Identification of variable amplitude fatigue loading based on bivariate probability mass functions

Document Type

Conference Proceeding

Date of Original Version



Fatigue life estimation under variable amplitude (VA) loading remains one of the major unresolved engineering problems. When engineering structures experience VA loading, the corresponding load at the crack front will be highly non-stationary. Under such conditions, the load interaction effects greatly affect fatigue crack propagation rate and can cause significant variation in the corresponding fatigue life. Existing characterization methods focus on single overload effect and their extensions to sequence of overloads which does not reflect the irregularity of the realistic loading. In this paper, numerical simulations were used to compare the fatigue life under various synthetic load time histories. The resulting time to failure estimation, concerning statistically and spectrally identical deterministic loads and their stochastic surrogates, exhibits drastic difference which reflects the corresponding differences in the temporal structure of the loads, which are not differentiated by their linear characteristics. This indicates that the temporal dynamics of the load time histories have considerable influence on fatigue crack propagation and fracture. In order to differentiate temporal dynamics of the synthetic loading, new bivariate probability mass functions (BPMFs) are proposed. They can capture and preserve the underlying temporal dynamics of locally non-stationary variable amplitude load-time histories with identical linear statistics. And qualitative explanations of the potential of this proposed method for fatigue life estimation under VA loading are discussed.

Publication Title

Proceedings of the ASME Design Engineering Technical Conference