Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Comparative study of multiaxial fatigue life prediction methodologies based on critical plane approaches: a review of Maximum Variance Method (MVM) and optimization algorithms
Submission Author:
Felipe Maganha de Lima , DF
Co-Authors:
Felipe Maganha de Lima, JORGE FERREIRA
Presenter: Felipe Maganha de Lima
doi://10.26678/ABCM.COBEM2023.COB2023-1064
Abstract
This study aims to evaluate methodologies for predicting multiaxial fatigue life based on random loading using critical plane approaches and optimized search algorithms for the identification of critical planes. Various multiaxial fatigue studies assert that most mechanical components and structures are often under multiaxial stress states, which, when caused by dynamic loads, can lead to the initiation and propagation of fatigue cracks. For instance, aircraft loads in flight, wind loads in wind turbines, wave action in offshore environments, liquid flow, and industrial pipe vibrations. Therefore, multiaxial fatigue life prediction has evolved with new strategies to ensure reliability and safety in mechanical designs. Additionally, the development of an efficient life prediction algorithm can optimize design time by providing preliminary numerical simulations of multiaxial fatigue failure before experimental and service operational tests. The literature review also addresses topics related to random multiaxial fatigue, such as stochastic processes and simulated signal reconstruction techniques based on standard deviation, spectral characteristics, and Power Spectral Density (PSD) reference functions of the original signal. Initially, the critical planes are mapped by the algorithm based on their variances using the Maximum Variance Method (MVM), in which normal and shear stresses act as driving forces, and gradient-based optimizers perform the selection of the final critical plane. Then, decoupled damage model approaches are applied for numerical life prediction based on the selected critical plane, material mechanical characteristics, and spectral characteristics of the signal. In the time domain, the Rainflow cycle counting method combined with the Palmgren-Miner damage accumulation rule, the Serensen-Kogayev correction, and the Modified Wöhler Curves Method (MWCM) to calibration of the model when the driving force is the shear stress. In the frequency domain, spectral analysis is performed using the Rayleigh distribution in the damage model, which can be combined with the Serensen-Kogayev correction. Finally, based on the results obtained, we can state that the methodologies used proved to be efficient in predicting multiaxial fatigue life with numerical estimates that were very close to the fatigue life of the material. For this reason, this work will contribute significantly to the application of reliable and low computational cost methodologies in multiaxial fatigue analysis, being an important step to guarantee the safety and reliability of mechanical structures subjected to cyclic loads.
Keywords
multiaxial fatigue, life prediction, Maximum Variance Method, driving forces, decoupled damage model
