The random nature of the microstructural attributes in fabricated composite material gives rise to developing a representative volume element which incorporates the aforementioned randomness. In this paper we present a methodology for generating a realistic volume element with discrete defects that are statistically equivalent random defects in a Sic/SiNC samples. Data extracted from two dimensional cross-sectional micrographs leads to the geometric statistical characterization of stochastic micro-scale discontinuities. Three dimensional discrete ellipsoidal voids are then introduced to a defect-free eight-harness weave model based on the obtained statistical characteristics, using a Monte-Carlo simulation. The geometric parameters of the generated defects result in target statistics equivalent to that of the defects in the micrographs.