An alternative optimization model and robust experimental design for the Assignment Scheduling Capability for the Unmanned Aerial Vehicles (ASC-U) simulation

Abstract

The Modeling, Virtual Environments, and Simulations Institute (MOVES) and the United States Army Training and Doctrine Command (TRADOC) Analysis Center (TRAC) at the Naval Postgraduate School, Monterey, California, developed the Assignment Scheduling Capability for Unmanned Aerial Vehicles (ASC-U) discrete event simulation to aid in the analysis of future U.S. Army Unmanned Aerial Vehicle (UAV) requirements. TRAC selected ASC-U to provide insight into the programmatic decisions addressed in the U.S. Army UAV-Mix Analysis that directly affects future development and fielding of UAVs to include the Future Combat System. ASC-U employs a discrete event simulation coupled with the optimization of a linear objective function. At regular intervals, ASC-U obtains an optimal solution to an assignment problem that assigns UAVs to mission requirements that are available or will be available at some time in the future. This thesis presents an alternative optimization model, explores 23 simulation factors, and provides sensitivity analysis for how UAV coverage may degrade in the presence of adverse random events. Integer programming, experimental design, and an innovative Optimized Flexible Latin Hypercube (OFLH) design are used to evaluate a representative sample from an Army 2018 scenario. The conclusions suggest the following: the alternative optimization model developed in this thesis can successfully maximize ASC-U value without the use of a heuristic; smaller optimization intervals do not guarantee higher total value when the heuristics are included; to maximize total value, Early Return should be set to FALSE and Secondary Areas should be set to TRUE; an OFLH is valuable for robust analysis of simulation models containing many factors; and as the model factors change over predefined ranges, the solution quality is consistent.