Multi-scale statistical approach of the elastic and thermal behavior of a thermoplastic polyamid glass fiber composite

The strong heterogeneity and the anisotropy of composite materials require a rigorous and precise analysis as a result of their impact on local properties. First, mechanical tests are performed to determine the macroscopical behavior of a polyamid glass fiber composite. Then we focus on the influence of the heterogeneities of the microstructure on thermal and mechanical properties from finite element calculations on the real microstructure, after plane strain assumptions. 100 images in 10 different sizes (50, 100, 150, 200, 250, 300, 350, 400, 450, 600 pixels) are analysed. The influence of the area fraction and the spatial arrangement of fibers is then established. For the thermal conductivity and the bulk modulus the fiber area fraction is the most important factor. These properties are improved by increasing the area fraction. On the other hand, for the shear modulus, the fibers spatial arrangement plays the paramount role if the size of the microstructure is smaller than the RVE. Therefore, to make a good prediction from a multi-scale approach the knowledge of the RVE is fundamental. By a statistical approach and a numerical homogenization method, we determine the RVE of the composite for the elastic behavior (shear and bulk moduli), the thermal behavior (thermal conductivity), and for the area fraction. There is a relatively good agreement between the effective properties of this RVE and the experimental macroscopical behavior. These effective properties are estimated by the Hashin-Shtrikman lower bound.