Statistical investigation of a field coupling to random twisted pair using Design of Experiment

In this paper, the theory of non-uniform multiconductors transmission lines (NUMTL) is used to calculate the radiated susceptibility (RS) in a random differential twisted wire pair (TWP) located above a ground plane. To model the random variations in the geometry of the twisted pair, a predictive statistical approach based on the Design of Experiment (DoE) is proposed. Unlike other stochastic methods such as Monte Carlo method, the advantage of DoE is its efficiency in terms of computation and storage resources. Indeed, it allows with limited number of simulations to interpret concretely the relationship between the output (induced current or voltage) and inputs (randomly varying geometric factors) through the response surface model. The robustness of the model is validated by a statistical test based on analysis of variance ANOVA.