Sensitivity analysis of material parameters on the micro-scale stress distribution in fiber reinforced composites

Knowing the mechanical behavior of fiber-reinforced composites are important since unidirectional (UD) composite materials have been widely used in many industries. Different methods in macro-scale and micro-scale have been used to determine the properties of the composite. While macro-scale methods fail to predict properties under transverse loading, many studies were done in micro-scale methods where the fiber and matrix are modeled separately. Finite element method is the most widely used method for micro-scale analysis. It is known that matrix properties dominate the failure behavior of the UD composites. Since these distributions are affected by the composite material properties, it is important to see the relation between them. The stress distribution is highly dependent on the distribution of the fiber. Therefore, random packing methods were developed to capture the real composite structure. With each run, fiber locations changes. Understanding the stresses and concentrations due to selected material properties is possible with a sensitivity analysis. In this thesis, the effect of material parameters on micro-scale stress distribution in fiber reinforced composites in transverse loading was analyzed. Different micro-scale models were generated due to generated fibers in the matrix being different than the actual structure. The models were prepared and solved by a commercial finite element software, ABAQUS. Material parameters that can be selected by the designer or engineer were analyzed within a preliminary analysis and were defined to the models. Their results were compared to see their importance in the sensitivity analysis. It was seen that fiber Poisson's ratio does not have any meaningful effect on stress concentration within the ranges given in the study. Chosen material parameters and their values taken from the literature were defined to the models. To calculate the relation between the stress concentration results gathered and the material parameters, a commonly known statistical analysis method called parametric correlation was used and correlation coefficients were interpreted. It was seen that composites Young's modulus ratio has lower relation with stress concentration than fiber's volume ratio. A parameter called overstressed volume percentage was introduced in the thesis. The correlation coefficient of this parameter with the input parameters was also calculated and interpreted.