Anda di halaman 1dari 136

A Thesis

entitled
Correlations among Tensile and Cyclic Deformation Properties for Steels and
Implications on Fatigue Life Predictions
by
Zachary Lopez
Submitted to the Graduate Faculty as partial fulfillment of the requirements for the
Master of Science Degree in Mechanical Engineering

_________________________________________
Dr. Ali Fatemi, Committee Chair


_________________________________________
Dr. Phillip White, Committee Member


_________________________________________
Dr. Lesley Berhan, Committee Member






_________________________________________
Dr. Patricia R. Komuniecki, Dean
College of Graduate Studies




The University of Toledo
May 2012









































Copyright 2012, Zachary Lopez

This document is copyrighted material. Under copyright law, no parts of this document
may be reproduced without the expressed permission of the author.
iii
An Abstract of

Correlations among Tensile and Cyclic Deformation Properties for Steels and
Implications on Fatigue Life Predictions

by

Zachary Lopez

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the
Master of Science Degree in Mechanical Engineering

The University of Toledo

May 2012


Cyclic loading can significantly alter the monotonic deformation resistance of a
material. Detailed knowledge of the cyclic deformation behavior of a material is essential
for reliable life calculations of cyclically loaded structures. Experimental determination
of cyclic deformation properties requires significantly more time and cost, as compared
to determination of tensile properties from a simple tension test. The objective of this
study was to derive a method for estimating the cyclic deformation properties of steels
from common tensile properties and hardness. A variety of correlations among hardness,
tensile properties, and cyclic deformation properties were found for fifty-seven steels
from the AISI Bar Steel Fatigue Database included in this study. Several correlations
from the literature among the aforementioned properties were also examined and
compared to the correlations proposed in this study using a different set of data from
sixty-six steels from SAE J1099 Technical Report. This data set was chosen to provide an
unbiased basis of comparison among the methods and to verify the predictive accuracy of
the proposed correlations. The proposed correlations successfully predicted the cyclic
iv
deformation properties of steels from both data sets with higher accuracy, compared to
most of the correlations from the literature. Most importantly, the proposed methods
require only hardness and three monotonic tension properties: ultimate tensile strength
(S
u
), yield strength (S
y
), and modulus of elasticity (E); all of which are either readily
available, or can easily be obtained. Fatigue life predictions based on two proposed
methods were compared using experimental data from the AISI database and from the
SAE J1099 report. It is shown that the proposed methods predict the fatigue life of most
of the steels reasonably well.















v


Acknowledgements

First I would like to extend my sincere appreciation to Dr. Ali Fatemi, my thesis and
research advisor, for his support, encouragement, and guidance on this thesis study. I
would like to thank Dr. Phillip White and Dr. Lesley Berhan for serving on my defense
committee. I would also like to thank Dr. Tom Oakwood from the American Iron and
Steel Institute (AISI) Fatigue Task Group for his assistance regarding the AISI bar steel
fatigue database.

vi


Table of Contents

Abstract ............................................................................................................................. iii
Acknowledgements ............................................................................................................v
Table of Contents ............................................................................................................. vi
List of Tables .................................................................................................................. ix
List of Figures .....................................................................................................................x
List of Abbreviations ................................................................................................... xviii
List of Symbols ............................................................................................................... xix
1 Introduction ..........................................................................................................1
1.1 Background ..........................................................................................................1
1.1.1 Importance of Fatigue and Cyclic Deformation .........................................1
1.1.2 Dependence of Cyclic Deformation Behavior on Microstructure ..............2
1.1.3 Significance of Cyclic Deformation Property Correlations with
Tensile Properties and Their Use in Design ...............................................3
1.2 Some Previous Studies on Cyclic Deformation Behavior Predictions .................5
1.3 Purpose of This Study ...........................................................................................9
2 Experimental Data .....................................................................................................16
2.1 Materials ........................................................................................................16
2.2 Determination of Monotonic and Cyclic Deformation Properties ......................17
2.2.1 Monotonic Tension Properties ..................................................................17
vii
2.2.2 Cyclic Deformation Properties .................................................................19
2.3 Data Exclusion Method Used .............................................................................20
3 Correlations among Tensile and Cyclic Deformation Properties..........................33
3.1 Correlations among Monotonic Properties and Hardness ..................................33
3.2 Classification of Materials Based on the S
u
/S
y
Ratio..........................................35
3.3 Comparison of Deformation Behavior between Steels of the Same
Hardness .............................................................................................................36
3.4 Correlations among Monotonic Properties or Hardness and the Cyclic
Strength Coefficient ...........................................................................................38
3.5 Correlations among Monotonic Properties or Hardness and the Cyclic
Strain Hardening Exponent ................................................................................41
3.6 Correlations among Monotonic Properties or Hardness and the Cyclic
Yield Strength .....................................................................................................42
3.7 Comparisons between Selected Methods............................................................44
3.7.1 Comparison of Cyclic Deformation Behavior Predictions
(Hardening of Softening) ..........................................................................44
3.7.2 Comparison of Cyclic Strength Coefficient Predictions ...........................46
3.7.3 Comparison of Cyclic Strain Hardening Exponent Predictions ...............47
3.7.4 Comparison of Cyclic Yield Strength Predictions ....................................48
3.8 Approximations of the Cyclic Stress-Strain Curve .............................................49
4 Implication of the Proposed Cyclic Deformation Prediction Method on
Fatigue Life Estimations............................................................................................90
4.1 Fatigue Life Prediction from Proposed Cyclic Stress-Strain Estimation
viii
Method and the Roessle-Fatemi Strain-Life Method .........................................90
4.2 Comparisons of Fatigue Life Estimations Based on the Experimental and
Predicted Cyclic Deformation Curves and the Experimental Monotonic
Deformation Curve .............................................................................................92
5 Summary and Conclusions .....................................................................................111
References .......................................................................................................................116

ix


List of Tables

2.1 Monotonic and cyclic deformation properties of selected steels from the
AISI bar steel fatigue database [10] .......................................................................25
2.2 Monotonic and cyclic deformation properties of selected steels from SAE
J1099 [11] ..............................................................................................................27
3.1 Cyclic deformation behavior predictions from various tensile properties
for selected steels from the AISI bar steel fatigue database [10]. ..........................53
3.2 Summary of the correctly predicted cyclic deformation behavior of
selected steels from the AISI bar steel fatigue database [10] by various
tensile properties. ...................................................................................................55



x


List of Figures

1-1 A plot of superimposed monotonic and cyclic stress-strain curves used to
illustrate both cyclic hardening and softening .......................................................11
1-2 An overview of the fatigue design process and the role of cyclic stress-
strain relationship ...................................................................................................12
1-3 Plot of S
u
/S
y
versus stress ratio from [4] ................................................................13
1-4 Estimation of the cyclic strength coefficient from (a) the monotonic
strength coefficient in [7], and (b) Brinell hardness in [8] ....................................14
1-5 Estimation of the cyclic strain hardening exponent from Brinell hardness
in [8] .......................................................................................................................15
2-1 Example of (a) mid-life hysteresis loop, and (b) true stress amplitude
versus true plastic strain amplitude plot used to obtain K' and n' ..........................29
2-2 Example of a steel material excluded from the study due to (a) significant
scatter of true stress versus true plastic strain amplitude data, and (b)
maximum plastic strain amplitude less than 0.10% ...............................................30
2-3 AISI steel data in a plot of (a) experimental versus calculated cyclic strain
hardening exponent from compatibility Equation 2.10, and (b)
experimental versus calculated cyclic strength coefficient from
compatibility Equation 2.9 .....................................................................................31
xi
2-4 SAE J1099 steel data in a plot of (a) experimental versus calculated cyclic
strain hardening exponent from compatibility Equation 2.11, and (b)
experimental versus calculated cyclic strength coefficient from
compatibility Equation 2.10 ...................................................................................32
3-1 Superimposed AISI and SAE steel data in a plot of (a) ultimate tensile
strength, (b) yield strength, and (c) monotonic strength coefficient versus
Brinell hardness. ....................................................................................................57
3-2 AISI and SAE steel data from various material groups superimposed in a
plot of ultimate tensile strength versus yield strength ...........................................57
3-3 AISI and SAE steel data of various material groups superimposed in a plot
of (a) ultimate tensile strength, (b) yield strength, and (c) monotonic
strength coefficient versus Brinell hardness ..........................................................59
3-4 Superimposed AISI and SAE steel data of various material groups shown
in a plot of S
u
/S
y
versus Brinell hardness...............................................................59
3-5 A comparison of the cyclic deformation behavior for two steels with the
same hardness superimposed in a plot of (a) true stress versus true plastic
strain amplitude, and (b) true stress amplitude versus true strain amplitude .........60
3-6 Superimposed AISI steel data of various material groups shown in a plot
of cyclic strength coefficient versus (a) Brinell hardness, (b) ultimate
tensile strength, (c) monotonic yield strength, and (d) monotonic strength
coefficient ..............................................................................................................62
3-7 Superimposed AISI data of various material groups in a plot of cyclic
strength coefficient versus S
u
/S
y
............................................................................63
xii
3-8 AISI data of various material groups superimposed in a plot of cyclic
strain hardening exponent versus (a) Brinell hardness, (b) ultimate tensile
strength, (c) monotonic yield strength, and (d) monotonic strain hardening
exponent .................................................................................................................65
3-9 AISI steel data of various material groups superimposed in a plot of cyclic
strain hardening exponent versus (a) S
y
/S
u
, and (b) S
u
/S
y
.....................................66
3-10 AISI steel data of various material groups superimposed in a plot of cyclic
yield strength versus (a) Brinell hardness, (b) ultimate tensile strength, and
(c) monotonic yield strength ..................................................................................68
3-11 AISI and SAE steel data superimposed in a plot of predicted versus
experimental cyclic strength coefficient for (a) Equation 1.8 based on K
[7], (b) Equation 1.9 based on HB [8], (c) Equations 3.9 and 3.10 based on
S
u
, and (d) Equations 3.7 and 3.8 based on HB .....................................................70
3-12 AISI and SAE steel data superimposed in a plot of predicted versus
experimental cyclic strain hardening exponent for (a) Equations 1.5
through 1.7 based on n, S
u
, S
y
, and
f
[7], (b) Equation 1.10 from [8], (c)
Equation 3.18 based on S
y
/S
u
, (d) Equations 3.25 and 3.26 based on S
y

and HB, and (e) Equations 3.27 and 3.28 based on S
y
and S
u
...............................73
3-13 AISI and SAE steel data superimposed in a plot of predicted versus
experimental cyclic yield strength for (a) Equation 3.20 based on HB, (b)
Equation 3.21 based on S
u
, (c) Equations 3.22 and 3.23 based on S
y
, (d)
Equation 2.9 with K' as a function of HB, (e) Equation 2.9 with K' as a
function of S
u
, and (f) Equation 1.11 based on RA and S
u
from [9] .....................76
xiii
3-14 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1020, (b) SAE 1038, (c) SAE 1038, (d) SAE 1541, (e)
SAE 1141Nb, (f) SAE 1050M, (g) SAE 10V45, and (h) SAE 10V45. .................77
3-15 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1151V, (b) SAE 1045, (c) SAE 1141V, (d) SAE
1141V, (e) SAE 1141AL, (f) SAE 5150, (g) SAE 15V41, and (h) SAE
1090M. ...................................................................................................................78
3-16 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1090, (b) SAE 1090, (c) SAE 1090, (d) SAE 4620, (e)
SAE 8620, and (f) SAE 8620. ................................................................................79
3-17 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 9310, (b) SAE 1141AL, (c) SAE 41B17M (PS19), (d)
SAE 15B35, (e) SAE 4620, (f) SAE 8822, (g) SAE 8822, and (h) SAE
5140........................................................................................................................80
3-18 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 4320, (b) SAE 10B21, (c) SAE 4140, (d) SAE 4140,
(e) SAE 4140, (f) SAE 4140, (g) SAE 4140, and (h) SAE 4140 ...........................81
xiv
3-19 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 86B20, (b) SAE 4130AL, (c) SAE 51B60, (d) SAE
9254, and (e) SAE 9254AL ...................................................................................82
3-20 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1022, (b) SAE C-70, (c) SAE 1141Nb, (d) SAE
15V24, (e) SAE 1038, (f) SAE 1141V, (g) SAE 5120, and (h) SAE 9310 ...........83
3-21 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1090M, (b) SAE 8620, (c) SAE 20MnCr5, (d) SAE
20MnCr5, (e) SAE 20MnCr5, and (f) SAE 86B20 ...............................................84
3-22 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE DDQ, (b) SAE DDQ+, (c) SAE DDQ+, (d) SAE
DQSK, (e) SAE 1005, (f) SAE 1005, (g) SAE 1008, and (h) SAE 10V40 ...........85
3-23 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE CQ, (b) SAE HF 50, (c) SAE HF 50, (d) SAE HF 50,
(e) SAE HF 60, (f) SAE HF 60, (g) SAE HF 50, and (h) SAE HF 60 ..................86
3-24 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
xv
including, (a) SAE HF 60, (b) SAE HF 60, (c) SAE HF 80, (d) SAE HF
80, (e) SAE HF 80, (f) SAE HF 80, (g) SAE HF 80, and (h) SAE 51V45............87
3-25 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1005, (b) SAE 1004, (c) SAE HF 50, (d) SAE HF 50,
(e) SAE HF 50, (f) SAE HF 50, (g) SAE HF 60, and (h) SAE HF 80 ..................88
3-26 Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE HF 80 and (b) SAE 1040 ........................................................89
4-1 Overview of the fatigue life calculation procedure used .......................................97
4-2 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
135 HB to 215 HB .................................................................................................98
4-3 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
215 HB to 270 HB .................................................................................................99
4-4 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
xvi
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
270 HB to 350 HB ...............................................................................................100
4-5 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
185 HB to 305 HB ...............................................................................................101
4-6 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
320 HB to 370 HB ...............................................................................................102
4-7 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
390 HB to 585 HB ...............................................................................................103
4-8 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with 1.2 < S
u
/S
y
< 1.4 and hardness range
of 195 HB to 270 HB ...........................................................................................104
xvii
4-9 Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with 1.2 < S
u
/S
y
< 1.4 and hardness range
of 270 HB to 400 HB ...........................................................................................105
4-10 Fatigue life estimations for steels in the AISI database based on the
predicted cyclic deformation curve (method 1). ..................................................106
4-11 Fatigue life estimations for steels in the AISI database based on the
predicted cyclic deformation curve (method 2). ..................................................107
4-12 Fatigue life estimations for steels in the AISI database based on the
experimental monotonic deformation curve ........................................................108
4-13 Fatigue life estimations for steels in SAE J1099 based on the predicted
cyclic deformation curve (method 1). ..................................................................109
4-14 Fatigue life estimations for steels in SAE J1099 based on the predicted
cyclic deformation curve (method 2) ...................................................................110

xviii


List of Abbreviations


AISI ............................American Iron and Steel Institute
ASTM ........................American Society for Testing and Materials

FEA ............................Finite Element Analysis

H .................................Cyclic Hardening Behavior
HB ..............................Brinell Hardness

M ................................Mixed-Mode Behavior

S .................................Cyclic Softening Behavior
SAE ............................Society of Automotive Engineers
S-N .............................Stress-Life
St ................................Stable Behavior

-N ..............................Strain-Life

xix


List of Symbols


A
o
...............................Original cross section area
A
f
................................Cross section area at fracture
b..................................Fatigue strength coefficient
c ..................................Fatigue ductility exponent
e ..................................Engineering strain
E .................................Modulus of elasticity
EL ...............................Percent elongation
K .................................Monotonic strength coefficient
K' ................................Cyclic strength coefficient
n..................................Monotonic strain hardening exponent
n' .................................Cyclic strain hardening exponent
2N
f
..............................Number of reversals to failure
RA ..............................Reduction in area
S .................................Engineering stress
S
y
................................Monotonic yield strength
S
y
' ...............................Cyclic yield strength
S
u
................................Ultimate tensile strength

