Sensitivity analysis on dependencies in dynamic availability models for large systems

Assumptions about independence are made when modeling the availability and reliability of large systems. The challenge, then, is to show that the loss of fidelity due to the independence assumptions is not significant. A sensitivity analysis that can cover all the situations for a dynamic system is the focus of this study. We attack this problem by conducting a series of sensitivity analyses on the dependencies identified in the system hierarchy. By showing the importance of the dependencies that have been modeled and the insignificance of the differences caused by the independence assumptions, the appropriateness of the modeling approach is demonstrated. This sensitivity analysis thus earns the analysts confidence in the model.