Exploratory Analysis to Address Deep Uncertainty – Using Calibratable System Models for Exploratory Simulation of Complex Missions

Traditional optimization methods and sometimes even the underlaying concept of experimentation design for optimization problems are not always successful in complex solution spaces. This is not only driven by the complexity itself, but also by the deep uncertainty in estimating the model parameters. Applying exploratory analysis to better understand the solution space has been recognized as a promising approach for several years, but with the new generation of computers the community now has the means to practically apply it as well. We define the terms, derive a conceptual model, provide an application case from the defense domain, and show the possibility to generalize the results, namely that parameter sweeps are possible and lead to deeper insights of the solution space, allowing for stable and robust solutions that often are more important in practice than optimized point solutions.