A ...............................True strain range
A
e
..............................Elastic strain range
A
p
..............................Plastic strain range
A ...............................True stress range
.................................True fracture ductility parameter
..................................True strain

a
.................................Strain amplitude

e
.................................True elastic strain

f
.................................True fracture ductility

f
' ................................Fatigue ductility coefficient

p
.................................True plastic strain
.................................True stress

a
................................True stress amplitude

f
................................True fracture strength


1
Chapter 1
Introduction

1.1 Background
1.1.1 Importance of Fatigue and Cyclic Deformation
Many books and articles have suggested that 50 to 90 percent of mechanical
failures are fatigue failures [1]. Most of these failures occur suddenly with little or no
warning, and in some cases result in catastrophic loss. Consequently, considerable effort
has been made to determine the physical nature of fatigue.
According to the American Society for Testing and Materials (ASTM), fatigue is
defined as the progressive, localized, and permanent structural change that occurs in a
material subjected to repeated or fluctuating strains at nominal stresses with maximum
values less than the tensile strength of the material [1]. Without repeated plastic
deformation fatigue will not occur. Although the nominal behavior is typically elastic,
cyclic plastic deformation commonly occurs at stress concentrations or at the microscopic
level.
In 1886, Bauschinger showed that a single loading reversal resulting in inelastic
deformation reduced the yield strength in the opposite loading direction [2]. It is now
well understood that the fatigue process can alter the stress-strain behavior of a material.
2
The magnitude of cyclically induced changes can be seen when comparing the materials
monotonic and cyclic stress-strain curves. If a material cyclically hardens, the cyclic
stress-strain curve will lie above the monotonic stress-strain curve. If a material cyclically
softens, the cyclic stress-strain curve will lie below the monotonic stress-strain curve, as
shown in Figure 1-1. It may be seen in this figure that use of the monotonic stress-strain
curve for a cyclic stress-amplitude of 1000 MPa would result in predicting fully elastic
strains, when in fact considerable cyclic plastic strain is present. This is made apparent by
the significant difference that exists between the strains estimated from the monotonic
and cyclic deformation curves (0.5% for monotonic vs. 1.35% for cyclic). Clearly, due to
the cyclic loading conditions that most components endure, the cyclic deformation
properties associated with individual materials are of utmost importance to design
engineers.
1.1.2 Dependence of Cyclic Deformation Behavior on Microstructure
Cyclic induced plastic deformation is generally produced by the motion of
dislocations. The degree to which a material cyclically hardens or softens is therefore
determined by the motion of dislocations, their mutual interactions, and interactions with
other types of lattice defects. Dislocation motion can be influenced by precipitates,
foreign particles, and grain boundaries [3].
For materials that are initially soft, their dislocation density increases in the
presence of cyclic plastic deformation. This increase in dislocation density results in a
decrease in dislocation mobility. Therefore, the material is said to cyclically harden as a
result of this increase of plastic deformation resistance [1].
3
Conversely, for initially hard materials, dislocation mobility increases under the
presence of inelastic strain cycling. In this case, cyclic plastic deformation causes the
initial dislocation structure to rearrange into a configuration such that there is less
resistance to dislocation mobility. Therefore, the material is said to cyclically soften [1].
To illustrate the difference of cyclic deformation behavior between hard and soft
steels, two different steels have been selected. The materials are 1038 and 4140 steels
with Brinell hardness of 185 and 409, respectively. In Figure 1-1, the monotonic and
cyclic stress-strain curves are superimposed for both steels. It can be seen from this figure
that both the monotonic and cyclic stress-strain curves of 4140 steel are above those of
1038 steel. This is to be expected considering 4140 steel is harder than 1038 steel. Also,
4140 steel cyclically softens whereas 1038 steel cyclically hardens. In Section 3.7.1, the
relationship between cyclic deformation behavior (cyclic hardening or softening) and
Brinell hardness is examined with the AISI steel database.
1.1.3 Significance of Cyclic Deformation Property Correlations with
Tensile Properties and Their Use in Design
Material selection for cyclically loaded structures is one of the most important
design decisions. In order to properly simulate the stress-strain or load-deflection
response of cyclically loaded structures, the cyclic deformation properties of the material
must be known. Determination of accurate cyclic deformation properties typically
requires significantly more time and cost, as compared to monotonic tension or hardness
tests which are relatively simple and inexpensive. Also, the properties obtained from
tension tests are more commonly available in the literature, as compared to the cyclic
4
deformation properties. Therefore, reliable correlations between tensile and cyclic
deformation properties with reasonable accuracy are very valuable to many industries.
An overview of the fatigue design process is illustrated in Figure 1-2. As shown
in this figure, there are several stress-strain relationships listed including Hookes Law,
the monotonic stress-strain relationship based on the monotonic strength coefficient (K)
and strain hardening exponent (n), and the cyclic stress-strain relationship based on the
cyclic strength coefficient (K') and strain hardening exponent (n'). It was shown in
Section 1.1.1 that the stress-strain response of a cyclically loaded material can be
significantly different from its monotonic stress-strain response. Therefore, the cyclic
stress-strain relationship should be used, especially when plastic strain is present. The
cyclic stress-strain relationship can be determined from experimental values of K' and n'
or predicted ones from tensile properties. A cyclic stress-strain equation along with the
external loading condition and geometry can then be used in conjunction with Finite
Element Analysis (FEA) to determine the stress and strain distributions within a
component. The stress-life approach (S-N) is commonly used to predict the fatigue life of
high strength or hardened steels since the strains are primarily elastic. The strain-life
approach (-N) can better characterize the fatigue behavior of a material subjected to
plastic deformation compared to the stress-life approach. Strains at notches or sharp
geometrical contours can be examined along with their effect on fatigue life by using this
approach. In the absence of experimental strain-life fatigue properties, several methods
currently available in the literature can be used to estimate the strain-life curve and
ultimately predict the fatigue life of the component.
5
In this study, new correlations among hardness, monotonic tensile data, and cyclic
deformation properties are proposed. The validity of existing correlations currently
available in the literature among hardness, monotonic tensile properties, and cyclic
deformation properties will be examined using the steels included in this study. The
results from these correlations will then be compared to those developed in this study.
Calculated fatigue lives based on the predicted cyclic stress-strain relationship will also
be compared to those based on the experimental cyclic stress-strain relationship.
1.2 Some Previous Studies on Cyclic Deformation Behavior
Predictions
Several researchers have developed correlation models to predict the cyclic
deformation behavior of steels from monotonic tensile properties and hardness. Manson
et al. [4] discovered a correlation between the hardening or softening behavior and the
ratio of ultimate tensile strength to yield strength (S
u
/S
y
) from sixteen materials including
steel, aluminum, and titanium alloys. The steel materials tested were AISI 4130 (soft and
hard), AISI 4340 (annealed and hard), AISI 52100, AISI 304 ELC (annealed and hard),
AISI 310 (annealed), and AM 350 (annealed and hard). Tensile and strain-controlled
fatigue tests were performed for each alloy to determine the monotonic and cyclic stress-
strain curves. Based on their results, all materials for which S
u
/S
y
1.4 hardened under
cyclic straining and those for which S
u
/S
y
1.2 cyclically softened. Cyclic hardening and
softening occurred at intermediate values for which 1.2 < S
u
/S
y
< 1.4. This correlation is
shown in Figure 1-3 [4].
Landgraf et al. [5] found a similar correlation between the cyclic hardening or
softening behavior and the strain hardening exponent (n) for several metallic materials
6
including three conditions of copper, two aluminum alloys, a titanium alloy, and several
steel alloys covering a range of hardness. The steel materials tested were Man-Ten steel,
SAE 4340 steel, SAE 1045 steel, and SAE 4142 steel. Based on their results, when the
strain hardening exponent is greater than 0.2, cyclically hardening can be expected. When
the strain hardening exponent is less than 0.1, cyclic softening can be expected. At
intermediate values of n between 0.1 and 0.2, cyclic stability or mixed-mode behavior
can be expected.
In a study conducted by Zhang et al. [6], a fracture ductility parameter () based
on the true fracture ductility coefficient (
f
) and reduction in area was introduced to
predict cyclic hardening or softening behavior:
f
f
o
o
f o
RA
A
A
A
A A
c o ) ( ln =

= (1.1)
where A
o
is the initial cross-section area and A
f
is the fracture cross-section area. Based
on the 40 alloys (aluminum, titanium, and steel) used in their study [6], they found that
when the fracture ductility parameter was less than 2% or between 20% and 65.4%, the
material cyclically softens. When the fracture ductility parameter was greater than 65.4%
or between 2% and 20%, the material cyclically hardens.
Several approaches have been proposed to predict cyclic deformation behavior by
estimating the cyclic strength coefficient (K') and cyclic strain hardening exponent (n')
from monotonic tensile properties and hardness. In a study conducted by Zhang et al. [7],
a method for estimating the cyclic strength coefficient (K') and cyclic strain hardening
exponent (n') from monotonic tensile properties was proposed. To estimate the cyclic
strain hardening exponent, three characteristics based on the monotonic strain hardening
exponent (n), true fracture ductility parameter (), true fracture strength (
f
), yield
7
strength (S
y
), and ultimate tensile strength (S
u
) were determined empirically. These
parameters were shown to provide the following three characteristics:
1) n' > n if < 20% and if 6 . 1 <
y
f
S
o
(1.2)
2) n' < n if < 20% and if 6 . 1 >
y
f
S
o
(1.3)
3)
n
n
S
S
S
y
u
y
f
'
~
o
if > 20% (1.4)
Based on these three characteristics, the following relationships for estimating the cyclic
strain hardening exponent (n') were derived:
|
|
.
|

\
|
+ = '
y
u
S
S
n n 1 1 06 . 1 | for < 5% or for 10% < 20% (1.5)
|
|
.
|

\
|
+ = '
u
f
S
n n
o
| 1 1 06 . 1 for 5% < < 10% and (1.6)
n
S
S
n
u f
y
|
|
.
|

\
|

= '
o
for > 20% (1.7)
In Equations 1.5 and 1.6, = 1 for
f
/S
y
< 1.6 and = -1 for
f
/S
y
> 1.6. In a similar
manner, an equation relating the cyclic strength coefficient (K') in MPa to the monotonic
strength coefficient (K) in MPa was derived in [7] as:
1220 57
545 . 0
= ' K K (1.8)
Equation 1.8 was obtained by a least-squares fit through data of seventeen alloys
(aluminum, steel, and titanium), and is shown in Figure 1-4(a). Equations 1.5 through 1.8
were shown to provide reasonable estimations of K' and n' of twenty-two aluminum,
steel, and titanium alloys in [7]. According to their results, the cyclic strength coefficient
8
and cyclic strain hardening exponent estimated by this method deviated from the
experimental values by as much as 27% and 34%, respectively [7].
Basan et al. [8] investigated correlations between experimentally determined
cyclic deformation properties and hardness of quenched and tempered low-alloy steel
42CrMo4. Experimental data from forty 42CrMo4 steels gathered from the literature and
their own test results were used to obtain the correlations. The Brinell hardness of these
steels ranged from 186 to 670. A good correlation was found to exist between the cyclic
strength coefficient (K') in MPa and Brinell hardness (HB) as:
75 . 376 ) ( 1173 . 0 ) ( 009 . 0
2
+ + = ' HB HB K (1.9)
Equation 1.9 was obtained by a polynomial least-squares fit for which R
2
= 0.703. This
correlation is shown in Figure 1-4(b). A plot of the cyclic strain hardening exponent vs.
Brinell hardness is provided in Figure 1-5. It may be seen from this figure that no
correlation could be found between the cyclic strain hardening exponent and hardness, so
the median value was used to represent this parameter [8]:
1087 . 0 = ' n (1.10)
The predictive accuracy of this method was evaluated by determining the percentage of
estimated stress amplitudes falling within various scatter-bands of specified factors.
These factors represent the percentage of deviation between stress-amplitudes calculated
from experimental values of K' and n' and predicted values of K' and n'. According to
their results, 59% of the data falls within a factor of 10%, 25% of the data falls between
factors of 10% and 20%, and 13% of the data falls between factors of 20% and 30% [8].
Li et al. [9] proposed an equation to estimate the cyclic yield strength (S
y
') of
steels from two monotonic tensile properties. In their study, experimentally determined
9
monotonic and cyclic properties from twenty-seven alloy steels from [10, 11] were used
to derive an expression relating cyclic yield strength to ultimate tensile strength (S
u
) and
percent reduction in area (RA). This correlation is defined by the following expression:
16 . 0
) 1 ln(
002 . 0
) 1 (
|
|
.
|

\
|

+ =
'
RA
S RA S
u y
(1.11)
Results from this equation were shown in [9] to deviate from the experimental cyclic
yield strength by 14% at most.
1.3 Purpose of This Study
This thesis evolved from the work conducted for the Bar Steel Fatigue Database
of the American Iron and Steel Institute (AISI). For this thesis study, data from fifty-
seven bar steels from the AISI fatigue database are used. Tensile and fatigue properties of
all the materials listed in the AISI database were generated by the University of Waterloo
(Ontario, Canada) and The University of Toledo.
The objective of this study is to derive a method for estimating the cyclic
deformation properties of steels from common tensile properties and hardness. A variety
of correlations among hardness, tensile data, and cyclic deformation properties are
examined for the fifty-seven steels included in this study and compared to several
correlations from the literature using a different set of steel data from SAE J1099 [11].
The proposed correlations successfully predict the cyclic deformation properties of steels
from both data sets more accurately compared to the correlations from the literature.
Most importantly, the proposed methods require only three monotonic tension properties:
Brinell hardness (HB) or ultimate tensile strength (S
u
), yield strength (S
y
), and modulus
10
of elasticity (E); all of which are either readily available, or can easily be obtained for a
steel. The overall goal of this thesis is to provide a prediction model which can be used to
estimate the cyclic deformation behavior of any steel material from tensile properties and
hardness. Such a cyclic deformation model can be used in conjunction with fatigue life
prediction models currently available in the literature and is shown to significantly
improve upon life predictions based on the monotonic stress-strain curve.











11



Figure 1-1: A plot of superimposed monotonic and cyclic stress-strain curves used to
illustrate both cyclic hardening and softening.
























12











































Figure 1-2: An overview of the fatigue design process and the role of cyclic stress-
strain relationship.
Loads Geometry Stress-Strain Relationship:
-Hookes Law (E)
-Monotonic stress-strain relationship (K, n)
-Cyclic stress-strain relationship (K', n')
-based on experimental K' and n'
-based on predictions from tensile properties
Finite Element Analysis or Stress/Strain Analysis
Stress-Life Approach:
-Experimental S-N curve
-Predicted S-N curve
Strain-Life Approach:
-Experimental -N curve
-Predicted -N curve
Fatigue Life
Stress Distribution Strain Distribution
13



Figure 1-3: Plot of S
u
/S
y
versus stress ratio from [4].

















14

(a)

(b)


Figure 1-4: Estimation of the cyclic strength coefficient from (a) the monotonic
strength coefficient in [7], and (b) Brinell hardness in [8].

15

(a)

Figure 1-5: Estimation of the cyclic strain hardening exponent from Brinell hardness
in [8].



















