A1 Journal article (refereed), original research

Evaluation of Strain–Life Fatigue Curve Estimation Methods and Their Application to a Direct-Quenched High-Strength Steel


Publication Details
Authors: Dabiri Mohammad, Ghafouri Mehran, Rohani Raftar Hamidreza, Björk Timo
Publisher: Springer Verlag (Germany)
Publication year: 2018
Language: English
Related Journal or Series Information: Journal of Materials Engineering and Performance
ISSN: 1059-9495
eISSN: 1544-1024
JUFO-Level of this publication: 1
Open Access: Not an Open Access publication

Abstract

Methods to estimate the strain–life curve, which were divided into three categories: simple approximations, artificial neural network-based approaches and continuum damage mechanics models, were examined, and their accuracy was assessed in strain–life evaluation of a direct-quenched high-strength steel. All the prediction methods claim to be able to perform low-cycle fatigue analysis using available or easily obtainable material properties, thus eliminating the need for costly and time-consuming fatigue tests. Simple approximations were able to estimate the strain–life curve with satisfactory accuracy using only monotonic properties. The tested neural network-based model, although yielding acceptable results for the material in question, was found to be overly sensitive to the data sets used for training and showed an inconsistency in estimation of the fatigue life and fatigue properties. The studied continuum damage-based model was able to produce a curve detecting early stages of crack initiation. This model requires more experimental data for calibration than approaches using simple approximations. As a result of the different theories underlying the analyzed methods, the different approaches have different strengths and weaknesses. However, it was found that the group of parametric equations categorized as simple approximations are the easiest for practical use, with their applicability having already been verified for a broad range of materials.


Last updated on 2019-13-03 at 12:00