Interactive evaluative model trading for resilient systems decisions

The Interactive Model-Centric Systems Engineering research effort is interested in developing knowledge necessary to leverage the increasing involvement of computational models in system design. One of the key means of leveraging a model-centric environment is the trading of models, which can reveal insights about the system that are difficult or impossible to see when considering only a single model. Prior work has demonstrated this technique on the value models used to determine the “goodness” of alternatives based on their performance and cost attributes. This paper extends the model trading paradigm to evaluative models: those that calculate the attributes themselves. The concept is demonstrated on a matching Space Tug case study with four different model implementations, each of which results in a different set of Pareto-efficient solutions. Analysis across implementations, as opposed to within a single model, reveals interesting design insights, of which some are physicsdriven and others are identified as model artifacts. © 2016 Adam M. Ross, Matthew E. Fitzgerald, and Donna H. Rhodes.