Numerical aspects of micromechanical fatigue simulations of FRP - Limits of degradation and adaptive cycle jump approach

Abstract

Finite element based micromechanical fatigue damage models are a promising approach for getting a deeper insight into the underlying physical principles of fatigue damage in fibre reinforced plastics (FRP). The fatigue damage behaviour of FRP is very progressive and accompanied by load redistributions. Capturing these effects requires a transient analysis in conjunction with the usage of progressive damage models. In addition to this, for economic needs, the FE simulation should be computationally efficient. To meet the different and partly contrasting demands on fatigue damage analyses, the most models make use of numerical tweaks. For example, considering the fully damaged state of a material point, the material parameters are tried to be reduced to zero, but – for numerical convergence and stability reasons – they are degraded to a small fraction of the initial properties, only. Additionally, besides the gradual material degradation depending on the number of applied load cycles, also sudden failure within a load cycle due to exceeding the strength limit of the material may takes place which, nevertheless, is not considered in each micromechanical fatigue damage model. Although a transient analysis tracking the cycles of the fatigue load is necessary, passing through every single cycle is computationally extremely expensive. A solution to this challenge is the cycle jump approach introducing another numerical parameter: The cycle jump determines the number of cycles over which the damage state is extrapolated. Regardless of the underlying damage model quality, the aforementioned parameters might have such a high influence that they might spoil the overall model quality and predictability. Thus the influence of the following numerical parameters on the obtained progressive damage behaviour of FRP is investigated for transverse tensile fatigue loads: 1) whether or not the model considers static failure within a simulated load cycle, 2) the degree of material property degradation after sudden failure and 3) the size of the cycle jump. The simulated damage behaviour is evaluated using experimentally observed crack patterns published in the literature. The results reveal a significant influence of the degree of material degradation and of the cycle jump on the simulated matrix crack formation at both, higher and lower fatigue loads. Static failure within a simulated load cycle primarily affects the damage behaviour at higher fatigue loads. The presentation gives recommendations of the parameter choice for plausible progressive fatigue damage simulation results. Regarding the cycle jump, an adaptive algorithm is proposed and implemented. This approach leads to plausible fatigue damage results paired with a significant reduction of computation time comparing to a cycle-by-cycle analysis.