On the challenge of designing a robust military force

Abstract

In the military field, detailed simulation models are often used to get insight in the consequences of choices regarding the composition of a force and the extent to which it can be kept deployable. With detailed models, the main idea is to capture all real-world complexity relevant for decision-making processes. However, detailed models tend to be ‘bad’ in the evaluation of uncertainty in a system. This can be problematic, as the outcomes of detailed models usually depend on numerous dubious assumptions that are too buried to be understood by hand. As simplified models can better cope with uncertainty, this study examined the combined use of a detailed model and a simplified model (this approach is also known as ‘multi-resolution modelling’) to improve the robustness of a naval force support system. In order to work with models with different levels of detail, an existing detailed model that reflects certain elements of the force support system of the Dutch Royal Navy is used to develop a new simpler model.

The key takeaway of this study is that multi-resolution modelling can be considered useful as the detailed model and the simplified model were found to be mutually dependent in their design, validity, and use to improve the robustness of a naval force support system. As the detailed model contains a more accurate depiction of the system of interest, it formed together with a set of aggregation functions and cross-validation tests the basis for the design and validity of the aggregated model. The detailed model also remains important for the evaluation of the real-world effectiveness of promising ‘high-level’ actions discovered by the simplified model. Vice versa, by means of exploratory analysis, the simplified model revealed some potentially problematic assumptions in the detailed model which has implications for its design and validity. Also, the simplified model gave insight into (a) problematic scenarios that lead to a low sailing availability, (b) patterns in the robustness of policies across different usage profiles, and (c) potential leverage points to increase the robustness of the naval force support system. This study concludes with a number of scientific research recommendations involving challenges in strong consistency and the composability of models at different levels of detail.