Kannapha Amaruchkul The 4 th International Conference on Operations and Supply Chain Management, Hongkong & Guangzhou, Jul.25 to Jul.31, 2010 Value of Information in Air-Cargo Revenue Management Kannapha Amaruchkul School of Applied Statistics National Institute of Development Administration (NIDA) 118 Serithai Road, Bangkapi, Bangkok 10240, Thailand EMAIL: kamaruchkul@gmail.com Abstract: Air-cargo capacity is random and affected by the number of passengers carried, because both cargo shipments and passenger bags are carried in a belly of a plane. The fewer the passengers carried, the higher the cargo capacity. Current seats sold usually provide some information on the passengers carried, and consequently the cargo capacity. Records of the passengers carried and the seats sold are readily available in a passenger revenue management system. We propose mathematical models to evaluate monetary benefits, if different levels of information in the passenger revenue management system are shared by the cargo revenue management system. At a minimum level, an airline constructs a prior distribution of random cargo capacity from a historical record of passengers carried. At a higher level, the airline updates the distribution of cargo capacity based on the number of seats sold. A numerical example that illustrates the proposed methodology is also provided. Keywords: Air-Cargo; Revenue Management; Stochastic Model Applications I. Introduction Air-cargo operations are a significant source of revenue for passenger airlines, most of which carry cargo shipments in the belly of their aircraft. The air-cargo industry is expected to grow five percent annually during 2007–2027 [6]. The growth is propelled by global free trade, and the emerging implementation of supply chain management strategies, which emphasize on short lead times. Despite its importance, revenue management (RM) systems for managing air-cargo spaces are much less developed than those for controlling prices and availability of passenger tickets. Air-cargo RM is more complex than passenger RM. Cargo capacity is a multi-dimensional quantity; two important dimensions are weight and volume. The flight may be full with respect to the weight capacity but not the volume capacity, or vice versa. In contrast, the total passengers on board are constrained by a one-dimensional quantity, the number of seats on the plane. Moreover, cargo capacity is random and may be affected by various factors, such as the amount of fuel, the length of the runway, the weather condition at the departure time, and the number of passengers carried as well as their bags. The belly of the plane carries both passenger bags and air-cargo shipments. The cargo capacity depends on the passengers carried. The smaller the passenger load factor (the percentage of seats occupied), the larger the cargo capacity. For instance, with full passenger load of 290 seats, Airbus A330-300 has 10 tons of cargo capacity: If the load is 90%, then the cargo capacity could be increased by 2.6 tons. (The passenger weight--the average weight of the passenger plus both normal baggage allowance and excess baggage--is assumed to be 90 kilograms; see [10].) Since the capacity constraint depends on both the number of passengers and the cargo, “decisions for both passenger and cargo are interrelated and ideally should be coordinated by a single RM system” [14, p. 563]. Currently, air-cargo and passenger RM systems are separated; each locally maximizes its own expected contribution. The integrated (single) RM system would attempt to maximize the total contributions from both passenger and cargo, i.e., to attain global optimization. To operationalize the integrated RM system requires data to accurately characterize and model both passenger and cargo demand. The optimal total contribution achievable through the integrated system is at least the combined contributions from the two isolated systems. However, the centralized system requires a huge setup cost, and the operating cost of running the sophisticated centralized system may exceed that of running the decentralized system. It is important to determine whether the expected incremental contribution outweighs the fixed setup and operating costs. Information on both passenger and cargo is visible in the centralized system, whereas in the decentralized system, no information is shared. Between the decentralized and centralized systems, there is also a spectrum of systems with different levels of information sharing. Recall that the cargo capacity depends on the number of seats sold. At a minimum level, cargo RM could construct a distribution of the random cargo capacity from a historical record of passenger load, which is available in the passenger RM system. At a greater level of information sharing, passenger RM could provide the current number of passenger bookings to cargo RM so that the distribution of the cargo capacity could be updated periodically. The greater level of information sharing, the higher expected total contribution and the higher operating cost. In this article, we propose quantitative modeling to measure the monetary value of information sharing. We consider a cargo booking problem on a single-leg flight with the goal of maximizing the expected net contribution, which is the total margin contributions of accepted 370