Fatigue in Materials: Fundamentals, Multiscale Characterizations and Computational Modeling: Data-Driven Investigations of Fatigue
Sponsored by: TMS Structural Materials Division, TMS: Computational Materials Science and Engineering Committee, TMS: Integrated Computational Materials Engineering Committee, TMS: Mechanical Behavior of Materials Committee
Program Organizers: Garrett Pataky, Clemson University; Ashley Spear, University of Utah; Jean-Briac le Graverend, Texas A&M University; Antonios Kontsos, Drexel University; Brian Wisner, Ohio University
Tuesday 8:30 AM
February 25, 2020
Room: 11A
Location: San Diego Convention Ctr
Session Chair: Ashley Spear, University of Utah
8:30 AM Break
8:50 AM Invited
Statistical Modeling of Censored Life Data: D. Gary Harlow1; 1Lehigh University
Frequently fatigue life tests are censored prior to failure, particularly for high cycles. Censoring is utilized because the number of cycles required to reach failure exceed the allotted physical time or costs. The major issue is that censoring occurs for extrema where high reliability is required. New materials often have limited life data, especially for typical operating conditions. Accurate statistical modeling of censored life data are critical, albeit more challenging than modeling complete data. A methodology where variability can be managed for censored life data is illustrated. The variability is assumed to be cumulative from all sources. The procedure focuses on the fusion of experimental data over different applied loading conditions. The approach is contrasted with classical statistical characterization techniques for life analysis. The major conclusion is that using all available data, including censored data, improves estimation of life distribution functions, leading to more accurate prediction of reliability.
9:10 AM
Fatigue Crack Growth in Structural Cast Aluminum Alloys: Microstructural Mechanisms, Modeling Strategies, and Integrated Design: Anthony Spangenberger1; Diana Lados1; 1Worcester Polytechnic Institute
Advanced design for fatigue crack growth (FCG) resistance requires development and integration of phenomenological models across a range of size scales and mechanistic regimes. To address this need, numerical and computational models of small and long FCG have been developed for materials having a ductile matrix and brittle reinforcing phases, and are supported by experimental testing of cast A356 aluminum alloy. A combined physical/statistical predictive model of microstructurally small crack growth has been developed on the basis of transforming long FCG data using easily measured microstructural parameters for secondary phase-controlled behavior. Complementary to this, an extended finite element model of microstructurally-based small and long FCG was developed as part of a through-process approach to component development and material selection for enhanced structural integrity. Both models are validated by conventional FCG testing, and supported by novel characterization methods for properties at the microstructure scale.
9:30 AM
Marked 2-point Spatial Correlations of Microstructure Neighborhoods Surrounding Fatigue Hot-spots in Ti-6Al-4V: Adrienne Muth1; Surya Kalidindi1; Reji John2; Adam Pilchak2; David McDowell1; 1Georgia Institute of Technology; 2Air Force Research Laboratory
Formation of fatigue cracks at the subgrain scale is statistically rare, especially when plastic deformation is highly heterogeneous. Fatigue Indicator Parameters (FIPs) for Ti64 are computed using crystal plasticity simulations for ensembles of statistical volume elements. The maximum extreme value (EV) FIP locations are identified, and 2-point spatial correlations are applied at two different characteristic length scales to investigate the coupled effects of microstructure attributes in the neighborhood of these fatigue hot-spots. ‘Fine’ scale 2-point statistics includes the volume immediately surrounding a subgrain averaged EV FIP. ‘Coarse’ scale convolutes two masked volumes, one containing the volume elements within an EV FIP location, the second of a selected radius surrounding an EV FIP grain, ensuring either the head or tail of every 2-point vector occurs in an EV FIP grain, termed marked 2-point correlations. To reduce dimensionality of the associated 2-point correlations, principal component analysis is subsequently applied.