16
Chapter 2
Experimental Data


2.1 Materials
Fifty-seven bar steels from the American Iron and Steel Institute Bar Fatigue
Database [10] were used to determine correlations among monotonic tensile data,
hardness, and cyclic deformation properties. As mentioned previously in Chapter 1,
tensile and strain-controlled fatigue tests used to determine the various material
properties were performed by the University of Waterloo and the University of Toledo. A
wide variety of steels was used to show that the proposed correlations can work for any
steel material. The ranges of ultimate tensile strength (S
u
) and Brinell hardness (HB) for
these steels are 502 MPa to 2450 MPa and 135 to 584, respectively. Additional tensile
and cyclic deformation properties of sixty-six steels from the SAE J1099 report [11] were
included in this study to confirm the predictive accuracy of the proposed correlations.
The ultimate tensile strength and Brinell hardness ranges of these steels are 279 MPa to
2241 MPa and 80 to 595, respectively.
Summaries of the steels and their available relevant properties used in this study
from the AISI database [10] and SAE J1099 [11] are provided in Tables 2.1 and 2.2,
respectively. The following items are listed: SAE designation, iteration number (from
17
AISI), Brinell hardness (HB), ultimate tensile strength (S
u
), monotonic yield strength
(S
y
), modulus of elasticity (E), percent elongation (EL), reduction in area (RA),
monotonic strength coefficient (K), monotonic strain hardening exponent (n), cyclic yield
strength (S
y
'), cyclic strength coefficient (K'), cyclic strain hardening exponent (n'), and
the ultimate to yield strength ratio (S
u
/S
y
). Not all monotonic properties were available
for some of the steels from SAE J1099 [11]. For those steels missing hardness, Brinell
hardness was calculated from ultimate tensile strength by the following equation from
[12]:
) ( 3 . 3 ) ( 0012 . 0
2
HB HB S
u
+ = (2.1)
The cyclic strength coefficient and cyclic strain hardening exponent values of some of the
SAE J1099 steels were also calculated, as noted in Table 2.2. The compatibility equations
used to calculate these values are introduced in Section 2.3.
In both Tables 2.1 and 2.2, the materials are organized into three groups based on
their ultimate tensile strength to yield strength ratio (S
u
/S
y
). The first group is comprised
of steels with S
u
/S
y
ratio greater than or equal to 1.4, the second group includes steels
with S
u
/S
y
ratios less than or equal to 1.2, and the third group have ratios between 1.2 and
1.4. Within each group, materials are listed according to increasing Brinell hardness.
2.2 Determination of Monotonic and Cyclic Deformation
Properties
2.2.1 Monotonic Tension Properties
The monotonic tension test is considered the most basic mechanical test. For this
type of test, a uniaxial tensile load is applied to a specimen until it breaks. The data
18
obtained from these tests provide information regarding the strength and ductility of a
material. These properties are typically determined using test standards such as ASTM
Standard E8 [13]. The yield strengths reported in Tables 2.1 and 2.2 were determined by
the 0.2% offset method. The monotonic strength coefficient (K) and strain hardening
exponent (n) are determined from true stress-strain data. True stress (o), true strain (c),
and true plastic strain (c
p
) are calculated from engineering stress (S) and engineering
strain (e), according to the following relationships which are based on constant volume
assumption:
) 1 ln( e + = c (2.2)
E
e p
o
c c c c = = (2.3)
The strength coefficient (K) and strain hardening exponent (n) are the intercept and slope
of the best line fit to true stress (o) versus true plastic strain (c
p
) data in log-log scale:
n
p
K

) (c o = (2.4)
In accordance with ASTM Standard E739 [14], when performing the least squares fit, the
true plastic strain is the independent variable and the true stress is the dependent variable.
To generate the K and n values, the range of data is chosen according to the definition of
discontinuous yielding specified in ASTM Standard E646 [15]. Therefore, the valid data
range occurs between the end of yield point extension and the strain at maximum load.
Once values of K and n have been determined, the true stress (o) - true strain (c)
curve is then often represented by the Ramberg-Osgood equation:
n
p e
K E
/ 1
|
.
|

\
|
+ = + =
o o
c c c (2.5)
19
2.2.2 Cyclic Deformation Properties
The cyclic deformation properties used in this study were determined from
constant amplitude strain-controlled fatigue tests. These tests were performed in
accordance with ASTM Standard E606 [16]. These properties were determined from the
steady-state stress-strain response. Steady-state response is defined as the material
deformation response when cyclic stability occurs. The stress-strain curve resulting from
one complete cycle (two reversals) after steady deformation response is reached is
referred to as the steady-state hysteresis loop and is commonly defined at the midlife. An
example of a midlife hysteresis loop is shown in Figure 2-1(a). As shown in this figure,
the calculated true plastic strain is determined by projecting two lines (loading and
unloading) through the horizontal axis at zero stress. These lines are a result of the least-
squares fit to the linear-elastic stress-strain data. The calculated true plastic strain range
(A
p
) is the distance between the two intersection points.
Similar to monotonic deformation behavior, the cyclic deformation behavior can
be characterized by the Ramberg-Osgood type equation:
n
p
e
K E
'
|
.
|

\
|
'
A
+
A
=
A
+
A
=
A
/ 1
2 2 2 2 2
o o
c
c c
(2.6)
As shown in Figure 2-1(b), the cyclic strength coefficient (K') and cyclic strain hardening
exponent (n') are the intercept and slope, respectively, of the best line fit to true stress
amplitude (Ao/2) versus calculated true plastic strain amplitude (Ac
p
/2) data in log-log
scale:
n
p
K
'
|
|
.
|

\
| A
' =
A
2 2
c
o
(2.7)
20
In accordance with ASTM Standard E739 [14], when performing the least squares fit, the
true plastic strain amplitude (Ac
p
/2) is the independent variable and the stress amplitude
(Ao/2) is the dependent variable. The true plastic strain amplitude is calculated by the
following equation:
E
p
2 2 2
o c
c
A

A
=
A
(2.8)
The plastic strain amplitude can also be measured directly from the midlife hysteresis
loop, as shown in Figure 2-1(a). The cyclic yield strength (S
y
') is determined by
substituting a value of 0.002 for the plastic strain amplitude in Equation 2.7 as follows:

n
y
K S
'
' =
'
) 002 . 0 ( (2.9)
If fatigue data are not available, a commonly used shortcut procedure to determine
the cyclic deformation response of a material is the incremental step test. By this method,
a specimen is subjected to increasing blocks of constant strain amplitude. At each strain
amplitude the cyclic stress-strain response is recorded after cyclic stabilization occurs.
Ideally, this method requires the use of only one specimen; however, a specimen may
break before stabilization is reached thus requiring multiple specimens to obtain the
complete family of hysteresis loops.
2.3 Data Exclusion Method Used
The ability of the cyclic deformation parameters (K' and n') to accurately
represent the actual cyclic stress-strain response of a material depends on the goodness of
the least-squares fit used to obtain K' and n'. The experimental data used in Figure 2-1(b)
fit the power expression very well since the R
2
correlation factor is close to 1. In this
21
example, there are several clusters of data. Each cluster consists of duplicate tests
performed at the same strain amplitude. A tight cluster of data indicates consistent
stabilized deformation behavior among duplicate tests. When a material yields
comparable results among duplicate tests of the same strain amplitude, then the cyclic
deformation parameters can be used with confidence to represent the actual cyclic stress-
strain response.
In contrast, significant variation among duplicate tests indicates lack of
repeatability, resulting from a variety of reasons. In this case, the cyclic deformation
parameters obtained will vary greatly depending on which data points are included in the
fit. Consequently, the cyclic deformation parameters obtained from such a fit may not
accurately represent the actual cyclic stress-strain response. Steels from the AISI bar steel
fatigue database [10] with poor least squares fit correlation factors (i.e. less than 0.85)
such as shown in Figure 2-2(a), were excluded from the correlations. Twenty-three out of
eighty-nine AISI steels were excluded from this study for this reason.
Additionally, AISI steels for which maximum plastic strain amplitude data were
less than 0.10% were excluded from the correlations. A steel material exhibiting this
cyclic behavior is shown in Figure 2-2(b). High hardness brittle steels typically behave in
this manner. These materials are incapable of withstanding significant plastic strain.
Therefore, the plastic strain amplitudes measured from the tests conducted for these steels
are too small to provide reliable cyclic deformation properties. This is not a problem,
however, because such high strength steels are typically used in applications in which the
external loading condition produces primarily elastic strains. In this case, the cyclic
22
stress-strain response can be reasonably approximated by Hookes law. Seven AISI steels
with R
2
factors greater than 0.85 were excluded from this study for this reason.
The final AISI data filtering step involved comparing the experimental cyclic
deformation parameters (K' and n') to the calculated values based on the strain-life fatigue
constants (
f
', b,
f
', c) by the following compatibility equations:
c
b
f
f
K
|
.
|

\
|
'
'
= '
c
o
(2.10)
c
b
n = ' (2.11)
The fatigue strength coefficient (
f
') and fatigue strength exponent (b) are obtained from
a least squares fit of true stress amplitude (/2) versus reversals to failure (2N
f
) data in
log-log scale:

b
f f
N ) 2 (
2
'
=
A
o
o
(2.12)
The fatigue ductility coefficient (
f
') and fatigue ductility exponent (c) are obtained from
a least squares fit of true plastic strain amplitude (Ac
p
/2) versus reversals to failure (2N
f
)
data in log-log scale:
c
f f
calculated
p
N
'
) 2 (
2
c
c
=
|
|
.
|

\
| A
(2.13)
Equations 2.10 and 2.11 are commonly referred to as the compatibility equations and are
obtained through inter-relations between Equations 2.7, 2.12, and 2.13. Theoretically, if
the three least squares fits used to obtain the cyclic deformation properties and strain-life
fatigue constants are perfect (i.e. R
2
= 1), then K' and n' calculated from Equations 2.10
and 2.11 will equal the experimental values from Equation 2.7. The magnitude of
23
variation between calculated and experimental values of K' and n' depends on the
goodness of all three fits. Therefore, a plot of experimental vs. calculated K' and n' data
can provide an additional tool to further assess the reliability of cyclic deformation
properties associated with the selected steels. In Figure 2-3(a), a plot of experimental vs.
calculated n' from Equation 2.11 is provided for the selected AISI steels. From this figure
it may be seen that most of the data fall within factors of 25%. Similarly, in a plot of
experimental vs. calculated K' from Equation 2.10 shown in Figure 2-3(b), most of the
data falls within factors of 25%. Those steels whose K' or n' values deviated by more
than 25% were examined further by generating the two plots used to obtain the strain-life
fatigue constants (Equations 2.12 and 2.13) for each. Significant scatter in true stress
amplitude vs. reversals to failure data and/or calculated true plastic strain amplitude vs.
reversals to failure data was observed which would explain the significant deviation
between experimental and calculated K' and n' values. It was concluded, however, that
since these steels satisfy the first two criteria which are based on the least squares fits
used to obtain K' and n', they should be included in the correlations.
Since the stabilized cyclic deformation data (i.e. true stress amplitude vs.
calculated true plastic strain amplitude data) were not available for steels from SAE
J1099 [11], a different exclusion procedure was used. Exclusion of this data was based
solely on agreement between the experimental and calculated (from Equations 2.10 and
2.11) values of K' and n'. Plots of experimental vs. calculated n' and K' are provided in
Figure 2-4. The SAE data shown in these figures is divided into two groups, as specified
in [11]: K' and n' calculated from the compatibility equations (open symbols) and
determined experimentally from a least squares fit of true stress vs. true plastic strain
24
amplitude data (closed symbols). As expected, the calculated values of K' and n' fall on
the 45 line. Factors of 25% and 35% were used to filter the cyclic strength
coefficient (K') and strain hardening exponent (n') experimental SAE data. Steels with n'
or K' data deviating by more than 25% and 35%, respectively, were excluded. A larger
factor (35%) was used to filter the K' data, compared to the factor used to filter the n' data
(25%). This is because the cyclic stress-strain curve is more dependent upon n' than K'.
Twenty out of fifty-four SAE steels with experimental K' and n' data were excluded for
this reason.
Due to the empirical nature of the cyclic deformation correlations proposed in this
study, it is of utmost importance to use only reliable cyclic deformation data to derive the
correlations. Consequently, the materials used in this study were carefully selected to
derive reliable correlations that can be used for a wide variety of steels.











25
Table 2.1: Monotonic and cyclic deformation properties of selected steels from the
AISI bar steel fatigue database [10].

SAE spec.
Iteration
# HB
Su
(MPa)
Sy
(MPa)
E
(GPa)
EL
(%)
RA
(%)
K
(MPa) n
Sy'
(MPa)
K'
(MPa) n' Su/Sy
Su/Sy 1.4
1020 23 135 502 295 207 50 64 933 0.239 317 1171 0.210 1.70
1038 18 163 582 331 201 44 54 1106 0.260 342 1340 0.220 1.76
1541 1 180 783 471 197 39 55 1165 0.143 421 1400 0.193 1.66
1038 19 185 652 359 219 38 53 1186 0.219 358 1420 0.222 1.81
1141Nb 13 199 659 418 220 35 53 1287 0.217 405 1448 0.205 1.58
1050M 3 205 821 450 211 43 50 1819 0.274 413 1987 0.253 1.83
1151V 27 205 761 452 206 26 51 1346 0.190 456 1473 0.189 1.68
10V45 21 212 765 465 213 26 48 1456 0.223 436 1642 0.213 1.64
1141V 15 217 725 450 214 32 49 1321 0.207 447 1467 0.191 1.61
1045 26 222 747 510 209 36 62 1334 0.199 440 1302 0.174 1.47
1141AL 11 223 771 457 216 40 57 1394 0.216 424 1515 0.205 1.68
1141V 17 229 789 494 220 30 47 1379 0.187 481 1441 0.177 1.60
5150 32 245 867 472 210 32 56 1630 0.207 497 1485 0.176 1.84
1090 6 259 1090 729 219 9 14 1780 0.162 535 1877 0.202 1.50
15V41 91 264 1071 666 193 22 33 1670 0.148 712 1575 0.128 1.61
10V45 22 269 909 606 216 24 42 1520 0.168 616 1691 0.162 1.50
1090 10 272 1124 760 214 18 38 1576 0.108 602 1964 0.190 1.48
1090M 9 279 1251 751 212 7 14 2273 0.193 637 1943 0.179 1.67
4620 47 289 998 688 208 34 58 1448 0.109 603 1824 0.178 1.45
1090 8 309 1147 678 213 7 22 1913 0.168 618 2163 0.202 1.69
8620 119 326 991 694 212 32 54 1624 0.140 601 1872 0.183 1.43
8620 120 352 1145 796 210 36 51 2132 0.165 705 2078 0.174 1.44
Su/Sy 1.2
4320 49 188 994 920 202 43 63 1127 0.031 652 799 0.032 1.08
9310 57 258 902 804 197 31 71 1136 0.055 616 1035 0.083 1.12
1141AL 12 277 925 814 227 32 59 1205 0.074 593 1260 0.121 1.14
41B17M (PS19) 72 277 872 827 213 39 68 1031 0.042 611 1029 0.084 1.05
15B35 45 286 940 866 219 37 65 1173 0.056 645 1094 0.085 1.08
4620 53 289 964 892 211 43 62 1200 0.056 667 1213 0.096 1.08
8822 100 297 946 884 212 37 67 1074 0.025 644 1154 0.094 1.07
5140 31 305 1039 957 218 28 53 1276 0.053 702 1329 0.103 1.09
10B21 24 322 1105 1062 212 33 71 1248 0.027 756 1109 0.062 1.04
4140 30 325 1043 929 207 31 61 1303 0.059 670 1460 0.125 1.12
86B20 74 336 1034 989 205 30 64 1213 0.037 705 1268 0.094 1.05
4140 93 353 1514 1363 201 19 48 1911 0.055 911 1614 0.092 1.11
8822 110 371 1723 1528 208 13 50 2175 0.057 1095 2055 0.101 1.13
4140 65 371 1401 1306 200 22 48 1680 0.041 852 1306 0.069 1.07
4140 66 390 1537 1330 204 20 42 2188 0.080 895 1591 0.093 1.16
4140 96 409 1248 1158 203 25 48 1499 0.044 856 1696 0.110 1.08
4140 97 417 1240 1167 207 25 47 1460 0.038 851 1617 0.103 1.06
4130AL 29 442 1483 1285 213 28 44 1984 0.067 915 2151 0.138 1.15
51B60 33 450 1970 1830 200 36 22 2332 0.039 1272 2490 0.108 1.08
9254 86 490 2020 1841 206 8 51 2511 0.050 1381 2316 0.083 1.10
26
9254AL 35 584 2450 2270 205 4 4 3322 0.042 1922 3322 0.088 1.08
1.2 < Su/Sy < 1.4
1022 25 195 604 457 200 48 69 971 0.161 348 1180 0.197 1.32
C-70 43 241 964 765 201 14 17 1315 0.090 572 1528 0.158 1.26
1141Nb 14 241 802 602 217 32 54 1199 0.126 481 1254 0.154 1.33
15V24 28 243 878 646 207 36 61 1318 0.129 615 1453 0.138 1.36
1038 20 245 743 560 218 43 69 1231 0.169 450 1242 0.163 1.33
1141V 16 252 797 610 215 34 58 1244 0.141 487 1280 0.154 1.31
5120 55 252 1008 780 214 46 58 1277 0.074 628 1784 0.168 1.29
20MnCr5 128 271 960 695 194 14 51 1477 0.121 567 1930 0.197 1.38
20MnCr5 129 301 1053 852 194 11 57 1762 0.117 613 2062 0.195 1.24
9310 127 352 1201 990 195 9 57 1796 0.094 798 2098 0.156 1.21
1090M 7 357 1388 1065 217 14 25 2202 0.168 730 2005 0.163 1.30
8620 121 382 1311 990 214 27 52 2276 0.138 927 2137 0.134 1.32
20MnCr5 130 390 1337 1071 194 10 52 1816 0.085 976 2231 0.133 1.25
86B20 82 401 1502 1198 206 60 60 2193 0.092 1075 2118 0.109 1.25






























