Applying the imprecise Dirichlet model in cases with partial observations and dependencies in failure data

Imprecise probabilistic methods in reliability provide exciting opportunities for dealing with partial observations and incomplete knowledge on dependencies in failure data. In this paper, we explore the use of the imprecise Dirichlet model for dealing with such information, and we derive both exact results and bounds which enable analytical investigations. However, we only consider a very basic two-component system, as analytical solutions for larger systems will become very complex. We explain how the results are related to similar analyses under data selection or reporting bias, and we discuss some challenges for future research.