A space-time upscaling technique for modeling high-cycle fatigue-damage of short-fiber reinforced composites

Characterizing short-fiber reinforced polymers under high-cycle fatigue loading is a tedious experimental task. To reduce the necessary experiments to a minimum, we introduce a computational strategy involving a mean-stress dependent fatigue-damage model for the stiffness degradation in short-fiber reinforced polymers. The key challenge in these materials is their inherent anisotropy which makes the necessary mechanical characterization process rather time-intensive, in particular for long-time experiments required for fatigue tests. Computational multiscale approaches may reduce the necessary mechanical tests to a bare minimum, offering significant savings in expense. We propose a mean-stress sensitive model to simulate the stiffness degradation in short-fiber reinforced composites subjected to fatigue loading. We start with a model formulated in time space and provide a multiple-set scale-bridging approach to arrive at a computationally efficient effective model. For a start, we describe a high-accuracy cycle-jump technique which permits us to simulate a large number of cycles, required for high-cycle fatigue. In a second step, we apply a model-order reduction in space to arrive at an effective model on component scale. Finally, we rely upon a fiber-orientation interpolation technique to produce an effective material model which covers all relevant fiber-orientation states throughout the component. Our approach utilizes a recently introduced compliance-based damage model for describing the stiffness degradation of the matrix material. We demonstrate the capability of the computational multiscale model to reproduce the stiffness degradation in fatigue experiments for different orientations, stress amplitudes, stress ratios between R = −1 and R = 0 and geometries with different notches.