27
Table 2.2: Monotonic and cyclic deformation properties of selected steels from SAE
J1099 [11].

SAE spec. HB
Su
(MPa)
Sy
(MPa)
E
(GPa)
RA
(%)
K
(MPa) n
Sy'
(MPa)
K'
(MPa) n' Su/Sy
Su/Sy 1.4
1015 80 414 228 207 68 - - 411 945-c 0.213-c 1.82
DDQ 82* 279 150 - - - - 369 1143 0.289 1.86
DDQ+ 86* 292 140 - - - - 292 641 0.201 2.09
1008 86 331 234 207 78 - 0.19 416 1443-c 0.318-c 1.41
DDQ+ 90* 306 152 - - - - 333 832 0.234 2.01
DQSK 90* 307 171 - - - - 322 694 0.196 1.80
1008 90 365 255 203 78 - 0.184 397 1234-c 0.29-c 1.43
1005 94* 321 226 - - - - 437 1208 0.260 1.42
1010 97* 331 200 203 80 534 0.185 334 867-c 0.244-c 1.66
1005 104* 356 234 - - - - 381 834 0.200 1.52
1008 106* 363 252 - - - - 667 1706 0.240 1.44
1020 108 393 255 186 64 400 0.072 403 1233-c 0.286-c 1.54
1020 109 441 262 203 62 738 0.19 559 1962-c 0.321-c 1.68
1025 157* 547 306 207 63 1142 0.281 464 1042-c 0.207-c 1.79
1045 225 752 517 - 44 - - 564 1022-c 0.152-c 1.45
10V40 225* 802 572 - - - - 762 1371 0.150 1.40
Su/Sy 1.2
CQ 103* 352 314 - - - - 297 419 0.088 1.12
HF 50 129* 448 383 - - - - 473 745 0.116 1.17
HF 50 129* 448 385 - - - - 478 785 0.127 1.16
HF 50 138* 479 403 - - - - 562 1014 0.151 1.19
HF 60 138* 479 431 - - - - 588 1029 0.143 1.11
HF 60 139* 481 416 - - - - 476 687 0.094 1.16
HF 50 142* 492 417 - - - - 594 1056 0.147 1.18
HF 60 144* 501 424 - - - - 407 531 0.068 1.18
HF 60 153* 533 459 - - - - 529 792 0.103 1.16
HF 60 153* 534 456 - - - - 604 1134 0.161 1.17
HF 80 176* 617 557 - - - - 698 1125 0.122 1.11
HF 80 178* 635 585 - - - - 665 984 0.100 1.09
HF 80 185* 654 580 - - - - 672 1091 0.124 1.13
HF 80 193* 681 605 - - - - 805 1387 0.139 1.13
HF 80 203* 719 642 - - - - 819 1285 0.115 1.12
1040 213* 759 637 - - - - 548 915 0.131 1.19
15B27 250 847 772 203 69 - 0.075 681 903-c 0.072-c 1.10
10B21 255 834 806 203 - - - 620 858-c 0.083-c 1.03
10B22 255 834 806 203 - - - 645 809-c 0.058-c 1.03
4130 259 896 778 221 67 - - 794 1264-c 0.119-c 1.15
15B27 264 916 854 203 67 - 0.065 765 1026-c 0.075-c 1.07
4140 293 938 848 207 - 1303 0.094 787 1084-c 0.082-c 1.11
10B21 318 1048 999 197 68 1295 0.054 809 1089-c 0.076-c 1.05
4340 318* 1171 1102 207 56 1358 0.036 792 1037-c 0.069-c 1.06
4130 366 1427 1358 200 55 - - 1086 1696-c 0.114-c 1.05
4142 380 1413 1378 207 48 - 0.051 1395 2266-c 0.124-c 1.03
1045 390 1344 1275 207 59 - 0.044 1166 1751-c 0.104-c 1.05
28
4142 400 1551 1447 200 47 - 0.032 1197 1756-c 0.098-c 1.07
4340 409 1468 1371 200 38 - - 1177 1996-c 0.135-c 1.07
5160 430 1584 1488 203 40 1941 0.0463 1333 1964-c 0.099-c 1.06
4142 450 1757 1584 207 42 - 0.043 1415 1997-c 0.088-c 1.11
4142 450 1929 1860 200 37 - 0.016 1354 1910-c 0.088-c 1.04
5160 456* 1755 1565 - - - - 1458 2065-c 0.089-c 1.12
4142 475 1929 1722 207 35 - 0.048 1661 2399-c 0.094-c 1.12
4140 475 2033 1895 200 20 - - 1438 1974-c 0.081-c 1.07
1045 500 1827 1689 207 51 - 0.047 1912 3371-c 0.145-c 1.08
1045 500 1956 1729 207 38 2352 0.041 1821 3366-c 0.157-c 1.13
51V45 535* 2108 1871 - - - - 1968 2799 0.090 1.13
1.2 < Su/Sy < 1.4
1005 95* 323 245 - - - - 436 1254 0.270 1.32
1004 110* 378 287 - - - - 386 781 0.180 1.32
HF 50 128* 441 361 - - - - 421 684 0.124 1.22
HF 50 128* 442 359 - - - - 459 761 0.129 1.23
HF 50 133* 461 375 - - - - 441 632 0.092 1.23
HF 50 141* 490 357 - - - - 397 481 0.049 1.37
HF 60 151* 525 434 - - - - 609 1152 0.163 1.21
HF 80 197* 697 569 - - - - 811 1287 0.118 1.22
HF 80 212* 756 579 - - - - 826 1389 0.133 1.31
4340 243 827 634 193 43 - - 693 1337-c 0.168-c 1.30
4340 275 1048 834 190 - - - 828 1249-c 0.105-c 1.26
1045 595 2241 1862 207 41 - 0.071 2468 3947-c 0.120-c 1.20


Note: c calculated from compatibility Equations 2.10 and 2.11, as stated in [11]
* calculated from Equation 2.1




















29

(a)

(b)

Figure 2-1: Example of (a) mid-life hysteresis loop, and (b) true stress amplitude
versus true plastic strain amplitude plot used to obtain K' and n'.

30

(a)

(b)

Figure 2-2: Example of a steel material excluded from the study due to (a) significant
scatter of true stress versus true plastic strain amplitude data, and (b)
maximum plastic strain amplitude less than 0.10%.
31

(a)

(b)

Figure 2-3: AISI steel data in a plot of (a) experimental versus calculated cyclic strain
hardening exponent from compatibility Equation 2.10, and (b)
experimental versus calculated cyclic strength coefficient from
compatibility Equation 2.9.
32

(a)


(b)

Figure 2-4: SAE J1099 steel data in a plot of (a) experimental versus calculated cyclic
strain hardening exponent from compatibility Equation 2.11, and (b)
experimental versus calculated cyclic strength coefficient from
compatibility Equation 2.10.
33
Chapter 3
Correlations among Tensile and Cyclic Deformation
Properties


3.1 Correlations among Monotonic Properties and Hardness
In this section, correlations among hardness and three common tensile strength
properties including ultimate tensile strength (S
u
), yield strength (S
y
), and strength
coefficient (K) are discussed. The same fifty-seven steel materials from the AISI bar steel
database [10] used to determine cyclic deformation property predictions were used to
determine these correlations. Data from sixty-six SAE J1099 [11] steels were used to
confirm the validity of these correlations.
A correlation was found to exist between ultimate tensile strength (S
u
) and Brinell
hardness (HB). The relationship between S
u
(in MPa) and HB can be described rather
well with the following second-order polynomial equation:
) ( 21 . 3 ) ( 0013 . 0
2
HB HB S
u
+ = (3.1)
for which the square of correlation coefficient was found to be R
2
= 0.86. Roessle and
Fatemi proposed a similar expression relating S
u
(in MPa) and HB using data from sixty-
nine steels in [12]. This expression is as follows:
) ( 3 . 3 ) ( 0012 . 0
2
HB HB S
u
+ = (3.2)
34
Roessle [12] approximated a curve shown in a plot from Dowling [17] which has S
u
(in
MPa) and HB data from low and medium strength carbon and alloy steels from [18]. This
approximation is represented by a straight line as:
) ( 45 . 3 HB S
u
~ (3.3)
In Figure 3-1(a), Equations 3.1 through 3.3 are superimposed with experimental
data from AISI [10] and SAE J1099 [11]. It can be seen from this figure that all three
curves converge with decreasing hardness. Equation 3.3 produces conservative
predictions compared to Equations 3.1 and 3.2 at high hardness. From comparison of
Equations 3.1 and 3.2 in Figure 3-1(a) it may be seen that both models yield nearly
identical predictions of ultimate tensile strength (S
u
). Given that these equations were
determined independently from two different sets of steel data, it can be concluded that
Equation 3.2 from [12] provides a valid correlation that can be used to predict S
u
of most
steel materials reasonably well.
The yield strength (S
y
) was also found to correlate with Brinell hardness (HB).
The least squares fit obtained from a second-order polynomial had good agreement with
the AISI data. The proposed correlation (with R
2
= 0.87) is given by:
) ( 62 . 1 ) ( 0039 . 0
2
HB HB S
y
+ = (3.4)
where S
y
is in MPa. Roessle examined a similar correlation between S
y
(in MPa) and HB
in [19]. This correlation is defined by the following expression:
) ( 33 . 1 ) ( 0044 . 0
2
HB HB S
y
+ = (3.5)
Equations 3.4 and 3.5 are superimposed with experimental data from AISI [10] and SAE
J1099 [11] in Figure 3-1(b). It can be seen from this figure that Equations 3.4 and 3.5
produce approximately the same predictions of S
y
.
35
Another correlation examined was that between the monotonic strength
coefficient (K) and Brinell hardness (HB). The following linear equation was used to fit
the AISI data:
365 ) ( 09 . 4 + = HB K (3.6)
for which the square of correlation coefficient was found to be R
2
= 0.59. Equation 3.6 is
superimposed with experimental data from AISI [10] and SAE [11] in Figure 3-1(c).
It can be seen from Figures 3-1(a), 3-1(b), and 3-1(c), that no significant
difference between the scatter of AISI [10] and SAE J1099 [11] data exists. Therefore, it
can be concluded that correlations obtained from least squares fits of AISI data are valid
since they can be used to predict monotonic properties of SAE J1099 steels with
approximately the same degree of accuracy. In addition, the scatter among ultimate
strength and yield strength data is less than that of strength coefficient data when plotted
against hardness as made apparent by comparison of the R
2
values. One explanation for
the significant scatter observed among monotonic strength coefficient data is discussed in
Section 3.3.
3.2 Classification of Materials Based on the S
u
/S
y
Ratio
The ratio S
u
/S
y
was used to classify the steels used in this study. As a result, all of
the steels have been placed in one of three groups. The first group includes steels with
S
u
/S
y
1.4, the second group includes steels with S
u
/S
y
1.2, and the third group
includes steels with intermediate values of 1.2 < S
u
/S
y
< 1.4. It is shown in the following
sections that this classification method improves correlations among cyclic deformation
properties and monotonic tensile properties/hardness.
36
Correlations among cyclic deformation properties, monotonic properties, and
hardness have been made individually for steels within these groups (based on the S
u
/S
y

