Application of the modal superposition technique combined with analytical elastoplastic approaches to assess the fatigue crack initiation on structural components

Abstract

Local fatigue approaches, such as, the stress-life, strain-life or energetic approaches defines a framework to estimate the fatigue crack initiation from notches of structural details. Various engineering structures, such as, bridges, wind towers, among others, are subjected to cyclic dynamic loadings which may substantially reduce the strength of these structures. Nowadays, the structural systems tend to be more complex being necessary to find computationally efficient solutions to perform their fatigue analysis, accounting for dynamic actions corresponding to long complex loading events (e.g. diversity of trains crossing a bridge), mainly if local approaches are envisaged. Thus, this paper aims at presenting and validating a generalization of a methodology based on modal superposition technique, for fatigue damage parameters evaluation, which can be applied in fatigue analysis using local approaches. This technique was applied recently in the context of fatigue crack propagation based on fracture mechanics, although it can be extended to compute the history of local notch stresses and strains at notches. A very important conclusion is that the technique can be explored for the case of local confined plasticity at notches whenever the global elastic behaviour of the component prevails. Local submodelling can be explored with this technique to avoid the necessity of large computational models. Local models are only needed to be run under linear elastic conditions for the selected modal shapes of the structure, being the local time history of fatigue damage variable computed by modal superposition for each loading event. That time history may be further post-processed for elastoplastic conditions using Neuber or Glinka's analyses. Comparisons with direct integration elastoplastic dynamic analysis confirmed the feasibility of the proposed approach.