0
TECHNICAL PAPERS

On the Fatigue Analysis of Non-Gaussian Stress Processes With Asymmetric Distribution

[+] Author and Article Information
Xiangyu Wang

Department of Mechanical Engineering, University of Delaware, Newark, DE 19716

J. Q. Sun1

Department of Mechanical Engineering, University of Delaware, Newark, DE 19716sun@me.udel.edu

1

Corresponding author.

J. Vib. Acoust 127(6), 556-565 (Jan 31, 2005) (10 pages) doi:10.1115/1.2110879 History: Received June 09, 2004; Revised January 31, 2005

This paper presents a multistage regression method to obtain an empirical joint probability density function (PDF) for rainflow amplitudes and means of non-Gaussian stress processes with asymmetric distribution. The proposed PDF model captures a wide range of non-Gaussian stress processes characterized by five parameters: Standard deviation, skewness, kurtosis, irregularity factor and mean frequency. It is shown that the fatigue prediction from the closed-form empirical PDF based on the multistage regression analysis agrees well with that from extensive Monte Carlo simulations. The ultimate purpose of the method is to provide a tool for engineers to predict the remaining fatigue life of a structure with direct measurements of stress responses at critical locations on a structure or stress responses at various hot-spots on a structure from the finite element analysis without conducting further Monte Carlo simulations.

FIGURES IN THIS ARTICLE
<>
Copyright © 2005 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Typical PSD of Gaussian and non-Gaussian stress processes. Solid line: PSD of non-Gaussian processes with kurtosis α4=14 and α3=1.067. Dashed line: PSD of a multiresonance Gaussian processes with coefficients Φ1=9.9145×102, ς1=0.01, ω1=60, Φ2=1.1101×105, ς2=0.04, ω2=250.

Grahic Jump Location
Figure 2

PDFs of the rainflow amplitude. Dashed line: Estimates from the adaptive kernel method. Solid line: The predictions from Eq. 31. (a) α3=1.515, α4=17, r=0.521, m=0.393. (b) α3=2.041, α4=14, r=0.711, m=0.605. (c) α3=−1.515, α4=10, r=0.799, m=0.698. (d) α3=−2.041, α4=20, r=0.320, m=0.162.

Grahic Jump Location
Figure 3

PDFs of the rainflow mean. Dashed line: Estimates from the adaptive kernel method. Solid line: The predictions from Eq. 18. (a) α3=0.509, α4=17, r=0.316, m=0.159. (b) α3=2.041, α4=10, r=0.512, m=0.381. (c) α3=−1.067, α4=14, r=0.414, m=0.279. (d) α3=−2.041, α4=24, r=0.713, m=0.608.

Grahic Jump Location
Figure 4

Symmetry of PDFs of rainflow mean with a same value of skewness but opposite sign. (a) α4=24, r=0.436, m=0.298. Solid line: α3=0.509. Dashed line: α3=−0.509. (b) α4=14, r=0.798, m=0.697. Solid line: α3=1.067. Dashed line: α3=−1.067. (c) α4=10, r=0.601, m=0.480. Solid line: α3=1.515. Dashed line: α3=−1.515. (d) α4=20, r=0.320, m=0.162. Solid line: α3=2.041. Dashed line: α3=−2.041.

Grahic Jump Location
Figure 5

Examples of δ2 as a function of xδ2=(r−m)∕(r+m) for different stress processes. Symbols: The regression analysis of the rainflow mean distribution. Lines: The prediction from Eq. 23. (a) α4=10. ∗ and solid line: α3=0. 엯 and dashed line: α3=0.509. ◻ and dashdot line: α3=1.515. (b) α4=14. ∗ and solid line: α3=0. 엯 and dashed line: α3=1.067. ◻ and dashdot line: α3=2.041. (c) α4=17. ∗ and solid line: α3=0.509. 엯 and dashed line: α3=1.515. ◻ and dashdot line: α3=2.041. (d) α4=24. ∗ and solid line: α3=0. 엯 and dashed line: α3=1.067. ◻ and dashdot line: α3=1.515.

Grahic Jump Location
Figure 6

The regression results of ϵi(i=1,2) with respect to α3 in the third stage regression. Symbols: Data points from the second stage regression. Lines: The prediction from Eq. 21. ∗ and solid line: α4=10. 엯 and dashed line: α4=14. ◻ and dotted line: α4=17. ◇ and dashdot line: α4=20. ▿ and –∙– line: α4=24.

Grahic Jump Location
Figure 7

The regression results of μi1 and μi2(i=1,2) with respect to α4 in the forth stage regression. ∗: The third stage regression. Solid line: The prediction from Eq. 22.

Grahic Jump Location
Figure 8

Examples of the regression results of the parameters δ1, δ3, δ4, and δ5. Symbols: The regression analysis of the rainflow mean distribution. Lines: The prediction from the regression models. (a) α4=17. ∗ and solid line: α3=0. 엯 and dashed line: α3=1.067. ◻ and dashdot line: α3=2.041. (b) α4=24. ∗ and solid line: α3=1.067. 엯 and dashed line: α3=1.515. ◻ and dashdot line: α3=2.041. xδ5=r+m. (c) α4=14. ∗ and solid line: α3=1.067. 엯 and dashed line: α3=1.515. ◻ and dashdot line: α3=2.041. xδ4=(2r2−m)∕(0.2r+1). (d) α4=20. ∗ and solid line: α3=0.509. 엯 and dashed line: α3=1.067. ◻ and dashdot line: α3=1.515.

Grahic Jump Location
Figure 9

PDF of the rainflow amplitude of stress processes with kurtosis α4=10, irregularity factor r=0.799, mean frequency m=0.699, and different skewness values. Solid line: Prediction from Eq. 31. ▿: α3=0. ●: α3=0.509. 엯: α3=1.067. ∗: α3=1.515. ◻: α3=2.041. ◇: α3=−0.509. ×: α3=−1.067. +: α3=−1.515. ▵: α3=−2.041.

Grahic Jump Location
Figure 10

Validation of the regression model for the PDF of rainflow amplitude. Solid line: The prediction from Eq. 31. Dashed line: The estimate from the adaptive kernel method. (a) α3=−2.041, α4=12, r=0.798, m=0.696. (b) α3=0.509, α4=22, r=0.325, m=0.166.

Grahic Jump Location
Figure 11

Validation of the regression model for the PDF of rainflow mean. Solid line: The prediction from Eq. 18. Dashed line: The estimate from the adaptive kernel method. (a) α3=2.041, α4=22, r=0.794, m=0.693. (b) α3=−1.067, α4=16, r=0.520, m=0.392.

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In