Air cargo revenue and capacity management

The air cargo industry has substantially grown over the past few years, driving the need of a structured environment with the explicit goal of maximizing revenues. The air cargo supply chain is composed of shippers, freight forwarders, and airlines. The shippers send their shipment to freight forwarders, who are responsible for contacting the airlines and procuring space to ship the cargo according to the shippers' needs. Currently, the process takes more time than needed due to a lack of coordination between freight forwarders and airlines; it is said that an integrator, a freight forwarder which owns its own fleet, moves an international shipment two or three times faster than a traditional freight forwarder/airline team [10].; The scope of this thesis is to propose a structured methodology for improving the airlines' and freight forwarders' actions when confronted with accepting demand and acquiring capacity respectively. We develop methods to tackle two air cargo revenue management problems: space allocation and show-up rate estimation.; The space allocation problem is defined as distributing the available capacity for free sale among incoming cargo bookings over the booking horizon such that the revenue at the end of the booking period is maximized. We use bid price methods to accept/reject incoming bookings: if the rate of the booking is lower than the bid price value then the booking is rejected. We show that a good approach to deal with the demand lumpiness encountered in the cargo industry is to split the cargo into two categories: small, which contains small packages and mail, and large, which contains the bulk of commercial cargo. Whereas the small cargo demand behavior can be approximated with the passenger demand behavior, and techniques from the passenger sector could be adapted for the small cargo business, the large cargo demand behavior shows similarities to wholesale retail and calls for different methods. We model the small cargo revenue management problem using a model from the passenger business and propose a new algorithm to solve it, which had a superior running time among the few algorithms known to solve the same model in the passenger business. The large cargo revenue management problem is solved via dynamic programming. In our simulations, when the demand is extremely lumpy, i.e., the cargo loads vary widely, the conjugated solutions from the two models result in up to 60% more revenue than the first come first serve method used in practice.; The second air cargo revenue management problem is estimating the cargo show-up rate, which is the ratio of cargo handed in at departure over bookings on hand. The show-up rate is used in the overbooking models to estimate the capacity available for free sale before the departure date. In the passenger business, the current practice is to estimate the show-up rate based on a Normal distribution. We show the Normal distribution is not suitable for the cargo business and propose a discrete distribution based on wavelet estimation. In a simulation study conducted for a set of real world demand date, the average yearly savings resulting from using the discrete estimator for a fleet of 300 flights per day and an average of cargo capacity per departure of 13,000 kilograms was {dollar}16,425,000.; Besides the airline's revenue management problems, we solve the capacity management problem for the freight forwarder. The freight forwarder bids for cargo space on flights offered by the airlines several months before the actual departure date of the aircraft. The committed capacity has to be confirmed a few days before the departure date. In spite of its importance, there are no known solutions to this problem. We propose modeling the problem as a perishable inventory problem with backlog and lead time. The lead time is the time between which the freight forwarder orders the capacity and the time the aircraft is scheduled to take off. During this time, new demand from shipp...