Digital twins are virtual representations of subsystems within a system of systems. They can be utilized to model and predict performance and condition degradation throughout a system’s life cycle. Condition based mainte-nance, or the performance of system maintenance based on the subsystem states, is often facilitated by the implementation of digital twins. An open challenge is selecting the subsystems that require digital twins. We establish a generic process for determining a set of priority-based system components requiring digital twin development for condition based maintenance purposes. The priority set, which we term the “triage” set, rep-resents the set of components that when monitored through a digital twin lead to the greatest increase in total system reliability and simultaneously represent the minimal cost set of components for implementing a digital twin. While we focus our process on an unmanned underwater vehicle (UUV), where we frame the design problem as a multiobjective optimization problem utilizing experimentally determined data and metrics from the model of a real UUV system, the process is generic enough that it could be utilized by any system looking at cost and reliability estimates for leveraging digital twin technology.