Simulation has become an indispensable support in the design process. This is particularly evident in the design of complex, multidomain, and multiphysics systems. In these cases in particular, though, the setup of the scheme to be simulated is a challenge in itself. In this paper, we propose an approach to the selection of the right level of detail of the models given the design objectives. This method is based on stochastic analysis and the results are lumped in indices that allow for a ranking of available models of each component. In detail, the main contribution is the formalization of a method to describe difference between different models of the same component. A case study is proposed to demonstrate the concept via the design of the overcurrent protection of an induction machine.