Survivability-based force optimization: Enhancements to a mix model methodology

The purpose of this investigation is to improve force structure modeling and decision-making recommendations for military force planners. Operations Research Analysts at the U.S. Army Training and Doctrine Command (TRADOC) Analysis Center at White Sands Missile Range (TRAC-WSMR), New Mexico use a force mix methodology to develop force structures. Model data is derived from CASTFOREM, a stochastic combat model. CASTFOREM simulations are based on single system upgrades or single additions to a military force and do not address multiple force modifications. The conceptual problem explored in this research is how to model the interactions that occur when multiple force modifications are made simultaneously.; A military force's effectiveness is based on two factors: its own survivability and its ability to defeat an opposing force. These two factors are not independent and are instead correlated. With U.S. military superiority, the ability to defeat an opposing force is presupposed. By optimizing on survivability, the ability to defeat an opposing force is also improved.; This dissertation describes a methodology that assumes the ability to defeat an opposing force and optimizes force survivability. Analysts have been unable to model the synergy between weapon systems and auxiliary systems, such as sensors, in a mix analysis. The potential synergy between weapon systems and auxiliary systems is also modeled in this methodology. Optimizing on survivability allows the benefits of auxiliary systems to be included in the model. Synergism between weapon systems often implies diminishing returns of incremental force improvements. To adjust for diminishing marginal returns, benefits for incremental improvements are normalized.; Processing of model input from CASTFOREM is resource intensive. Reducing the effort required to manipulate the data increases time available for analysis. Heuristics are compared to determine the best algorithm for data processing. The heuristics address the condition of statistically overlapping groups, resulting from ANOVA testing and multiple comparison procedures. The best heuristic was programmed to expedite the process, resulting in savings of 10 staff-hours per CASTFOREM simulation. The research and implementation of this heuristic were the first steps in the process improvement of this problem.