A Dynamic Programming Approach for Network Air Cargo Revenue Management

Please enable javascript in your browser.

Aref Kashfi, Samin | 2014

738 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 45514 (01)
University: Sharif University of Technology
Department: Industrial Engineering
Advisor(s): Modarres, Mohammad
Abstract:
This research aims to establish a novel approach to meet the capacity utilization challenge of the freight forwarders in order to maximize the expected revenue in a cargo network. The two-dimensional cargo capacities are assumed to be stochastic and limited in each leg. A stochastic dynamic programming model is developed to demonstrate the maximization of expected revenue over a finite planning horizon, in which each fare class request is to be either accepted or rejected. Although the main problem is computationally intractable due to multidimensional state variable, it can be decomposed into one-dimensional capacity control sub-problems for each leg. A heuristic algorithm is developed to provide the freight forwarder with dynamic acceptance/rejection policy based on the decomposition model. The performance of the heuristic algorithm is evaluated in comparison with estimator models of upper bound and lower bound. The results indicate the performance of the heuristic policy which can come up with improved feasible solutions
Keywords:
Revenue Management ; Air Transport ; Dynamic Programming ; Cordon-Based Pricing ; Capacity Control ; Dynamic Programming Decomposition