In this paper, we discuss the defect evaluation in a conductive material, by a method based on stochastic finite
elements. This later allows then to obtain a hazard distribution of one or more physical properties of the material. To do this,
we have considered two kinds of distributions of random variables; Gaussian and Lognormal ones. Those variables will
represent the distribution of the electrical conductivity in the area of the defect to be investigated and an evaluation of the
impact on the reliability of the device will be done. A stochastic finite element calculation code is developed
and the variation of the impedance is then deduced by doing a scan along the surface of the conductive plate. The computational was performed in the two cases successively, Gaussian and Lognormal distributions. A post-treatment is
performed by the calculation of the reliability index. This results Allowed us to evaluate the probability of failure
associated with the nature of the fault present on the conductive plate for both cases of random variables.
The study is completed by the evaluation of the impact of the defect thickness on the reliability index using lognormal
distribution of the random variable. The results obtained appear to be interesting.