Optimization models for allocation of air strike assets with persistence

Abstract

This thesis addresses the critical process of assigning strike aircraft to targets once the targets have been identified: How do we optimally employ available aircraft and weapons on the current set of targets, and how can we modify a previously optimized assignment list to face changes in the tactical situation? Our contribution to the strike-planning problem includes (1) a static allocation model in which each aircraft makes at most one sortie during the planning time horizon, (2) a dynamic model in which each aircraft may make more than one sortie during that horizon, and (3) extensions of these models with "persistence incentives," which discourage major plan changes in the results when partial but important changes in the tactical situation necessitate reoptimization. These optimization models are mixed-integer programs that solve in seconds on a personal computer for realistic scenarios with three weapons types, 156 aircraft at seven bases, and 100 potential targets. In a scenario in which two new high-priority targets arise and must be added to an air tasking order with eight original targets, persistence incentives reduce the number of major plan changes from five to two.