Probabilistic multiscale modeling of 3D randomly oriented and aligned wavy CNT nanocomposites and RVE size determination

This paper presents a probabilistic multiscale approach to model the random spatial distribution of local elastic properties arising from the heterogeneous waviness and orientation of CNT fillers within a 3D microscale continuum representative volume element (RVE) of a CNT-reinforced polymer matrix. The proposed direction-sensitive 3D Karhunen-Loève expansion (KLE) provides the basis to simulate stochastic variations in the CNT orientations rather than assuming a perfect alignment. Computational homogenization analyses are carried out to investigate the effects of the statistical parameters of random CNT waviness on the microscale continuum RVE size. The proposed probabilistic multiscale modelling framework allows to consider the uncertainty associated with the use of CNT nanocomposites for various applications.