance, packing, and local delivery to consignees. The medium-term models will focus on the airlines' decisions about which combinations of capacity and price to offer for sale to the freight forwarders. Both short and medium term problems will be formulated as Markov Decision Processes and structural properties of the optimal solutions will be identified, where possible. Algorithms that generate good heuristic solutions and bounds on the airline's optimal expected revenue will be developed and their accuracy will be tested through extensive numerical experiments. If successful, the results of this research may affect the way in which airlines think about and manage cargo capacity. Since air-cargo revenue management is an uncharted area for academic researchers, the proposed work will also open up new avenues for future investigation. It will lead to the development of new models and computational techniques for bounding and testing heuristic solutions to large scale stochastic optimization problems. Another goal of this research is to demonstrate the potential benefits of using stochastic optimization based solutions for managing cargo capacity and to generate new managerial insights. Finally, the proposed work will contribute to the training of graduate and undergraduate students.