Computational methods for multiaxial fatigue analysis

The computational fatigue analysis tool presented here is based on available models for multiaxial cyclic plasticity, notch correction, cycle counting, and damage estimation. Improvements and additions to current models are proposed and included into a complete fatigue analysis code.; First, an integrated notch correction and plasticity model is developed. The model computes multiaxial stresses and strains from local or nominal strain measurements. It is shown that Köttgen's notch correction method may be combined with any plasticity model to create a single procedure for determining notch stresses and strains. As an example, an algorithm combining an infinite-surface Mróz model with Köttgen's notch model is presented.; Next, two published data sets are analyzed using McDowell's method for deriving measured values of the backstress rate from experimental data. Several cyclic plasticity models are compared by rating their ability to predict the direction and magnitude of the backstress rate. The stress rate is shown to be superior to other models in its ability to predict the direction of the backstress rate. No model is found that can uniquely predict the magnitude of the backstress rate for all loadings, though a model that outperforms the others is proposed.; Damage assessment for multiaxial loading is considered next. Current uniaxial rainflow methods are shown to be inadequate for counting cycles during multiaxial loading. A multiaxial rainflow algorithm is then presented. The algorithm correctly identifies minima and maxima on auxiliary channels while counting cycles on a single channel per the critical plane method.; A robust numerical implementation of damage models for use in the critical plane method is also discussed. An algorithm is proposed that establishes an implicit relationship between the damage parameter and the unknown life, resulting in a relation that needs only uniaxial strain-life and stress-strain material properties as input.; Finally, the predictions of the analysis tool are compared to experimental data published in the literature. The tool is shown to correctly predict strains for notched shaft experiments, stresses for thin-walled tube experiments, and lives for thin-walled tube experiments. However, the results also show the limitations of the method.