ratio). In Figure 3-2, a plot of ultimate tensile strength (S
u
) vs. yield strength (S
y
) is
provided for AISI and SAE J1099 data to illustrate this classification method. Different
symbols and colors are used to distinguish the data according to their S
u
/S
y
ratio.
Hereafter, these same symbols will be used in all the figures that include correlations.
3.3 Comparison of Deformation Behavior between Steels of
the Same Hardness
In general, the strength of steel increases with increasing hardness, as was shown
in Section 3.2. Although correlations among three monotonic properties (S
u
, S
y
, and K)
and hardness were found, the strongest correlations were observed for ultimate tensile
strength (S
u
) and yield strength (S
y
) vs. Brinell hardness (HB). The correlations between
S
u
and HB from the literature are shown again in Figure 3-3(a), but in this plot the data
are grouped according to their S
u
/S
y
ratio. It may be seen from this figure that no
distinction among steels of different S
u
/S
y
material groups exists. However, in a plot of
yield strength (S
y
) vs. Brinell hardness (HB), a clear distinction among steels of different
S
u
/S
y
material groups exists. This plot is provided in Figure 3-3(b). Similarly, in Figure 3-
3(c), this distinction can be observed in a plot of monotonic strength coefficient (K) vs.
Brinell hardness (HB). In this figure, the strength coefficient data of the S
u
/S
y
1.4 group
is greater than the strength coefficient data of the S
u
/S
y
1.2 group at any hardness.
Strength coefficient data of the intermediate group (1.2 < S
u
/S
y
< 1.4) is generally in
between these two groups. Since the strength coefficient is one of the properties that
describes the deformation behavior, it can be concluded that scatter among strength
37
coefficient data is attributed to differences in deformation behavior between steels of the
same hardness but different S
u
/S
y
ratios. This is illustrated in a plot of S
u
/S
y
vs. Brinell
hardness as shown in Figure 3-4. In this figure, the S
u
/S
y
ratios vary significantly among
steels of the same hardness.
A significant difference in the cyclic stress-strain behavior between two steels of
the same hardness can also exist. To illustrate this, two steels of the same hardness but
different S
u
/S
y
ratios have been selected. The materials are 1090 steel with a S
u
/S
y
ratio of
1.50 and 9310 steel with a ratio of 1.12. In Figure 3-5(a), a plot of true stress amplitude
vs. true plastic strain amplitude in log-log scale is provided. It can be seen from this
figure that the cyclic strength coefficient (K') and the cyclic strain hardening exponent
(n') of the two steels are significantly different. In Figure 3-5(b), the monotonic and
cyclic stress-strain curves are superimposed for both steels. The monotonic and cyclic
stress-strain curves were obtained from the Ramberg-Osgood equation of which
experimental values of K, n, K', and n' were used. Several observations can be made
regarding this figure. The monotonic and cyclic yield strengths of 1090 steel are lower
than those of 9310 steel. The slopes of the stress-strain curves (monotonic and cyclic) of
1090 steel are higher than those of 9310 steel and both steels cyclically soften, but 9310
steel softens to a larger degree. Clearly, significant differences exist in both the
monotonic and cyclic deformation properties between these two steels of the same
hardness. Therefore, it may be concluded that hardness alone cannot be used to predict
the cyclic deformation properties of steels with high accuracy.
It was shown in Figures 3-3(b) and 3-3(c) that correlations obtained from data
within the material groups provide stronger agreement between yield strength or strength
38
coefficient and hardness, compared to the correlations obtained from all the data.
Although these correlations are not useful because they require monotonic properties to
determine the various material groups, they indicate the possibility for improving the
predictive accuracy of cyclic deformation correlations by grouping steels according to
their S
u
/S
y
ratios.
3.4 Correlations among Monotonic Properties or Hardness
and the Cyclic Strength Coefficient
In Section 3.3 it was shown that correlations among yield strength (S
y
) or
monotonic strength coefficient (K) and hardness are improved when considered within
different material groups. A similar trend was found from correlations between the cyclic
strength coefficient (K') and various monotonic properties and hardness. Within the
various material groups, the cyclic strength coefficient (K') was found to be dependent
upon Brinell hardness (HB), ultimate tensile strength (S
u
), yield strength (S
y
), and
monotonic strength coefficient (K) to varying degrees.
In Figure 3-6(a), a plot of the cyclic strength coefficient (K') vs. Brinell hardness
(HB) is provided. It can be seen from this figure that a definitive separation exists
between data of S
u
/S
y
> 1.2 and S
u
/S
y
1.2. For the entire range of hardness shown, the
cyclic strength coefficient values of steels with S
u
/S
y
> 1.2, for the most part, are greater
than those of steels with S
u
/S
y
1.2. A line represents data from the S
u
/S
y
> 1.2 group
reasonably well. A good agreement exists between the data of steels with S
u
/S
y
1.2 and
a second-order polynomial least squares fit. The correlations are given by (units: MPa):
613 ) ( 09 . 4 + = ' HB K for S
u
/S
y
> 1.2 (3.7)
39
705 ) ( 26 . 1 ) ( 10 8 . 9
2 3
+ = '

HB HB K for S
u
/S
y
1.2 (3.8)
In Figure 3-6(b), a plot of the cyclic strength coefficient (K') vs. ultimate tensile
strength (S
u
) is provided. The data in this figure produce similar trend curves as those
observed in Figure 3-6(a). This is expected given the strong correlation between Brinell
hardness (HB) and ultimate tensile strength (S
u
) which was shown in Figure 3-1(a). The
correlations are (units: MPa):
593 ) ( 16 . 1 + = '
u
S K for S
u
/S
y
> 1.2 (3.9)
619 ) ( 23 . 0 ) ( 10 0 . 3
2 4
+ + = '

u u
S S K for S
u
/S
y
1.2 (3.10)
In Figure 3-6(c), a plot of the cyclic strength coefficient (K') vs. monotonic yield
strength (S
y
) is provided. The data in this figure produce similar correlations as those
observed in Figures 3-6(a) and 3-6(b). The K' values of steels with S
u
/S
y
> 1.2 are greater
than those of steels with S
u
/S
y
1.2 for the entire range of S
y
. The correlations are given
by (units: MPa):
937 ) ( 15 . 1 + = '
y
S K for S
u
/S
y
> 1.2 (3.11)
514 ) ( 43 . 0 ) ( 10 3
2 4
+ + = '

y y
S S K for S
u
/S
y
1.2 (3.12)
In Figure 3-6(d), a plot of cyclic vs. monotonic strength coefficient is provided.
Correlations obtained from a linear least squares fit were determined for the material
groups individually. The correlations are as follows (units: MPa):
500 ) ( 77 . 0 + = ' K K for S
u
/S
y
> 1.2 (3.13)
41 ) ( 94 . 0 + = ' K K for S
u
/S
y
1.2 (3.14)
Although the square of correlation coefficients suggest a dependence of the cyclic
strength coefficient (K') on the monotonic strength coefficient (K) upon separation of
40
material groups, it can be seen from this figure that most of the data is clustered about the
diagonal line (K' = K). This is particularly true for steels with monotonic strength
coefficient values less than 1,800 MPa, as suggest by Roessle in [19].
The dependence of the cyclic strength coefficient (K') on S
u
/S
y
was also
investigated. It can be seen from Figure 3-7 that no correlation exists between these
parameters.
From comparison of Figures 3-6(a) through 3-6(d), it can be seen that correlations
which use ultimate tensile strength (S
u
) to predict the cyclic strength coefficient (K')
provide the best agreement between the data within each material group and the least
squares fit. The correlations shown in Figure 3-6(b) for K' vs. S
u
provide better
agreement between the data and the least squares fit for both groups, compared to the
correlations shown in Figure 3-6(c) for K' vs. S
y
.
From comparison of Figures 3-6(a) for K' vs. HB and 3-6(b) for K' vs. S
u
it can be
seen that the relative positions of the data and prediction curves are very similar for both
plots. This is expected considering the strong correlation between ultimate tensile
strength (S
u
) and Brinell hardness (HB) that was shown in Figure 3-1(a). As was
mentioned in Section 3.1, this correlation works well for all material groups and is,
therefore, independent of the S
u
/S
y
ratio. Based on this observation, it may be concluded
that correlations which use ultimate tensile strength and Brinell hardness to predict the
cyclic strength coefficient (K') yield very similar results. This may serve as a useful tool
for situations in which one of the two properties (S
u
and HB) is not available. Therefore,
the correlations found between K' and HB and between K' and S
u
are proposed in this
study.
41
3.5 Correlations among Monotonic Properties or Hardness
and the Cyclic Strain Hardening Exponent
Roessle [19] examined a correlation between the cyclic strain hardening exponent
(n') and the monotonic strain hardening exponent (n) as follows:
10 . 0 ) ( 42 . 0 + = ' n n (3.15)
This correlation was based on experimental data of 20 steels from the AISI database [10]
and resulted in poor agreement between the data and linear least squares fit. In Figure 3-
8(d), a plot of cyclic strain hardening exponent (n') vs. monotonic strain hardening
exponent (n) is provided. It can be seen from this figure that steels with S
u
/S
y
1.2 have
relatively small n and cyclic n' values, whereas steels with S
u
/S
y
1.4 have relatively
large n and n' values. The steels with 1.2 < S
u
/S
y
< 1.4 have somewhat intermediate n and
n' values. A linear least squares fit to all of the data produced the best correlation among
these properties as follows:
07 . 0 ) ( 64 . 0 + = ' n n (R
2
= 0.78) (3.16)
This correlation resembles the correlation examined by Roessle in [19] which is defined
by Equation 3.15. It is clear from this figure that Equations 3.15 and 3.16 provide an
improved estimation of n' compared to n' estimated from n' = n.
No correlations were found to exist between the cyclic strain hardening exponent
(n') and Brinell hardness (HB), ultimate strength (S
u
), or yield strength (S
y
), as shown in
Figures 3-8(a) through 3-8(c), respectively. However, a correlation between n' and the
S
y
/S
u
ratio obtained from a linear least squares fit had good agreement with the data. In
Figure 3-9(a), a plot of the cyclic strain hardening exponent (n') vs. S
y
/S
u
is provided. The
proposed correlation between n' and S
y
/S
u
is as follows:
42
40 . 0 33 . 0 +
|
|
.
|

\
|
= '
u
y
S
S
n (R
2
= 0.79) (3.17)
A similar correlation was found between the cyclic strain-hardening exponent (n') and the
inverse of S
y
/S
u
. This correlation is shown in Figure 3-9(b). The proposed correlation
between n' and S
u
/S
y
is as follows:
08 . 0 17 . 0
|
|
.
|

\
|
= '
y
u
S
S
n (R
2
= 0.74) (3.18)
Although the scatter in data from Figure 3-8(d) for n' vs. n is comparable to the
scatter in data from Figures 3-9(a) for n' vs. S
y
/S
u
and 3-10(b) for n' vs. S
u
/S
y
, correlations
which use the ultimate tensile strength (S
u
) and yield strength (S
y
) are preferable because
the monotonic strain hardening exponent (n) is not as commonly available. From
comparison of the correlations shown in Figures 3-9(a) and 3-9(b) it can be seen that the
correlation between n' and S
y
/S
u
defined by Equation 3.17 provides the best agreement
between the data and the linear least squares fit. Therefore, this correlation is proposed.
3.6 Correlations among Monotonic Properties or Hardness
and the Cyclic Yield Strength
The cyclic yield strength (S
y
') is calculated from Equation 2.9. This property can
be used to determine the stress amplitude at which significant cyclic plastic deformation
starts to occur under cyclic loading. Therefore, the cyclic yield strength, if successfully
predicted from monotonic properties or hardness, can serve as an important property for
fatigue design. In this section, correlations between the cyclic yield strength and
monotonic properties or hardness are discussed.
43
A plot of the cyclic yield strength (S
y
') vs. Brinell hardness (HB) is shown in
Figure 3-10(a). It can be seen from this figure that a second-order polynomial represents
all of the data well (R
2
= 0.89). This correlation is given by (units: MPa):
) ( 49 . 1 ) ( 10 5 . 2
2 3
HB HB S
y
+ =
'
(3.19)
In Figure 3-10(b), a plot of the cyclic yield strength (S
y
') vs. ultimate tensile strength (S
u
)
is provided. It can be seen from this figure that there is good agreement between all the
data and the second-order polynomial least squares fit (R
2
= 0.94). This correlation is
given by (units: MPa):
) ( 54 . 0 ) ( 10 0 . 8
2 5
u u y
S S S + =
'
(3.20)
This correlation resembles that shown in Figure 3-10(a). This is expected considering the
strong correlation between the ultimate tensile strength and hardness. Since both of these
correlations provide a good estimation of S
y
', both are proposed.
In Figure 3-10(c), a plot of the cyclic yield strength (S
y
') vs. the monotonic yield
strength (S
y
) is provided. It can be seen from this figure that a strong correlation exists
between the data of steels with S
u
/S
y
1.2 and a second-order polynomial least squares
fit for which R
2
= 0.99. A linear fit represents combined data from the other two groups
well (R
2
= 0.88) for which S
u
/S
y
> 1.2. Therefore, two equations which predict S
y
' from
S
y
are proposed. The proposed correlations are as follows (units: MPa):
82 ) ( 75 . 0 + =
'
y y
S S for S
u
/S
y
> 1.2 (3.21)
526 ) ( 15 . 0 ) ( 10 0 . 3
2 4
+ =
'
y y y
S S S for S
u
/S
y
1.2 (3.22)
44
3.7 Comparisons between Selected Methods
In this section, comparisons are made using the methods introduced in Section 1.2
from the literature and the correlations found in this study. Data from AISI [10] and SAE
J1099 [11] listed in Tables 2.1 and 2.2 are used to compare cyclic deformation properties
predicted from the various correlations. It should be mentioned that since the correlations
were obtained from fittings of the AISI data used in this study, good predictions based on
the proposed correlations are expected. However, since the SAE J1099 data were not
used for deriving the proposed correlations, the different prediction methods can be
compared for this data without a bias for any of the methods used.
3.7.1 Comparison of Cyclic Deformation Behavior Predictions
(Hardening or Softening)
Three methods from the literature for predicting cyclic hardening/softening
behavior were introduced in Section 1.2. In this section, these methods are compared
using the fifty-seven steels from AISI [10]. The tensile parameters used by the methods
from literature of these steels as well as predictions based on Brinell hardness and
ultimate tensile strength are listed in Table 3.1.
The actual cyclic hardening/softening behavior is defined through the relative
position between the experimental tensile stress-strain curve and the cyclic stress-strain
curve based on the experimental cyclic strength coefficient (K') and strain hardening
exponent (n'). The experimental tensile stress-strain curve is used for comparison with the
cyclic curve, rather than the tensile curve based on the experimental strength coefficient
(K) and strain hardening exponent (n), since the Ramberg-Osgood equation cannot often
45
account for yielding behavior. When the cyclic stress-strain curve is above the tensile
stress-strain curve, the material behaves in a cyclic hardening manner (H), but when the
cyclic stress-strain curve lies below the tensile stress-strain curve, the material behaves in
a cyclic softening manner (S). When the tensile and cyclic stress-strain curves intersect
(i.e. the cyclic curve initially below the monotonic curve, then above at larger strains), the
material behaves in a mixed-mode manner (M). For a mixed-mode material, softening is
followed by hardening. The three aforementioned cyclic behaviors were observed for the
fifty-seven steels from AISI [10] used in this study. In Table 3.1, these steels are grouped
according to their experimental cyclic behaviors. Within each group (Hardening,
Softening, and Mixed-Mode), steels are listed by increasing hardness.
It was mentioned in Section 1.1.3 that hard steel materials typically cyclically
soften, whereas soft steel materials typically cyclically harden. Given the strong
correlation between ultimate tensile strength and Brinell hardness that was shown in
Figure 3-1(a), it may also be concluded that high strength steels cyclically soften and low
strength steels cyclically harden. This hypothesis was tested using the AISI steel
materials. With the exception of two materials, all the steels either softened or had mixed-
mode behavior. Results from Table 3.1 indicate that 89% of the cyclically softening
materials have S
u
greater than 920 MPa and 92% have HB greater than 250. Of the steels
exhibiting mixed-mode behavior, 94% have S
u
less than 920 MPa and 88% have HB less
than 250. Based on these results, it may be concluded that, as a rough guide, steels with
S
u
less than 920 MPa and HB less than 250 can be expected to behave in a mixed-mode
manner (i.e. initial softening, followed by hardening), and steels with S
u
greater than 920
MPa and HB greater than 250 can be expected to cyclically soften. Percentages of
46
successfully predicted behaviors (hardening, softening, and mixed-mode) based on S
u
,
HB, and three methods from the literature are listed in Table 3.2. It can be seen from this
table that the methods from the literature result in poor cyclic behavior predictions.
Overall, predictions based on S
u
or HB, as previously suggested, provide the most
correctly predicted cyclic behaviors.
3.7.2 Comparison of Cyclic Strength Coefficient Predictions
The cyclic strength coefficient (K') predicted from Equations 1.8 and 1.9, and
Equations 3.7 through 3.10 are compared in this section. For each method, data from
AISI [10] and SAE J1099 [11] are superimposed in a plot of predicted vs. experimental
K'. Cyclic strength coefficient values predicted from Equations 1.8 [7] and 1.9 [8] are
plotted in Figures 3-11(a) and 3-11(b), respectively. K' values predicted from ultimate
tensile strength (Equations 3.9 and 3.10) and Brinell hardness (Equations 3.7 and 3.8) are
plotted in Figures 3-11(c) and 3-11(d), respectively. The equations used to calculate the
predicted K' values are shown in each figure. For comparison purposes, a 45 line and
scatter bands of factors 10%, 20%, and 30% are included in each plot, along with the
percentages of AISI, SAE, and total data within each factor.
It can be seen in Figures 3-11(a) through 3-11(d) that the methods proposed in
this study had the most data (from AISI and SAE J1099) within all deviation factors
compared to the methods from literature. From comparison of Figures 3-12(c) and 3-
12(d), it can be seen that prediction of K' from Brinell hardness or from ultimate strength
have about the same degree of accuracy.
47
3.7.3 Comparison of Cyclic Strain Hardening Exponent Predictions
In Section 3.5, a good correlation was shown between the cyclic strain hardening
exponent (n') and S
y
/S
u
which is defined by Equation 3.17. The cyclic strain hardening
exponent can also be estimated indirectly by substituting predicted values of the cyclic
strength coefficient (K') and cyclic yield strength (S
y
') into Equation 2.9. By solving for
n', the following equation is derived:
|
|
|
.
|

