Yield Management Impacts on Airline Spill Estimation

Abstract
The correct estimation of spill, or passenger demand turned away, is an integral part of the determination of optimal aircraft capacities in the airline fleet assignment process. While making advances in the solution of the large-scale fleet assignment optimization problem, airlines have continued to use an aggregate approach to spill estimation. This aggregate approach ignores the effects of yield management practices that have been widely implemented by airlines during the past decade. In this paper, we illustrate the importance of incorporating the effects of yield management booking limits into the methodology used to estimate both the number of passengers spilled at a given aircraft capacity and their associated revenue value. We describe an approach to spill estimation that makes use of the detailed demand information provided by yield management systems, and we present recursive algorithms that can be used to obtain more accurate spill estimates in cases when multiple booking classes are used. Numerical examples are presented to illustrate the extent to which the outcomes of the different estimation approaches differ, suggesting that these differences can be large enough to have an impact on optimal fleet assignment.