CONCURRENT MULTI-PART MULTI-EVENT DESIGN REFRESH PLANNING MODELS INCORPORATING SOLUTION REQUIREMENTS AND PART-UNIQUE TEMPORAL CONSTRAINTS
Nelson III, Raymond Stanley
Sandborn, Peter A
MetadataShow full item record
Technology obsolescence, also known as DMSMS (Diminishing Manufacturing Sources and Material Shortages), is a significant problem for systems whose operational life is much longer than the procurement lifetimes of their constitute components. The most severely affected systems are sustainment-dominated, which means their long-term sustainment (life-cycle) costs significantly exceed the procurement cost for the system. Unlike high-volume commercial products, these sustainment-dominated systems may require design refreshes to simply remain manufacturable and supportable. A strategic method for reducing the life-cycle cost impact of DMSMS is called refresh planning. The goal of refresh planning is to determine when design refreshes should occur (or what the frequency of refreshes should be) and how to manage the system components that are obsolete or soon to be obsolete at the design refreshes. Existing strategic management approaches focus on methods for determining design refresh dates. While creating a set of feasible design refresh plans is achievable using existing design refresh planning methodologies, the generated refresh plans may not satisfy the needs of the designers (sustainers and customers) because they do not conform to the constraints imposed on the system. This dissertation develops a new refresh planning model that satisfies refresh structure requirements (i.e., requirements that constrain the form of the refresh plan to be periodic) and develops and presents the definition, generalization, synthesis and application of part-unique temporal constraints in the design refresh planning process for systems impacted by DMSMS-type obsolescence. Periodic refresh plans are required by applications that are refresh deployment constrained such as ships and submarines (e.g., only a finite number of dry docks are available to refresh systems). The new refresh planning model developed in this dissertation requires 50% less data and runs 50% faster than the existing state-of-the-art discrete event simulation solutions for problems where a periodic refresh solution is required.