Numerical Simulation Of Progressive Damage Of Fiber-reinforced Composites

Fiber-reinforced composites have been used in numerous key structuralcomponents because of their high strength and damage resistance. Ironically, peoplehave not understood quite well some of the essential mechanical behavior of thecomposites in service due to their complex microstructures. It has been observed thatstructural components made of unidirectional fiber-reinforced composites can retaintheir load-bearing capacity after local damage initiates. Understanding of thepost-damage mechanical behavior of the composites, however, is rather inadequatecompared to that concerning the damage-free mechanical behavior of the materials. Inorder to exploit the potential of fiber-reinforced composites, it is necessary for us toextend our limited knowledge of the post-damage behavior of the composites.This thesis is aimed at constructing a progressive damage model for unidirectionalfiber-reinforced composites to simulate their mechanical behavior after the initiation ofdamage. Longitudinal damage dominated by fibers and transverse damage dominatedby matrix are considered. The maximum stress criterion and Puck’s action planefracture criteria are chosen as the initiation criteria for the longitudinal and transverseprogressive damage, respectively. The second-order damage tensor in theMurakami-Ohno model is employed here to represent the accumulation of damage inthe composites. A damage evolution law is hereby built based upon the exponentialstrain softening behavior and the energy equivalence hypothesis proposed by Sidoroff.Dependence of the finite element simulation results on mesh is eliminated through theintroduction of the crack band model. The irreversible nature of damage andone-sidedness effect of microcracks perpendicular to the fibers are taken into account.The corresponding damage model is implemented via coding a user-defined materialmechanical behavior (UMAT) subroutine of Abaqus. Element tests are conducted inAbaqus/Standard to verify the model as well as the code. The results demonstrate thatthe progressive damage model is correctly achieved by this code.