\
|
'
'
= '
K
S
n
y
log 37 . 0 (3.23)
Since good correlations between K' and S
u
or HB (Equations 3.7 through 3.10) and
between S
y
' and S
y
(Equations 3.21 and 3.22) were found for each material group,
substitution of predicted K' and S
y
' into Equation 3.23 may provide a very good
estimation of n'. After substitution of Equations 3.7 and 3.8 and Equations 3.21 and 3.22
into Equation 3.23, the following relationships are obtained based on S
y
(in MPa) and
HB:
|
|
.
|

\
|
+
+
=
613 ) ( 09 . 4
937 ) ( 75 . 0
log 37 . 0 '
HB
S
n
y
for S
u
/S
y
> 1.2 (3.24)
|
|
.
|

\
|
+
+
=

705 ) ( 26 . 1 ) ( 10 8 . 9
526 ) ( 15 . 0 ) ( 10 0 . 3
log 37 . 0 '
2 3
2 4
HB HB x
S S x
n
y y
for S
u
/S
y
1.2 (3.25)
Substitution of Equations 3.9 and 3.10 and Equations 3.21 and 3.22 into Equation 3.23,
the following relationships are obtained based on S
y
(in MPa) and S
u
(in MPa):
|
|
.
|

\
|
+
+
=
593 ) ( 16 . 1
937 ) ( 75 . 0
log 37 . 0 '
u
y
S
S
n for S
u
/S
y
> 1.2 (3.26)
48
|
|
.
|

\
|
+ +
+
=

619 ) ( 23 . 0 ) ( 10 0 . 3
526 ) ( 15 . 0 ) ( 10 0 . 3
log 37 . 0 '
2 4
2 4
u u
y y
S S x
S S x
n for S
u
/S
y
1.2 (3.27)
These methods along with two methods from the literature (Equations 1.5 through 1.7
from [7], and 1.10 from [8]) are compared in a plot of predicted vs. experimental cyclic
strain hardening exponent (n'). It should be mentioned that only AISI data are shown in
Figure 3-12(a) since Equations 1.5 through 1.7 require the true fracture strength (
f
) to
predict n' and this property is not available for the steels listed in SAE J1099 [11].
Figures 3-12(b) through 3-12(e) have both AISI and SAE J1099 data included.
From comparison of Figures 3-12(a) through 3-12(e) it can be seen that the
prediction methods based on Equation 3.17 shown in Figure 3-12(c) and Equations 3.26
and 3.27 shown in Figure 3-12(e) had the most data (from AISI and SAE J1099) within
all factors, compared to the other methods. Although both prediction methods result in a
similar degree of accuracy, Equation 3.17 is simpler to use.
3.7.4 Comparison of Cyclic Yield Strength Predictions
Three methods to predict the cyclic yield strength (S
y
') directly from Brinell
hardness (HB), ultimate tensile strength (S
u
), and yield strength (S
y
) were introduced in
Section 3.6 and are defined by Equations 3.19 through 3.22. The cyclic yield strength can
also be estimated indirectly by substituting predicted values of the cyclic strength
coefficient (K') and strain hardening exponent (n') into Equation 2.9. By this method, K'
is estimated from HB and S
u
via Equations 3.7 through 3.10, while n' is estimated from
S
y
/S
u
via Equation 3.17. These proposed methods along with a method from the literature
49
(Equation 1.11 from [9]) are compared in this section. For each method, a plot of
predicted vs. experimental cyclic yield strength (S
y
') was made.
Predicted cyclic yield strength data from the various methods are shown in
Figures 3-13(a) through 3-13(f). From comparison of these figures it can be seen that the
direct correlations based on Equations 3.21 and 3.22 shown in Figure 3-13(c) and
predictions based on Equation 1.11 from [9] shown in Figure 3-13(f) provide the best
overall predictions for combined AISI and SAE J1099 data.
3.8 Approximations of the Cyclic Stress-Strain Curve
It was mentioned in Chapter 1 that the primary purpose of this study was to
develop a method to predict the cyclic stress-strain curve of any steel material based on
monotonic properties and/or hardness. The methods proposed in this study involve
estimating the cyclic strength coefficient (K') and the cyclic strain hardening exponent
(n') individually from monotonic properties and/or hardness. Based on the results
discussed in Section 3.7, the correlations that produce the best estimation of the cyclic
stress-strain curve are proposed in this section. Predicted cyclic stress-strain curves based
on the proposed methods are superimposed with the experimental cyclic and monotonic
stress-strain curves of the AISI steels. These plots are used to show the predictive
accuracy of the proposed method in relation to the experimental cyclic stress-strain curve.
Plots of superimposed cyclic stress-strain curves were also generated for SAE J1099
steels that have experimental K' and n' data.
50
The proposed estimation method is as follows. For steels of which S
u
/S
y
is greater
than 1.2, the cyclic strength coefficient (K') and cyclic strain hardening exponent (n') can
be estimated from the following equations (where S
u
, S
y
, and K' are in MPa):
613 ) ( 09 . 4 ' + = HB K (3.28)
593 ) ( 16 . 1 ' + =
u
S K (3.29)
|
|
.
|

\
|
+
+
=
593 ) ( 16 . 1
937 ) ( 75 . 0
log 37 . 0 '
u
y
S
S
n (3.30)
For steels of which S
u
/S
y
is less than or equal to 1.2, the cyclic strength coefficient
(K') and cyclic strain hardening exponent (n') can be estimated from the following
equations (where S
u
, S
y
, and K' are in MPa):
705 ) ( 26 . 1 ) ( 10 8 . 9
2 3
+ = '

HB HB K (3.31)
619 ) ( 23 . 0 ) ( 10 0 . 3
2 4
+ + = '

u u
S S K (3.32)
|
|
.
|

\
|
+ +
+
=

619 ) ( 23 . 0 ) ( 10 0 . 3
526 ) ( 15 . 0 ) ( 10 0 . 3
log 37 . 0 '
2 4
2 4
u u
y y
S S x
S S x
n (3.33)
The cyclic strain hardening exponent (n') can also be estimated from the
following equation (where S
u
and S
y
are in MPa):
40 . 0 33 . 0 +
|
|
.
|

\
|
= '
u
y
S
S
n
(3.34)
Equation 3.34 can be used for steels of any S
u
/S
y
ratio.
In Figures 3-14 through 3-21, the predicted cyclic stress-strain curves of steels
from the AISI database [10] are superimposed with their experimental monotonic and
cyclic stress-strain curves. Predicted cyclic stress-strain curves are superimposed with
experimental cyclic stress-strain curves of steels from SAE J1099 [11] in Figures 3-22
51
through 3-26. The experimental monotonic and cyclic stress-strain curves were plotted
using Equations 2.5 and 2.6, respectively, for which experimental values of monotonic
and cyclic strength coefficient and strain hardening exponent were substituted. The
predicted cyclic stress-strain curves were obtained by substituting predicted values of K'
and n' into Equation 2.6. K' values calculated from Equations 3.9 and 3.10 and n' values
calculated from Equations 3.26 and 3.27 were used to obtain the curves labeled Cyclic
Pred. 1. K' values calculated from Equations 3.9 and 3.10 and n' values calculated from
Equation 3.18 were used to obtain the curves labeled Cyclic Pred. 2.
It was mentioned in Chapter 1 that the experimental monotonic stress-strain curve
based on Equation 2.5 is sometimes used to estimate the cyclic stress-strain relationship
in the absence of cyclic deformation properties. It can be seen from Figures 3-14 through
3-19 that many steels exhibit significant cyclic hardening or softening. This is made
apparent by comparison of the experimental monotonic and cyclic stress-strain curves.
From superposition of all four stress-strain curves, the accuracy of the predicted cyclic
stress-strain curves can be easily determined by observation of their relative positions.
The stress-strain curves of AISI database steels with S
u
/S
y
1.4 are shown in
Figures 3-14 through 3-16. It may be seen from these figures that there is no distinct
advantage from using one prediction method over the other. In Figure 3-22, the cyclic
stress-strain curves of SAE J1099 steel data with S
u
/S
y
1.4 are shown. Prediction
method 1 provides a better approximation of the experimental cyclic stress-strain curve
for most of the SAE J1099 database steels in this figure, compared to prediction method
2. Although method 1 provides a better approximation compared to prediction method 2
52
for most of the SAE J1099 database steels in this group, the method based on Equation
3.34 Cyclic Pred. 2 is simpler to use.
In Figures 3-17 through 3-19, the stress-strain curves of AISI steel materials with
S
u
/S
y
1.2 are shown. All of the steel materials included in this material group cyclically
soften. Prediction method 1 more closely approximates the experimental cyclic stress-
strain curve compared to prediction method 2 for most of the AISI database steels in this
group. In Figures 3-23 and 3-24, the cyclic stress-strain curves of SAE J1099 database
steels with S
u
/S
y
1.2 are shown. More experimental cyclic stress-strain curves of the
steels from these figures are closely approximated by the prediction method 2 than by
prediction method 1.
Steel materials from the AISI database in the intermediate material group for
which 1.2 < S
u
/S
y
< 1.4 are shown in Figures 3-20 and 3-21. Prediction method 1 is
shown to provide a better approximation of the cyclic stress-strain curve compared to
prediction method 2 and the monotonic stress-strain curve for most of the AISI database
steels in this group. In Figures 3-25 and 3-26, the cyclic stress-strain curves of SAE
J1099 database steels with 1.2 < S
u
/S
y
< 1.4 are shown. Prediction method 1 provides a
better approximation of the experimental cyclic stress-strain curve for most of the SAE
J1099 database steels in this group compared to prediction method 2.








53
Table 3.1: Cyclic deformation behavior predictions from various tensile properties
for selected steels from the AISI bar steel fatigue database [10].










SAE spec.
Iteration
# S
u
HB S
u
/S
y
n

(%)
5150 32 867 245 1.84 0.207 44.6
15V41 91 1071 264 1.61 0.148 12.9
4320 49 994 188 1.08 0.031 62.4
C-70 43 964 241 1.26 0.090 3.4
1141Nb 14 802 241 1.33 0.126 43.2
5120 55 1008 252 1.29 0.074 52.2
1141V 16 797 252 1.31 0.141 52.2
9310 57 902 258 1.12 0.055 85.0
1090 6 1090 259 1.50 0.162 2.8
20MnCr5 128 960 271 1.38 0.121 37.8
1090 10 1124 272 1.48 0.108 18.8
1141AL 12 925 277 1.14 0.074 53.1
41B17M (PS19) 72 872 277 1.05 0.042 77.3
1090M 9 1251 279 1.67 0.193 1.4
15B35 45 940 286 1.08 0.056 6.5
4620 53 964 289 1.08 0.056 61.9
4620 47 998 289 1.45 0.109 52.4
8822 100 946 297 1.07 0.025 74.9
20MnCr5 129 1053 301 1.24 0.117 47.0
5140 31 1039 305 1.09 0.053 42.2
10B21 24 1105 322 1.04 0.027 86.0
4140 30 1043 325 1.12 0.059 61.3
8620 119 991 326 1.43 0.140 41.9
86B20 74 1034 336 1.05 0.037 64.5
8620 120 1145 352 1.44 0.165 36.1
9310 127 1201 352 1.21 0.094 47.0
4140 93 1514 353 1.11 0.055 31.2
1090M 7 1388 357 1.30 0.168 7.5
8822 110 1723 371 1.13 0.057 33.6
4140 65 1401 371 1.07 0.041 33.7
8620 121 1311 382 1.32 0.138 38.3
4140 66 1537 390 1.16 0.080 25.3
20MnCr5 130 1337 390 1.25 0.085 38.5
86B20 82 1502 401 1.25 0.092 54.3
4140 96 1248 409 1.08 0.044 30.7
4140 97 1240 417 1.06 0.038 30.2
4130AL 29 1483 442 1.15 0.067 26.6
51B60 33 1970 450 1.08 0.039 4.3
9254 86 2020 490 1.10 0.050 36.2
9254AL 35 2450 584 1.08 0.042 0.0
H M M H H S
H S S H M H
S M M - M S
S S S - S S
S S M S S S
S S M - S H
S S S H M H
S S S - M S
S M S - M S
S M S S S H
S M S S S H
S S S H M S
S S S H M H
S S S S S S
S S S H M S
S S S S S H
S S S S S H
S S S S S S
S S S S S H
S S S S S S
S S S - M S
S S S S S S
S S S H M S
S S S - S S
S S S H M S
S S S S S S
S S S S S S
S S S S S S
S S S S S S
S S S - M H
S S S - S S
S S S - S S
S S S - M S
S S S S S S
S S S S S S
S S S S S H
S S S S S S
S S S S S S
S S S S S S
S S S S S S
Experimental
Behavior
Prediction
based on S
u
Prediction
based on HB
Prediction
based on S
u
/S
y
Prediction
based on n
Prediction
based on
54


Note: H: Hardening
S: Softening
M: Mixed-Mode




























1020 23 502 135 1.70 0.239 64.1
1038 18 582 163 1.76 0.260 43.2
1541 1 783 180 1.66 0.143 44.1
1038 19 652 185 1.81 0.219 42.4
1022 25 604 195 1.32 0.161 79.7
1141Nb 13 659 199 1.58 0.217 42.4
1050M 3 821 205 1.83 0.274 34.7
1151V 27 761 205 1.68 0.190 35.4
10V45 21 765 212 1.64 0.223 33.6
1141V 15 725 217 1.61 0.207 34.3
1045 26 747 222 1.47 0.199 62.0
1141AL 11 771 223 1.68 0.216 51.3
1141V 17 789 229 1.60 0.187 28.2
15V24 28 878 243 1.36 0.129 55.0
1038 20 743 245 1.33 0.169 80.0
10V45 22 909 269 1.50 0.168 21.0
1090 8 1147 309 1.69 0.168 4.3
M M M H H S
M M M H H S
M M M - M H
M M M H H S
M M M H M S
M M M H H S
M M M H H S
M M M H H S
M M M H H S
M M M H M S
M M - M S
M M M H M S
M M M H H S
M S S H M H
M M M - M H
M M S H M S
M M M H M S
M
55
Table 3.2: Summary of the correctly predicted cyclic deformation behavior of
selected steels from the AISI bar steel fatigue database [10] by various
tensile properties.

Experimental
Cyclic Behavior
Number
of Steels
% Correctly
Predicted by
S
u

% Correctly
Predicted by
HB
% Correctly
Predicted by
S
u
/S
y

% Correctly
Predicted by
n
% Correctly
Predicted by

Hardening 2 None None 100% 50% 50%
Softening 38 89% 92% 55% 68% 74%
Mixed-Mode 17 94% 88% None 53% None
All 57 88% 88% 40% 63% 51%



























56


(a)


(b)

57

(c)

Figure 3-1: Superimposed AISI and SAE steel data in a plot of (a) ultimate tensile
strength, (b) yield strength, and (c) monotonic strength coefficient versus
Brinell hardness.


Figure 3-2: AISI and SAE steel data from various material groups superimposed in a
plot of ultimate tensile strength versus yield strength.

