An efficient framework for rapid life assessment in industrial applications: Fatigue crack growth. (August 2017)

• Record Type: Journal Article
• Authors: Ananthasayanam, Balajee; Loghin, Adrian; Subramaniyan, Arun K.
• Is Part Of: Engineering fracture mechanics. Volume 181(2017)
• Page Start: 7
• Page End: 28
• Publication Date: 2017-08
• Subjects: Crack propagation -- Fatigue -- Lifing -- 3DFAS -- Bayesian hybrid models -- Optimal Latin hypercube; Fracture mechanics -- Periodicals; Rupture, Mécanique de la -- Périodiques; Fracture mechanics; Periodicals; 620.112605
• DOI: 10.1016/j.engfracmech.2017.06.016
• Languages: English
• ISSNs: 0013-7944

Abstract

Highlights: Practical tool for engineers to quickly assess crack propagation life. Combines accuracy of FE and computational efficiency of handbook based solutions. Computationally efficient for probabilistic fracture mechanics. Suitable for Arbitrary Far-field 2D stress gradients and crack shape transitions. Abstract: In the last 30 years closed form solutions for stress intensity factors (Sih, 1973)[1] have been used extensively in crack propagation life assessment. While these low runtime solutions are very attractive for probabilistic type life assessments, they are constrained to simple geometries and loading conditions that would limit their application at component level. More recently, traditional 3D finite element techniques were developed to model incremental crack growth and estimate life. In contrast to the closed form solutions, finite element simulations can be more accurate capturing geometry and loading conditions but at a high runtime cost. An alternate method to make use of accurate 3D finite element based stress intensity factor solutions in a more efficient manner and to determine crack propagation life is proposed here. Bayesian hybrid modeling is utilized in this approach to provide a balance between the two techniques. Several Mode-I examples are presented herein to set a benchmark on accuracy and runtime efficiency of the proposed method.