9:50 AM
Multi-scale Modeling and Uncertainty Quantification of Fatigue Crack Nucleation in Titanium Alloys with Parametrically Homogenized Constitutive Models: Deniz Ozturk1; Shravan Kotha1; Somnath Ghosh1; 1Johns Hopkins University
Parametrically-Homogenized Constitutive Models (PHCM) and Fatigue Crack Nucleation Models (PHCNMs) for Titanium alloys have been previously developed from computational homogenization of CPFE analysis results performed on 3D polycrystalline RVEs. The constitutive parameters of PHCMs have direct dependence on the morphological and crystallographic variables of the microstructure, and PHCMs are incorporated in commercial FEA codes as user material subroutines for microstructure-sensitive modeling of structural components. An uncertainty quantification framework is developed for PHCMs for multi-scale modeling of deformation and fatigue crack nucleation in structural components, accounting for material microstructural uncertainty, model reduction errors and calibration data sparsity. Microstructural uncertainty is represented by statistical moments of microstructural parameters calculated from multiple EBSD scans. A novel uncertainty propagation method is developed to propagate the uncertainties in PHCM constitutive parameters and microstructural variables to the model output variables of interest, such as stress, strain or macroscopic fatigue measures, while avoiding expensive Monte Carlo simulations.
10:10 AM Break
10:30 AM Invited
A Machine Learning Approach to Predict Fatigue Damage and Crack Initiation Sites in a BCC Steel Microstructure: Ali Riza Durmaz1; Thomas Straub1; Chris Eberl1; 1Fraunhofer IWM
A material fatigue lifetime is determined by the crack formation process: damage accumulation in individual grains, micro crack initiation, and finally short crack formation. In the past years, a testing methodology for fatigue damage evolution investigation was developed. This methodology uses sensitive measurements of the resonant frequency for correlation with damage initiation. Building on this work, a multi modal approach has been developed employing in-situ optical sample surface images. The processed image data allows the creation of labels for machine learning (ML) methods. As ML features, various microstructure characteristics are extracted using ex-situ EBSD measurements processed with MTEX and complemented with crystal plasticity FEM (CPFEM). Feature importance and relationships are extracted from a random forest model. Semantical segmentation with UNET of post-mortem SEM images allows the distinction between extrusion and crack regions.The first comparison of experimentally obtained, CPFEM and ML predicted damage initiation sites will be presented.
10:50 AM
Probabilistic Dwell Fatigue Life Prediction of Microtextured Ti-6Al-4V: Sushant Jha1; James Larsen2; Reji John2; Adam Pilchak2; 1University of Dayton Research Institute; 2US Air Force Research Laboratory
Near-alpha and alpha+beta titanium alloys exhibit significant debit in life under dwell fatigue loading compared to cyclic fatigue when containing microtextured regions (MTRs). The debit in life has been attributed to both early initiation as well as faster small-crack growth through suitably oriented MTRs, via a faceted crack growth mechanism. From component lifing perspective, the effect of MTR on minimum dwell fatigue life is traditionally accounted for by use of knock-down factors that are often based on empirical methods. In this presentation, a probabilistic model that captures the effect of spatial distribution of MTRs on the small-crack growth behavior and distribution in dwell fatigue life will be discussed. This modeling approach attempts to directly account for the effects of MTR variables and also the loading, stress gradient, and surface treatment variables on minimum life. The sensitivity of model predictions to these variables will also be discussed.
11:10 AM
Variable Amplitude Fatigue Analysis Through an Approach based in the Equivalent Number of Cycles: Hernan Pinto1; Paola Moraga1; Matias Valenzuela1; Alvaro Pena1; Jose Garcia1; 1Pontificia Universidad Catolica de Valparaiso
Nowadays, for constant strain or stress amplitude, several researches have proposed different fatigue analysis models based on different mechanical, mathematical and statistical considerations; being the constant strain or stress amplitude fatigue analysis well covered, providing a good modeling of fatigue life and fatigue life predictions. Regarding variable amplitude analysis, the Palmgren - Miner rule it is the most used and commonly accepted for analyze variable amplitude, nevertheless, several researchers continue studying and proposing new models and approaches to deal with fatigue under variable stress or strain amplitudes.In this research, an approach to analyze the fatigue behavior under variable strain amplitude is proposed. The model arises from a well known and established statistical fatigue analysis model (The Weibull Strain Fatigue model), considers the application of isoprobability curves and introduce the concept of equivalent number of cycles. Finally the model is validated using previous existing variable amplitude fatigue data.