58

(a)


(b)


59

(c)

Figure 3-3: AISI and SAE steel data of various material groups superimposed in a plot
of (a) ultimate tensile strength, (b) yield strength, and (c) monotonic
strength coefficient versus Brinell hardness.


Figure 3-4: Superimposed AISI and SAE steel data of various material groups shown
in a plot of S
u
/S
y
versus Brinell hardness.
60

(a)


(b)

Figure 3-5: A comparison of the cyclic deformation behavior for two steels with the
same hardness superimposed in a plot of (a) true stress versus true plastic
strain amplitude, and (b) true stress amplitude versus true strain amplitude.
61

(a)


(b)


62

(c)

(d)

Figure 3-6: Superimposed AISI steel data of various material groups shown in a plot
of cyclic strength coefficient versus (a) Brinell hardness, (b) ultimate
tensile strength, (c) monotonic yield strength, and (d) monotonic strength
coefficient.

63


Figure 3-7: Superimposed AISI data of various material groups in a plot of cyclic
strength coefficient versus S
u
/S
y
.







64

(a)


(b)


65

(c)


(d)

Figure 3-8: AISI data of various material groups superimposed in a plot of cyclic
strain hardening exponent versus (a) Brinell hardness, (b) ultimate tensile
strength, (c) monotonic yield strength, and (d) monotonic strain hardening
exponent.

66

(a)


(b)

Figure 3-9: AISI steel data of various material groups superimposed in a plot of cyclic
strain hardening exponent versus (a) S
y
/S
u
, and (b) S
u
/S
y
.

67

(a)


(b)

68

(c)

Figure 3-10: AISI steel data of various material groups superimposed in a plot of cyclic
yield strength versus (a) Brinell hardness, (b) ultimate tensile strength, and
(c) monotonic yield strength.











69

(a)


(b)


70

(c)

(d)

Figure 3-11: AISI and SAE J1099 steel data superimposed in a plot of predicted versus
experimental cyclic strength coefficient for (a) Equation 1.8 based on K
[7], (b) Equation 1.9 based on HB [8], (c) Equations 3.9 and 3.10 based on
S
u
, and (d) Equations 3.7 and 3.8 based on HB.
71

(a)


(b)


72

(c)


(d)


73

(e)

Figure 3-12: AISI and SAE J1099 steel data superimposed in a plot of predicted versus
experimental cyclic strain hardening exponent for (a) Equations 1.5
through 1.7 based on n, S
u
, S
y
, and
f
[7], (b) Equation 1.10 from [8], (c)
Equation 3.18 based on S
y
/S
u
, (d) Equations 3.25 and 3.26 based on S
y
and
HB, and (e) Equations 3.27 and 3.28 based on S
y
and S
u
.











74

(a)

(b)


75

(c)

(d)


76

(e)

(f)

Figure 3-13: AISI and SAE J1099 steel data superimposed in a plot of predicted versus
experimental cyclic yield strength for (a) Equation 3.20 based on HB, (b)
Equation 3.21 based on S
u
, (c) Equations 3.22 and 3.23 based on S
y
, (d)
Equation 2.9 with K' as a function of HB, (e) Equation 2.9 with K' as a
function of S
u
, and (f) Equation 1.11 based on RA and S
u
from [9].
77

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-14: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1020, (b) SAE 1038, (c) SAE 1038, (d) SAE 1541, (e)
SAE 1141Nb, (f) SAE 1050M, (g) SAE 10V45, and (h) SAE 10V45.

78

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-15: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1151V, (b) SAE 1045, (c) SAE 1141V, (d) SAE
1141V, (e) SAE 1141AL, (f) SAE 5150, (g) SAE 15V41, and (h) SAE
1090M.
79



(a) (b)

(c) (d)

(e) (f)

Figure 3-16: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE 1090, (b) SAE 1090, (c) SAE 1090, (d) SAE 4620, (e)
SAE 8620, and (f) SAE 8620.



80

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-17: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 9310, (b) SAE 1141AL, (c) SAE 41B17M (PS19), (d)
SAE 15B35, (e) SAE 4620, (f) SAE 8822, (g) SAE 8822, and (h) SAE
5140.
81


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-18: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 4320, (b) SAE 10B21, (c) SAE 4140, (d) SAE 4140,
(e) SAE 4140, (f) SAE 4140, (g) SAE 4140, and (h) SAE 4140.
82


(a) (b)

(c) (d)

(e)

Figure 3-19: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE 86B20, (b) SAE 4130AL, (c) SAE 51B60, (d) SAE
9254, and (e) SAE 9254AL.










83


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-20: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1022, (b) SAE C-70, (c) SAE 1141Nb, (d) SAE
15V24, (e) SAE 1038, (f) SAE 1141V, (g) SAE 5120, and (h) SAE 9310.
84


(a) (b)

(c) (d)

(e) (f)

Figure 3-21: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1090M, (b) SAE 8620, (c) SAE 20MnCr5, (d) SAE
20MnCr5, (e) SAE 20MnCr5, and (f) SAE 86B20.










85


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-22: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.4,
including, (a) SAE DDQ, (b) SAE DDQ+, (c) SAE DDQ+, (d) SAE
DQSK, (e) SAE 1005, (f) SAE 1005, (g) SAE 1008, and (h) SAE 10V40.
86


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-23: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE CQ, (b) SAE HF 50, (c) SAE HF 50, (d) SAE HF 50,
(e) SAE HF 60, (f) SAE HF 60, (g) SAE HF 50, and (h) SAE HF 60.
87


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-24: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with S
u
/S
y
1.2,
including, (a) SAE HF 60, (b) SAE HF 60, (c) SAE HF 80, (d) SAE HF
80, (e) SAE HF 80, (f) SAE HF 80, (g) SAE HF 80, and (h) SAE 51V45.
88


(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 3-25: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE 1005, (b) SAE 1004, (c) SAE HF 50, (d) SAE HF 50,
(e) SAE HF 50, (f) SAE HF 50, (g) SAE HF 60, and (h) SAE HF 80.
89


(a) (b)

Figure 3-26: Experimental monotonic and cyclic stress-strain curves superimposed with
the predicted cyclic stress-strain curves for steels with 1.2 < S
u
/S
y
< 1.4,
including, (a) SAE HF 80 and (b) SAE 1040.




























90
Chapter 4
Implication of the Proposed Cyclic Deformation
Method on Fatigue Life Estimations

4.1 Fatigue Life Prediction from Proposed Cyclic Stress-
Strain Estimation Method and the Roessle-Fatemi Strain-
Life Method
It was mentioned in Chapter 1 that significant error may result from using the
monotonic stress-strain curve in replacement of the cyclic stress-strain curve to estimate
the cyclic deformation behavior. The difference between the experimental cyclic and
monotonic stress-strain curves of the selected steels from the AISI database [10] and SAE
J1099 database [11] was shown in Figures 3-14 through 3-26. It may be seen from these
figures that significant improvements of strain-amplitude predictions, in relation to the
monotonic curve, were made using the proposed cyclic deformation estimation method
for most of the AISI steels. Implications of the proposed estimations of the cyclic stress-
strain relationship on fatigue life predictions are discussed in this chapter.
Currently, there are several approaches to fatigue life predictions. Two commonly
used approaches (the stress-based and strain-based approaches) predict the life of a
component until the initiation of a small detectable crack. The strain-based approach (-
N) is preferably used for low to intermediate strength steels because of its ability to
account for the plastic deformation that may occur in localized regions (e.g., where
91
cracks begin, at stress concentrations, or where local yielding occurs). This approach
relates the reversals to failure (2N
f
) to the total strain amplitude (A/2). This relationship,
often referred to as the Coffin-Manson relationship, is as follows:
c
f f
b
f
f p
e
N N
E
) 2 ( ) 2 (
2 2 2
'
+
'
=
A
+
A
=
A
c
o c
c c
(4.1)
Just as the cyclic deformation properties of a material are often unavailable, the
strain-life fatigue constants (
f
', b,
f
', c) are also not as commonly available as tensile
properties or hardness. Therefore, a need exists for strain-life behavior approximations
based on commonly available or easily attainable properties such as hardness or tensile
properties. In a study conducted by Lee and Song [20], several methods for estimating the
strain-controlled fatigue constants were studied using data from 139 materials including
40 steels. Of the methods reviewed, the Roessle-Fatemi direct hardness method [12] was
found to estimate the strain-life fatigue properties successfully for high-alloy steels. This
method is advantageous in comparison to other fatigue life estimation methods because it
requires only hardness and the modulus of elasticity, both of which are either commonly
available, or easily measureable. This method is defined by the following equation [12]:
56 . 0
2
09 . 0
) 2 (
191000 ) ( 487 ) ( 32 . 0
) 2 (
225 ) ( 25 . 4
2

+
+
+
=
A
f f
N
E
HB HB
N
E
HB c
(4.2)

The strain amplitudes used in Equation 4.2 were determined using the following
technique. A set of strain amplitudes similar to those used in the strain-controlled fatigue
tests (covering a fatigue life range of about 2x10
2
to 2x10
6
reversals) was selected for
each AISI steel material. Since the strain-amplitudes used for fatigue testing were not
included in the SAE J1099 report [11], a set of strain amplitudes were selected for these
steels to provide the aforementioned fatigue life range. The stress amplitudes
92
corresponding to these strain amplitudes were calculated using Equation 2.6 in which the
experimental cyclic strength coefficient (K') and cyclic strain hardening exponent (n')
were used. Predicted strain amplitudes were then calculated from cyclic stress-strain
Equation 2.6 with the estimated K' and n' values for the aforementioned stress
amplitudes. Predicted strain amplitudes for the same stress amplitudes were also
calculated from Equation 2.5 using experimental values of the monotonic strength
coefficient (K) and strain hardening exponent (n). Therefore, three sets of strain
amplitudes corresponding to one set of stress amplitudes were derived for each material:
a set representing the actual strain amplitudes used in testing, a set calculated from
Equation 2.6 using predicted cyclic deformation properties, and a set calculated from
Equation 2.5 using experimental monotonic deformation properties. Correspondingly,
three sets of reversals to failure (2N
f
) were calculated from Equation 4.2, one for each set
of strain amplitudes. For clarification, this procedure is illustrated in Figure 4-1. In
Section 4.2, fatigue lives obtained from estimated cyclic deformation properties and from
experimental monotonic deformation properties are compared to fatigue lives obtained
from the experimental strain amplitudes.
4.2 Comparisons of Fatigue Life Estimations Based on the
Experimental and Predicted Cyclic Deformation Curves
and the Experimental Monotonic Deformation Curve
In this section, comparisons are made between the fatigue lives based on strain
amplitudes obtained from the experimental cyclic stress-strain curve, the predicted cyclic
stress-strain curve, and the experimental monotonic stress-strain curve for fifty-seven
steel materials selected from the AISI database [10]. Results from the predicted fatigue
93
lives of the SAE J1099 steels [11] shown in Figures 3-22 through 3-26 are also compared
in this section. For all prediction methods, plots of predicted vs. experimental fatigue life
(2N
f
) were made. The predicted fatigue life was derived for chosen stress amplitudes and
the corresponding strain amplitudes obtained from three sources: two predicted cyclic
stress-strain relationships based on Equation 2.6 and the experimental monotonic stress-
strain relationship based on Equation 2.5, as illustrated in Figure 4-1. The predicted
cyclic stress-strain relationship based on method 1 was obtained by substituting predicted
values of K' calculated from Equations 3.9 and 3.10 and n' calculated from Equations
3.26 and 3.27 into Equation 2.6. The predicted cyclic stress-strain relationship based on
method 2 was obtained by substituting predicted values of K' calculated from Equations
3.9 and 3.10 and n' calculated from Equation 3.17 into Equation 2.6. These relationships
are shown in Figures 3-14 through 3-26 and are labeled Cyclic Pred. 1 and Cyclic
Pred. 2, respectively. The experimental fatigue life was calculated from Equation 4.2 in
which strain amplitudes representative of testing conditions were used. All AISI steels for
which S
u
/S
y
1.4 were divided into three sets of plots (Figures 4-2 through 4-4), as were
the S
u
/S
y
1.2 AISI materials (Figures 4-5 through 4-7). The 1.2 < S
u
/S
y
< 1.4 AISI
steels were included in two sets of plots (Figures 4-8 and 4-9). For every set of plots,
unique symbols representing each material are included. The materials are listed
according to increasing hardness, and the hardness range for each group is shown in the
figures. For comparison purposes, a 45 line and scatter bands using life factors of 3 and
5 are included in all of the figures.
In Figures 4-2(a) through 4-4(a), a plot of the estimated fatigue life based on the
predicted cyclic deformation curve (method 1) vs. experimental fatigue life based on
94
representative strain amplitudes is provided for AISI steels of which S
u
/S
y
1.4. The
same steels are also shown in Figures 4-2(b) through 4-4(b), where the estimated fatigue
life based on the monotonic cyclic deformation curve is plotted against the experimental
fatigue life. From Figures 4-2(a) and 4-4(a) it may be seen that all of the fatigue lives
estimated from the predicted cyclic deformation curve fall within life factors of 3. In
contrast, the majority of fatigue lives estimated from the monotonic deformation curve
are conservative with only 13% of the data within a life factor of 3 as shown in Figure
4-2(b). This is expected given that all of the materials in this group have a S
u
/S
y
ratio
greater than or equal to 1.4 and are, thus, expected to cyclically harden. For cyclic
hardening materials the stain amplitude estimated from the monotonic stress-strain curve
corresponding to a given stress amplitude is greater than the strain amplitude obtained
from the cyclic stress-strain curve. Therefore, the corresponding fatigue life based on the
monotonic stress-strain curve is shorter than the experimental fatigue life.
In Figures 4-5(a) through 4-7(a), estimated fatigue life based on the predicted
cyclic deformation curve (method 1) is plotted against experimental fatigue life for AISI
steels of which S
u
/S
y
1.2. Results from Figures 4-5(a) and 4-6(a) indicate that
prediction method 1 provides the most accurate estimation of the cyclic stress-strain
curve of steels in the relatively low to intermediate hardness range (185 HB to 370 HB).
It may be seen from all the figures that most of the predicted fatigue lives are
conservative. Conversely, fatigue lives estimated from the experimental monotonic
curves are, for the most part, non-conservative as shown in Figures 4-5(b) through 4-7(b).
This is expected since all of the materials in this group cyclically soften. For cyclic
softening materials the stain amplitude estimated from the monotonic stress-strain curve
95
corresponding to a given stress amplitude is less than the strain amplitude obtained from
the cyclic stress-strain curve. Therefore, the corresponding fatigue life based on the
monotonic stress-strain curve is greater than the experimental fatigue life.
In Figures 4-8(a) and 4-9(a), estimated fatigue life based on the predicted cyclic
deformation curve (method 1) is plotted against experimental fatigue life for AISI steels
of which 1.2 < S
u
/S
y
< 1.4. It may be seen from Figures 4-8(b) and 4-9(b) that
significantly more fatigue life data estimated from the monotonic stress-strain curve are
within the factor bands compared to those from steels with S
u
/S
y
1.2. Nevertheless,
estimated fatigue lives based on the predicted cyclic stress-strain curve for most of these
steels are better than predictions based on the monotonic curve.
Fatigue life data based on prediction method 1 for all of the AISI steels included
in this study are plotted in Figure 4-10. Similarly, fatigue life predictions based on
method 2 for the AISI steels are shown in Figure 4-11. Estimated fatigue life data based
on the experimental monotonic deformation curve for the AISI steels are plotted in
Figure 4-12. From comparison of Figures 4-10 and 4-11 it is clear that fatigue life
estimations based on prediction method 1 are significantly better, compared to the fatigue
life estimations based on prediction method 2. At short and intermediate lives, both
prediction methods 1 and 2 provide significant improvement of fatigue live estimations,
compared to those based on the monotonic deformation curve. In contrast, there is no
significant improvement of estimated fatigue lives at long life (more than 10
5
reversals).
This is expected given the fact that at long lives strains are mostly elastic. Therefore, no
significant difference exists between the monotonic and cyclic stress-strain curves.
96
Estimated fatigue life data based on prediction methods 1 and 2 for the SAE
J1099 steels included in Figures 3-22 through 3-26 are plotted in Figures 4-13 and 4-14,
respectively. From comparison of these figures it can be seen that more fatigue life data
based on method 1 are within life factors of 3 and 5, compared to fatigue life data
based on method 2.
































