Using Advanced Tabu Search Approaches to Perform Enhanced Air Mobility Command Operational Airlift Analyses

Abstract

The descriptions below are intended to complement and expand upon the abstract contained in the cover page (Report Documentation Page, Standard Perform 298) for this final report.Each of the doctoral dissertations and final Masters degree documents, either reports are theses, completed during the report period, will be briefly summarized, where appropriate, giving the most significant advancements and conclusions contained in each document. The descriptions will be given in the order of presentation in the references given at the end of this report. First, we will describe the 4 dissertations completed during report period. These dissertations are available through The University of Texas at Austin.The work by Roesener [1] and Roesener et al. [27] documents an advanced tabu search (TS) approach to both the static and dynamic airlift loading problem. It represents an extension of Harwig's seminal work [5,19] on the two-dimensional orthogonal bin packing problem. However, in Roesener's research, additional constraints associated with military aircraft such as weight and balance had to be considered. Roesener's methodology was shown to be significantly superior to standard practices employed at the time of its development.The research of Yang [2] and Yang et al. [29] was the first documented research in applying TS to the correction of disruption management problems in the civilian airline industry. This highly innovative research employed both pure TS and a hybrid mathematical programming-TS approach to the aircraft grounding problem and to the reduced airport capacity problem to achieve superior results in both cases. While Kinney [8] and Kinney et al. [20] also used a hybrid methodology, the problems approached by Yang were considerably more restrictive in the context of the associated constraints.The research documented in Ciarleglio [3] and Ciarleglio et al. [22,23] constitute a major breakthrough and departure from previous work in TS. This research resulted in a fundamental reusable framework, the modular adaptive self learning TS (MASTS) for the application of advanced TS techniques. Surpassing the purview of just being a