Assessing reliability of fatigue indicator parameters for small crack growth via a probabilistic framework

Abstract : Microstructurally small cracks exhibit large variability in their fatigue crack growth rate. It is accepted that the inherent variability in microstructural features is related to the uncertainty in the growth rate. However, due to (i) the lack of cycle-by-cycle experimental data, (ii) the complexity of the short crack growth phenomenon, and (iii) the incomplete physics of constitutive relationships, only empirical damage metrics have been postulated to describe the short crack driving force metric (SCDFM) at the mesoscale level. The identification of the SCDFM of polycrystalline engineering alloys is a critical need, in order to achieve more reliable fatigue life prediction and improve material design. In this work, the first steps in the development of a general probabilistic framework are presented, which uses experimental result as an input, retrieves missing experimental data through crystal plasticity (CP) simulations, and extracts correlations utilizing machine learning and Bayesian networks (BNs). More precisely, experimental results representing cycle-by-cycle data of a short crack growing through a beta-metastable titanium alloy, VST-55531, have been acquired via phase and diffraction contrast tomography. These results serve as an input for FFT-based CP simulations, which provide the micromechanical fields influenced by the presence of the crack, complementing the information available from the experiment. In order to assess the correlation between postulated SCDFM and experimental observations, the data is mined and analyzed utilizing BNs. Results show the ability of the framework to autonomously capture relevant correlations and the equivalence in the prediction capability of different postulated SCDFMs for the high cycle fatigue regime.
Type de document :
Article dans une revue
Modelling and Simulation in Materials Science and Engineering, IOP Publishing, 2017, 25 (4), 045010, 29 p. 〈10.1088/1361-651X/aa6c45〉
Liste complète des métadonnées

Littérature citée [1 références]  Voir  Masquer  Télécharger

https://hal-mines-paristech.archives-ouvertes.fr/hal-01540936
Contributeur : Odile Adam <>
Soumis le : mardi 11 septembre 2018 - 15:22:19
Dernière modification le : lundi 12 novembre 2018 - 10:58:37

Fichier

Rovinelli_MSMSE-2017_offprint....
Fichiers produits par l'(les) auteur(s)

Identifiants

Citation

Andrea Rovinelli, Yoann Guilhem, Henry Proudhon, Ricardo A Lebensohn, Wolfgang Ludwig, et al.. Assessing reliability of fatigue indicator parameters for small crack growth via a probabilistic framework. Modelling and Simulation in Materials Science and Engineering, IOP Publishing, 2017, 25 (4), 045010, 29 p. 〈10.1088/1361-651X/aa6c45〉. 〈hal-01540936〉

Partager

Métriques

Consultations de la notice

283

Téléchargements de fichiers

5