97








































Figure 4-1: Overview of the fatigue life calculation procedure used.

Stress Amplitude
Strain Amplitude
based on the
Experimental Cyclic
Stress-Strain Curve
Strain Amplitude
based on the
Predicted Cyclic
Stress-Strain Curve
(2 Methods)
Strain Amplitude
based on the
Experimental Monotonic
Stress-Strain Curve
Fatigue Life
based on the Roessl-Fatemi Strain-Life Equation

98

(a)

(b)

Figure 4-2: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
135 HB to 215 HB.

99

(a)

(b)

Figure 4-3: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
215 HB to 270 HB.

100

(a)

(b)

Figure 4-4: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.4 and hardness range of
270 HB to 350 HB.
101

(a)

(b)

Figure 4-5: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
185 HB to 305 HB.

102

(a)


(b)

Figure 4-6: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
320 HB to 370 HB.
103

(a)

(b)

Figure 4-7: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with S
u
/S
y
1.2 and hardness range of
390 HB to 585 HB.

104

(a)

(b)

Figure 4-8: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with 1.2 < S
u
/S
y
< 1.4 and hardness range
of 195 HB to 270 HB.

105

(a)

(b)

Figure 4-9: Comparison of fatigue lives based on strain amplitudes obtained from
experimental cyclic stress-strain curve and (a) the predicted cyclic
deformation curve, and (b) the experimental monotonic stress-strain curve
for AISI database materials [10] with 1.2 < S
u
/S
y
< 1.4 and hardness range
of 270 HB to 400 HB.

106
F
i
g
u
r
e

4
-
1
0
:

F
a
t
i
g
u
e

l
i
f
e

e
s
t
i
m
a
t
i
o
n
s

f
o
r

s
t
e
e
l
s

i
n

t
h
e

A
I
S
I

d
a
t
a
b
a
s
e

b
a
s
e
d

o
n

t
h
e

p
r
e
d
i
c
t
e
d

c
y
c
l
i
c

d
e
f
o
r
m
a
t
i
o
n

c
u
r
v
e

(
m
e
t
h
o
d

1
)
.
























107























F
i
g
u
r
e

4
-
1
1
:

F
a
t
i
g
u
e

l
i
f
e

e
s
t
i
m
a
t
i
o
n
s

f
o
r

s
t
e
e
l
s

i
n

t
h
e

A
I
S
I

d
a
t
a
b
a
s
e

b
a
s
e
d

o
n

t
h
e

p
r
e
d
i
c
t
e
d

d
e
f
o
r
m
a
t
i
o
n

c
u
r
v
e

(
m
e
t
h
o
d

2
)
.

108























F
i
g
u
r
e

4
-
1
2
:

F
a
t
i
g
u
e

l
i
f
e

e
s
t
i
m
a
t
i
o
n
s

f
o
r

s
t
e
e
l
s

i
n

t
h
e

A
I
S
I

d
a
t
a
b
a
s
e

b
a
s
e
d

o
n

t
h
e

e
x
p
e
r
i
m
e
n
t
a
l

m
o
n
o
t
o
n
i
c

d
e
f
o
r
m
a
t
i
o
n

c
u
r
v
e
.

109























F
i
g
u
r
e

4
-
1
3
:

F
a
t
i
g
u
e

l
i
f
e

e
s
t
i
m
a
t
i
o
n
s

f
o
r

s
t
e
e
l
s

i
n

S
A
E

J
1
0
9
9

b
a
s
e
d

o
n

t
h
e

p
r
e
d
i
c
t
e
d

c
y
c
l
i
c

d
e
f
o
r
m
a
t
i
o
n

c
u
r
v
e

(
m
e
t
h
o
d

1
)
.

110






















F
i
g
u
r
e

4
-
1
4
:

F
a
t
i
g
u
e

l
i
f
e

e
s
t
i
m
a
t
i
o
n
s

f
o
r

s
t
e
e
l
s

i
n

S
A
E

J
1
0
9
9

b
a
s
e
d

o
n

t
h
e

p
r
e
d
i
c
t
e
d

d
e
f
o
r
m
a
t
i
o
n

c
u
r
v
e

(
m
e
t
h
o
d

2
)
.

111
Chapter 5
Summary and Conclusions

This study investigated correlations among the various monotonic tensile
properties, cyclic deformation properties, and hardness for steel materials. Experimental
data for fifty-seven steel materials from the AISI bar steel fatigue database [10] were
used to obtain the correlations. Several correlations from the literature to predict tensile
and cyclic deformation properties were reviewed. By applying the correlations from
literature to the steels used in this study, the validity of these correlations was evaluated.
Sixty-six additional steel materials from the SAE J1099 report [11] were also used to
confirm the predictive accuracy of the proposed correlations in comparison to the
correlations suggested in the literature. By using a total of 123 steel materials, a wide
variation of chemical composition and mechanical properties was covered. New
correlations to predict the cyclic deformation properties of steels based on common
tensile properties and hardness were proposed in this study. Implications of fatigue life
predictions based on the proposed correlations were investigated using steels from the
AISI database and the SAE J1099 database. It is important to mention that these
correlations can only estimate material properties and, therefore, should not replace
experimental testing when reliable properties and high confidence are required.
112
Based on the data analysis and discussions presented in the preceding chapters,
the following conclusions can be drawn:
1. A correlation between the ultimate tensile strength (S
u
) and Brinell hardness (HB)
was found. The second-order polynomial equation used to fit data from the AISI
database is nearly identical to the equation relating S
u
and HB proposed by
Roessle and Fatemi in [12]. Both equations represent data from the AISI database
and SAE J1099 database well and yield nearly identical predictions of S
u
based on
hardness. The equation proposed by Roessle and Fatemi [12] is more accurate in
predicting S
u
from HB for higher hardness steels, compared to the common linear
prediction method from the literature.
2. A correlation between the yield strength (S
y
) and Brinell Hardness (HB) was
found and is represented by a second-order polynomial fit to the data from the
AISI database. This equation is very similar to that suggested by Roessle in [19].
3. Cyclic softening can be expected for steel materials of relatively high strength or
hardness and mixed-mode behavior (i.e. softening followed by hardening) can be
expected for steel materials of relatively low strength or hardness. Based on
experimental data from the fifty-seven AISI steel materials used in this study,
89% of the steels which cyclically softened have ultimate tensile strength (S
u
)
greater than 920 MPa and 92% have Brinell hardness (HB) greater than 250. Of
the steels that exhibited mixed-mode behavior, 94% have S
u
less than 920 MPa
and 88% have HB less than 250.
4. Significant variation between the cyclic deformation properties from two steel
materials of the same hardness can exist. In general, steels for which S
u
/S
y
1.4
113
have larger cyclic strength coefficients (K') and cyclic strain hardening exponents
(n') compared to steels of the same hardness for which S
u
/S
y
1.2. Also, steels for
which S
u
/S
y
1.4 have smaller cyclic yield strengths (S
y
') compared to steels of
the same hardness for which S
u
/S
y
1.2. Therefore, hardness alone cannot be
used to accurately predict the cyclic deformation properties of steels.
5. The cyclic strength coefficient (K') can be roughly approximated as being equal to
the monotonic strength coefficient (K). However, significant error may result
from replacing the cyclic strain hardening exponent (n') with the monotonic one
(n). A reasonable correlation was found between n' and n using a linear least
squares fit.
6. Classification of steels into material groups based on the S
u
/S
y
ratio improved
correlations between hardness, monotonic tensile properties, and cyclic
deformation properties within each material group. The cyclic strength coefficient
(K') was found to be dependent upon Brinell hardness (HB) and ultimate tensile
strength (S
u
) when the data were separated into two material groups. In a similar
manner, the cyclic yield strength (S
y
') was found to be dependent upon the
monotonic yield strength (S
y
) within both material groups. The cyclic yield
strength value was also found to correlate with HB and S
u
data of steels from both
groups. For both of these correlations, a second-order polynomial equation
represents most of the data well.
7. The cyclic strain hardening exponent (n') value was found to correlate with the
S
y
/S
u
ratio. There was no correlation found between (n') and Brinell hardness,
yield strength, or ultimate tensile strength for the AISI steels used in this study.
114
Another method to predict n' from predicted K' and S
y
' was shown to provide
good results.
8. A comparison was made of several correlations that were obtained from the
literature for estimating the cyclic strength coefficient (K') and the cyclic strain
hardening exponent (n') from tensile properties and hardness. It was shown that
the proposed cyclic deformation correlations were better at predicting K' and n'
than the correlations suggested in the literature for both AISI and SAE J1099 data
sets. A method from the literature and two proposed methods for predicting S
y
'
from tensile properties were shown to provide comparable results based on both
AISI and SAE J1099 data sets.
9. Several methods were proposed for estimating the cyclic stress-strain curve by
predicting the cyclic strength coefficient (K') and cyclic strain hardening exponent
(n') from common tensile properties and hardness. Two sets of equations for
estimating K', one from Brinell hardness and another from ultimate tensile
strength, were obtained from least squares fits among data within two material
groups based on S
u
/S
y
. Similarly, a set of two equations were derived for
estimating n' from S
u
and S
y
. An equation for estimating n' from S
y
/S
u
was derived
from a linear fit to all of the data.
10. Fatigue lives based on predicted cyclic stress-strain curves from two proposed
methods were compared using experimental data from the AISI and SAE J1099
databases. Prediction method 1 utilizes a cyclic strain-hardening exponent (n')
value based on the ultimate tensile strength (S
u
) and yield strength, and prediction
method 2 utilizes n' based on the S
y
/S
u
ratio. Both prediction methods utilize a
115
cyclic strength coefficient (K') value based on S
u
. Predicted cyclic stress-strain
curves based on both methods 1 and 2 were shown to approximate the
experimental cyclic stress-strain curve for most AISI and SAE J1099 steels
reasonably well. The predicted cyclic stress-strain curve based on method 1
provided more estimated fatigue life data in life factors of 3 and 5, compared to
predictions based on method 2 for both AISI and SAE J1099 data sets.
11. The most significant deviations between the experimental monotonic and cyclic
stress-strain curves were observed for AISI steels with S
u
/S
y
1.2, as made
apparent by comparisons of fatigue life based on the monotonic stress-strain curve
versus fatigue life based on the cyclic stress-strain curve. The lowest percentages
of estimated fatigue life data within life factors of 3 and 5 based on the
monotonic stress-strain curve were observed for this material group.




116


References

[1] Stephens, R.I., Fatemi, A., Stephens, R.R., and Fuchs, H.O., Metal Fatigue in
Engineering, 2
nd
ed., John Wiley and Sons Inc., New York, NY, 2000.

[2] Bauschinger, J., On the Change of the Position of the Elastic Limit of Iron and
Steel under Cyclic Variations of Stress, Mitt. Mech.-Tech. Lab., Munich,
Germany, Vol. 13, No. 1, 1886.

[3] Klesnil, M. and Lukas, P., Fatigue of Metallic Materials, 2
nd
ed., Elsevier, New
York, 1992.

[4] Manson, S.S. Hirschberg, M.H., and Smith, R.W., Fatigue Behavior Under
Strain Cycling in Low and Intermediate Life Range, Technical Note D-1574,
National Aeronautics and Space Administration, Washington, 1963.

[5] Landgraf, R.W., Morrow, J.D., and Endo, T., Determination of the Cyclic Stress-
Strain Curve, Journal of Materials, Vol. 4, 1969, pp. 176-188.

[6] Zhang, Z., Li, J., Sun, Q., Qiao, Y., and Li, C., Two Parameters Describing
Cyclic Hardening/Softening Behaviors of Metallic Materials, Journal of
Materials Engineering and Performance, Vol. 18, 2009, pp. 237-244.

[7] Zhang, Z.P., Qiao, Y.J., Sun, Q., Li, C., and Li, J., Theoretical Estimation to the
Cyclic Strength Coefficient and the Cyclic Strain-Hardening Exponent for
Metallic Materials: Preliminary Study, Journal of Materials Engineering and
Performance, Vol. 18, 2009, pp. 245-254.

[8] Basan, R., Franulovic, M., and Smokvina Hanza, S., Estimation of Cyclic Stress-
Strain Curves for Low-Alloy Steel from Hardness, University of Rijeka, Rijeka,
Croatia, Journal for Theory and Practice in Metallurgy, Vol. 49, 2010, pp. 83-86.

[9] Li., J., Sun, Q., Zhang, Z., Li, C., and Qiao, Y.J., Theoretical Estimation to the
Cyclic Yield Strength and Fatigue Limit for Alloy Steels, Mechanical Research
Communications, Vol. 36, 2009, pp. 316-321.

[10] American Iron and Steel Institute (AISI), Bar Steel Fatigue Database,
barsteelfatigue.autosteel.org.

117
[11] Society of Automotive Engineers, Technical Report on Low Cycle Fatigue
Properties Ferrous and Non-Ferrous Materials, SAE J1099, Rev. 2002.

[12] Roessle, M.L. and Fatemi, A., Strain-Controlled Fatigue Properties of Steels and
Some Simple Approximations, International Journal of Fatigue, Vol. 22, 2000,
pp. 495-511.

[13] ASTM Standard E8-04, Standard Test Methods for Tension Testing of Metallic
Materials, Annual Book of ASTM Standards, American Society for Testing and
Materials, West Conshohocken, PA, Vol. 03.01, 2004, pp. 62-85.

[14] ASTM Standard E8-04, Standard Practice for Statistical Analysis of Linear or
Linearized Stress-Life (S-N) and Strain-Life (-N) Fatigue Data, Annual Book of
ASTM Standards, American Society for Testing and Materials, West
Conshohocken, PA, Vol. 03.01, 1995, pp. 670-676.

[15] ASTM Standard E646-00, Standard Test Method for Tensile Strain-Hardening
Exponents (n-values) of Metallic Sheet Materials, Annual Book of ASTM
Standards, American Society for Testing and Materials, West Conshohocken, PA,
Vol. 03.01, 2004, pp. 619-626.

[16] ASTM Standard E606-92, Standard Practice for Strain-Controlled Fatigue
Testing, Annual Book of ASTM Standards, American Society for Testing and
Materials, West Conshohocken, PA, Vol. 03.01, 2004, pp. 593-606.

[17] Dowling, N.E., Mechanical Behavior of Materials, Engineering Methods for
Deformation, Fracture, and Fatigue, Prentice-Hall Inc., Englewood Cliffs, NJ,
1993.

[18] Metals Handbook: Desk Edition, Boyer, H.E. and Gall, T.L., eds., American
Society for Metals, Metals Park, OH, 1985.

[19] Roessle, M.L., Correlations Among Microstructural Parameters, Tensile Data,
and Fatigue Properties for Steels, MS Thesis, The University of Toledo, Toledo,
OH, 1998.

[20] Lee, K. and Song, J., Estimation Methods for Strain-Life Fatigue Properties from
Hardness, International Journal of Fatigue, Vol. 28, 2006, pp. 386-400.

Anda mungkin juga menyukai