Mesh Selection for Progressive Failure Modeling of Carbon Fiber Panels in Mode III Shear Using an Explicit Finite Element Solver

Abstract

Modeling progressive failure of composite materials can be a challenging task. It is further complicated by mesh dependency of finite element solvers when implementing the Hashin criterion. As one continues to refine the mesh, the finite element analysis (FEA) continues to yield varying solutions; hence there is no converged solution for continuous mesh refinement. This is due to mesh dependency when modeling strain softening. The FEA package Abaqus attempts to mitigate this but the issue is not eliminated. Methods that address mesh dependency include experimental validation and numerical analysis. Experimental validation tailors the mesh to match a specific result. This mesh would then be applicable to other configurations with the same geometry and loading; however, experimental validation with increasingly complex parts for FEA is costly and time consuming. Numerical approaches to mesh selection do not require experimental validation; however, these methods may be computationally expensive depending on the analysis. A mesh selection strategy that does not require experimental validation, while computationally efficient, should be implemented for design purposes. This study investigates a mesh selection strategy based on a converged elastic solution; the coarsest mesh that converges to a solution in the linear-elastic portion of the material response is chosen for analysis. Previous studies using an implicit solver yielded good results for out-of-plane loading conditions; however this procedure has not been implemented for explicit solvers. In this study, an investigation was conducted to determine the appropriate mesh to model progressive damage for notched, carbon fiber composite panels in out-of-plane shear (mode III) using an explicit solver in Abaqus/Explicit. This study analyzed 20 ply thick panels and considered three stacking sequences: 10%, 30%, and 50% zero degree plies. The procedure initially disabled damage and identified the coarsest mesh that approached a converged elastic solution. Using this mesh, damage was enabled and the models were run with loading proceeding through damage initiation until failure. The panels’ material response were extracted from the finite element (FE) model and filtered in order to determine their maximum load-carrying ability. The FE predicted maximum loads were then compared to corresponding experimental data in order to validate the mesh selection procedure. This process is not limited to out-of-plane shear; the potential for this mesh selection method would allow for progressive failure simulation to be more applicable in the design process of composite structures instead of post-damage analysis.