Description
Accurate failure predictability in composite structures is a necessity when they are used in aerospace and automotive industries where safety is of utmost importance. Proven benefits of numerical simulations over experimental tests in overall efficiency and convenience has seen an increased focus on virtual design and testing methods. But, even with a significant progress in formulation of novel techniques, failure analysis of complex composite structures is still a major challenge. The material heterogeneity primarily causes these structures to fail in multiple failure modes but also gives rise to stochastic variations in characterization of geometric and material properties. Designs that exhibit competing failure mechanisms are sensitive to such variations and can show large deviations in progressive failure responses and make simulations unpredictable depending on the model fidelity and its capability to capture physics accurately. Conversely, it is possible to purposely make structural designs sensitive using optimization techniques and use it as validations tests to assess model capabilities and clearly understand the underlying failure mechanisms. Past studies have demonstrated that configurations exhibiting competing failure mechanisms can be attained using design optimization that maximizes damage dissipation energy. Such optimization techniques with appropriate constraints can be used to increase sensitivity of designs for certain assumptions and random variations. Whereas, it can maximize errors in the model and show weaknesses in analysis formulations. This work initially investigates progressive failure response in composite bolted joints under pure bearing loads for variations in edge distance to diameter ratio and different laminate stacking sequences. A three-dimensional model is analyzed with subroutine VUMAT using finite element analysis package ABAQUS™. Further, it demonstrates the use of numerical optimization to maximize the sensitivity of the failure response using a simplified two-dimensional formulation of the problem. The model is sampled and analyzed for maximum energy dissipation and maximum peak load and the optimal designs are subjected to small variations in geometry, material and simulation parameters. The increased sensitivity of the failure response to these parameters makes it easier to elucidate the contribution of failure mechanisms and the result of competing load paths on the failure response of the model.
