Author: Stephane J-C Bratu

Publisher:

Published: 1998

Total Pages: 124

ISBN-13:

A connecting passenger occupies a seat on each of the flight leg of his itinerary. Moreover, for a given fare class, the fare of a connecting passenger is lower than the sum of the fares of the local passengers on the traversed legs. If the demand is high, giving availability to a connecting passenger may displace local passengers and the airline would lose revenue. The objective of this thesis is to evaluate methods that airlines can use to better estimate the network revenue value of connecting passengers for the purpose of determining seat availability. In this thesis we analyze and compare two different ways of estimating the network revenue value of the connecting passengers. The first approach consists of estimating the displacement cost of the connecting passenger on all the traversed legs by the shadow prices associated with the capacity constraints of a network linear program (LP). The second one is a prorated fare convergence technique developed in this thesis. The fares of the connecting passengers are prorated on each of the traversed legs using an estimation of the expected marginal revenue of the last seat on the legs. The existence and uniqueness of the limit for each prorated fare sequence are also proven. We have compared the performance of different seat inventory control models that incorporate these two network revenue estimation techniques. The optimization/booking simulation uses demand forecasts from an airline's Yield Management historical database. The seat inventory control methods that use the network revenue value concepts perform up to 1.50% better than the existing fare class control approach at a high demand scenario (82% average load factor). Moreover, the prorated fare convergence technique performs better than the LP shadow price displacement cost approach especially if the demand is controlled by a bid price mechanism. Indeed, for a high demand scenario and a relatively high number of reoptimizations along the booking process, the prorated fare convergence method performs 0.12% better than the shadow price approach for a bid price control mechanism. Finally, the revenue difference between the two methods is both significant and robust with respect to demand variations.