Affordability Measurement and Prediction Technologies

This proposal builds upon work currently underway for the Office of Naval Research's Acquisition Reform Office. This body of work known as SNAP (Simulating New Acquisition Processes), consists of five interconnected simulation models that make up a "virtual acquisition process." The models, developed using the system dynamics methodology, can quantify the long-term tradeoffs among cost, performance and schedule measures. SNAP output includes key decision variables allowing program managers to test the consequences of alternative "what-if?" scenarios and to place high confidence in the resulting priorities for affordability actions. Although SNAP development is well underway, SNAP models lack funding to incorporate an essential feature -- Evolutionary Algorithms (EAs). EAs mimic biological evolution by repeatedly "mutating" selected parameter values and simulating the model to produce a generation of project solutions. During simulation the EA automatically evaluates tradeoffs among cost, schedule and performance variables, using feedback to alter both tasks and resources in a search for the most effective result. The addition of EAs to SNAP will result in an incredibly advanced affordability prediction tool that will not only correctly predict the total cost of ownership for any system but will also enable program managers to evolve better and better program plans with CAIV (Cost as an Independent Variable) guidelines.