Parametric probabilistic approach for cumulative fatigue damage using double linear damage rule considering limited data

João Paulo Dias (1) , Stephen Ekwaro-Osire (1) , Americo Cunha Jr (2) , Shweta Dabetwar (1) , Abraham Nispel (1) , Fisseha M Alemayehu (3) , Haileyesus B Endeshaw (4)

1 TTU - Texas Tech University [Lubbock]
2 UERJ - Universidade do Estado do Rio de Janeiro [Brasil] = Rio de Janeiro State University [Brazil] = Université d'État de Rio de Janeiro [Brésil]
3 West Texas A&M University
4 CSU - Colorado State University [Fort Collins]

This work proposes a parametric probabilistic approach to model damage accumulation using the double linear damage rule (DLDR) considering the existence of limited experimental fatigue data. A probabilistic version of DLDR is developed in which the joint distribution of the knee-point coordinates is obtained as a function of the joint distribution of the DLDR model input parameters. Considering information extracted from experiments containing a limited number of data points, an uncertainty quantification framework based on the Maximum Entropy Principle and Monte Carlo simulations is proposed to determine the distribution of fatigue life. The proposed approach is validated using fatigue life experiments available in the literature.

Mots clés: uncertainty quantification; cumulative fatigue damage; double linear damage rule; limited data experiments; Maximum Entropy Principle

Domaines: Mécanique des solides [physics.class-ph]; Mécanique des structures [physics.class-ph]; Génie mécanique [physics.class-ph]; Probabilités [math.PR]; Statistiques [math.ST]; Science non linéaire [physics]

João Paulo Dias, Stephen Ekwaro-Osire, Americo Cunha Jr, Shweta Dabetwar, Abraham Nispel, et al.. Parametric probabilistic approach for cumulative fatigue damage using double linear damage rule considering limited data. International Journal of Fatigue, 2019, 127, pp.246-258. ⟨hal-02158207⟩